Enterprise iRestructur'ing and Economic Po icy in Russia Edited by Simon Commander Qimiao Fan Mark E. Schaffer EDI DEVELOPMENT S UDIES Other EDI Development Studies (In order of publication) Infrastructure Delivery: Private Initiative and the Public Good Edited by Ashoka Mody Trade, Technology, and International Competitiveness Irfan ul Haque Corporate Governance in Transitional Economies: Insider Control and the Role of Banks Edited by Masahiko Aoki and Hyung-Ki Kim Unemployment, Restructuring, and the Labor Market in Eastern Europe and Russia Edited by Simon Commander and Fabrizio Coricelli Monitoring and Evaluating Social Programs in Developing Countries: A Handbookfor Policymakers, Managers, and Researchers Joseph Valadez and Michael Bamberger Agroindustrial Investment and Operations James G. Brown with Deloitte & Touche Labor Markets in an Era of Adjustment Edited by Susan Horton, Ravi Kanbur, and Dipak Mazumdar Vol. I-Issues Papers; Vol. 2-Case Studies Does Privatization Deliver? Highlightsfrom a World Bank Conference Edited by Ahmed Galal and Mary Shirley The Adaptive Economy: Adjustment Policies in Small, Low-Income Countries Tony Killick Financial Regulation: Changing the Rules of the Game Edited by Dimitri Vittas The Distribution of Income and Wealth in Korea Danny Leipziger and others Public Enterprise Reform: The Lessons of Experience Mary Shirley and John Nellis (Also available in French and Spanish) Privatization and Control of State-Owned Enterprises Edited by Ravi Ramamurti and Raymond Vernon Finance at the Frontier: Debt Capacity and the Role of Credit in the Private Economy J. D. Von Pischke EDI DEVELOPMENT STUDIES Enterprise Restructuring and Economic Policy in Russia Edited by Simon Commander Qimiao Fan Mark E. Schaffer The World Bank Washington, D. C. Copyright 1996 The Internatonal Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing October 1996 The Economic Development Institute (EDI) was established by the World Bank in 1955 to train officials concerned with development planning, policymaking, investment analysis, and project implementation in member developing countries. At present the substance of the EDI's work emphasizes macroeconomic and sectoral economic policy analysis. Through a variety of courses, seminars, and workshops, most of which are given overseas in cooperation with local institutions, the EDI seeks to sharpen analytical skills used in policy analysis and to broaden understanding of the experience of individual countries with economic development. Al- though the EDI's publications are designed to support its training activities, many are of interest to a much broader audience. EDI materials, including any findings, interpretations, and conclusions, are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center Inc., Suite 910,222 Rosewood Drive, Danvers, Massachusetts 01923, U. S. A. The backlist of publications by the World Bank is shown in the annual Index of Publications, which is available from Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or from Publications, Banque mondiale, 66, avenue d'Iena, 75116 Paris, France. Simon Commander is principal economist in the New Products and Outreach Division of the World Bank's Economic Development Institute; Qimiao Fan is an economist in the World Bank's Country Operations Division II of the Europe and Central Asia Country Department III; and Mark E. Schaffer is professor of economics at the Centre for Economic Research in Transition, Heriot Watt University in the United Kingdom. Library of Congress Cataloging-in-Publication Data Enterprise restructuring and economic policy in Russia / edited by Simon Commander, Qimiao Fan, Mark E. Schaffer. p. cm. - (EDI Development studies, ISSN 1020-105X) Includes bibliographical references and index. ISBN 0-8213-3725-4 1. Business enterprises-Russia (Federation)-Finance. 2. Corporate governance-Russia (Federation) 3. Russia (Federation)-Economic policy- 1991- I. Commander, Simon. II. Fan, Qimiao. m. Schaffer, Mark E. IV. Series. HG4189.2.E57 1996 338.6'44'0947-dc20 96-21160 CIP Contents Foreword xi Acknozwledgments xii Contributors xiii Abbreviations and Acronyms xiv 1. Introductioni 1 Simon Commander, Qimiao Fan, and Mark E. Schaffer Shocks and Restructuring 2 Employment Decisions and Worker Compensation 3 Firms' Budget Constraints 5 Privatization and Firm Behavior 8 Prospects and Summary 10 Part I. Employment, Wages, and the Provision of Social Benefits 2. How Ruissian Firms Make Their Wage antd Employmeint Decisions 1 5 Simon Commander, Sumana Dhar, and Ruslan Yemtsov Branch Evolution: Evidence from Official Data 17 Firm Evolution: Survey Evidence 21 Owtvnership Effects 31 Firm Objectives 35 Explaining Insider Influence 37 Bargaining in the Firm 39 iii iv Contents Conclusions 48 References 49 Notes 50 3. Social Benefits and the Russian Industrial Firm 52 Simon Commander, Une J. Lee, and Andrei Tolstopiatenko What Benefits Are Provided? 55 Costs of Social Benefits 68 Cost Recovery 70 Structure of Compensation, Benefit Pricing, and Incentives 73 Conclusions 76 Appendix 78 References 81 Notes 82 Part II. Financial Aspects of Enterprise Restructuring 4. "Arrears" in the Russian Enterprise Sector 87 Gilles Alfandari and Mark E. Schaffer Late Payment versus Bad Debts: From Financial Stress to Financial Distress 89 Definitions and Measurement 91 Arrears in Perspective: Aggregates and Intemational Comparisons 97 Macroeconomic Policy, Liquidity, and Financial Distress 108 Arrears and Liquidity 109 Concentration of Arrears in Financially Distressed Firms 110 Microevidence on Arrears 116 The Responses of Enterprises and Policy Implications 128 Appendix: Definitions and Coverage of Goskomstat Data 134 References 136 Notes 138 5. Firms, Banks, and Credit in Russia 140 Qimiao Fan, Une J. Lee, and Mark E. Schaffer General Features 142 The Distribution of Bank Debt and the "Bad Debt" Problem 145 Bank Credit Supply and Demand 154 Banks' Influence on Enterprise Decisions 157 Concluding Remarks and Some Policy Implications 162 References 164 Notes 165 Contents v 6. Government Financial Transfers to Industrial Enterprises and Restructutring 166 Gilles Alfandari, Qimiao Fan, and Lev Freinkman Overview 167 The Nature of Government Financial Transfers 171 Concentration of Financial Transfers 176 Financial Transfers to Enterprises: Implicit Government Objectives 182 Government Financial Transfers and Enterprise Performance 194 Conclusions and Policy Recommendations 197 Appendix 199 References 200 Notes 201 Part III. Corporate Governance and Competition 7. Ownership Strtuctures, Patterns of Control, and Enterprise Behavior in Russia 205 John S. Earle, Saul Estrin, and Larisa L. Leshchenko Alternative Ownership Forms and Enterprise Behavior: Some Hypotheses 206 Evolution of Governance Form 212 Institutional Features of Russian Privatization 213 Corporate Control in Russian Enterprises 215 Ownership and Control in Russian Firms 221 Ownership and Enterprise Behavior 225 Reorientation of Firms' Objectives and Restructuring 233 Conclusions 243 References 247 Notes 249 8. The Perfornance of De Novo Private Firms in Ruissiani Manifacthring 253 Andrea Richter and Mark E. Schaffer Sample Characteristics and Methodology 255 Basic Characteristics of De Novo Firms 258 Economic Performance of De Novo Firms 263 Performance Expectations 268 Russian De Novo Performance Compared with Their Polish Counterparts 269 Conclusions 270 References 272 Notes 273 vi Contents Appendix. The World Bank Survey of 439 Induistrial Enterprises 275 Une J. Lee Sample Selection 276 The Survey Instrument 277 Classification 278 Overview of Sampled Enterprises 279 Notes 283 Index 285 Tables 2-1. Structure of Industrial Employment 18 2-2. Decomposition of Unit Labor Costs in Industry, 1992-94 21 2-3. Descriptive Statistics: Mean and Coefficient of Variation for Firms 22 2-4. Firms Classed by Excess Employment, 1994 26 2-5. Probit Estimation Relating Below-Poverty-Line Wages to Firm and Other Attributes 28 2-6. Relative Influence in Decisionmaking, 1994 30 2-7. Employment Stability by Ownership Type 32 2-8. Employment Changes: Delayed Privatizers and High Government- Shareholding Firms Compared with Other Privatizers 34 2-9. Correlation Matrix between Labor Orientation and Shares of Different Actors 34 2-10. Firm Objectives, 1994 and 1990/91 35 2-11. Correlation Matrix among Firm Objectives 36 2-12. Ordered Logit Estimation Relating Firm Objectives to Firm and Other Attributes 38 2-13. Regression of the Change in Employment 42 2-14. Regression of the Change in Wage Rate 43 2-15. Ratio of Wage to Gross Profit per Worker, 1993 and 1994 45 2-16. Regression of the Ratio of Wage to Reservation Wage 47 3-1. Provision of Benefits, Mid-1994 and 1990/91 56 3-2. Ownership of Benefits, Mid-1994 58 3-3. Provision of Benefits and Size of Enterprise, Mid-1994 59 3-4. Provision of Benefits by Main Industrial Sectors, Mid-1994 60 3-5. Provision of Benefits and Ownership, Mid-1994 61 3-6. Provision of Benefits by Region, Mid-1994 62 3-7. Provision of Benefits and Firm Setting, Mid-1994 64 3-8. Number of Social Benefits, Ordered Logit Estimation 65 3-9. Change in Benefit Levels, Ordered Logit Estimation 66 3-10. Cost of Benefits, Mid-1994 69 Contenits vii 3-11. Adjusted Cost of Benefits, Mid-1994 70 3-12. Financial Transfers, OLS Regression 72 4-1. Simplified Balance Sheet of a Russian Enterprise 92 4-2. Structure of Liabilities and Receivables, and Portions in Arrears, Survey and Goskomstat Data 95 4-3a. Balance Sheet of Russian Industrial Sectors, 1 January 1995, General Items and Selected Assets 100 4-3b. Balance Sheet of Russian Industrial Sectors, 1 January 1995, Selected Liabilities 101 4-4. Arrears in Russian Industry since 1992 102 4-5. Trade Credit and Overdue Trade Credit in Western and Transition Economies 104 4-6. Partial Correlations of Assets and Liabilities in Arrears with Money Holdings III 4-7. Concentration of Liabilities and Arrears 114 4-8. Correlations of Arrears with Arrears 118 4-9. Simple and Partial Correlations between Arrears and Firm Characteristics 119 4-10. Arrears, Demand, and Response to Change in Demand 120 4-11. Arrears and Financial Indicators 121 4-12. Term Structure of Arrears 123 4-13. Frequency of Occurrence of Arrears 124 4-14. Cited Causes of Overdue Payables 126 4-15. Payment Priorities and Financial Distress: Ranking of Payment Obligations in Order of Urgency 127 4-16. Methods Used to Control Overdue Receivables 130 4-17. Obstacles to Pursuing Debtors 131 5-1. Ownership Cross-holdings in Banks and Firms 144 5-2. Bank Credit and Overdue Bank Credit in Russia, 1990-95 146 5-3. Bad Debt and the Rollover Problem: "Have You, in the Past Two Years, Failed to Repay or Service a Bank Debt on Time?" 149 5-4. Term Structure of Overdue Liabilities to Banks 150 5-5. Concentration of Bank Debt in Financially Distressed Firms 151 5-6. Characteristics of Firms with Significant Amounts of Overdue Bank Credit 153 5-7. Ease of Obtaining Bank Credit on Commercial Terms 154 5-8. Main Problems in Obtaining Bank Loans 155 5-9. Factors Affecting the Supply of Bank Credit, Ordered Logit Results 158 5-10. Which Firms Hold Bank Credit? 159 5-11. Bank Influence on Decisions of Enterprises 160 viii Contents 6-1. Average Size of Enterprises by Groups, 1994 170 6-2. Average Gross Transfers and Transfers per Unit 172 6-3. The Largest Recipients 181 6-4. Gini Indexes on Employment, Output, and Financial Transfers 182 6-5. Transfers and Ownership 185 6-6. Comparison of 1994 Recipients and Nonrecipients 186 6-7. Logistic Regressions, Controlling for Ownership and Sectors 187 6-8. Correlations with the Probability of Being a Recipient 190 6-9. Correlations with the Current Amount of Financial Transfers Received 191 6-10. Explained Variable: "Being a Recipient in 1994" 194 7-1. Comparison of the Impact of Alternative Ownership Forms in Attaining Objectives of Transition 208 7-2. Distribution of Ownership by Dominant Owner Type 217 7-3. Legal Form by Dominant Owner Type 218 7-4. Branch by Dominant Owner Type 219 7-5. Dominant Owner by Industry Sector Group 220 7-6. Region by Dominant Owner Type 220 7-7. Clarification of Property Rights: Influence of Actors by Dominant Owner Type 222 7-8. Decisions Concerning Employment: Hiring and Firing of Workers, Social and Nonwage Benefits 222 7-9. Decisions Concerning Employment: Hiring and Firing of Management, Managerial Compensation 223 7-10. Decisions Concerning Allocation of Profits, Major Investments, Sale or Lease of Major Assets, Financial Issues Generally 223 7-11. Correlation of Ownership and Influence 224 7-12. Depoliticization 228 7-13. Depoliticization Regressions 230 7-14. Depoliticization Regressions: Existence of Government Support 232 7-15. Depoliticization Regressions: Magnitude of Government Assistance 233 7-16. Responses on Importance of Management Strategies 235 7-17. Company Performance 237 7-18. Sales in 1994 239 7-19. Percentage of Sales Exported to Non-Former Soviet Union Economies 240 7-20. Capacity Utilization in 1994 241 7-21. Proportion of Capital Stock More than Fifteen Years Old 242 7-22. Full-Time Employment 242 7-23. Average Monthly Wage of Managers 243 7-24. Average Monthly Wage of Workers 244 Contents ix 8-1. Size Distribution of Firms 257 8-2. Geographical Distribution 259 8-3. Sectoral Distribution 260 8-4. Vintage of the Capital Stock 261 8-5. Wage and Labor Data 262 8-6. Output, Employment Growth, and Capacity Utilization 263 8-7. Job Creation and Job Destruction, Mid-1993 to Mid-1994 264 A-1. Distribution of Sample by Ownership 279 A-2. Distribution by Size of Enterprise 280 A-3. Distribution over Industrial Branches 280 A-4. Sample Distribution over Regions 281 A-5. Enterprises within the Military-Industrial Complex (MIC) 282 Figures 2-1. Industrial Output and Employment, 1991-94 16 2-2. Change in Employment Related to Change in Output, by Industry, 1990/91-1994 19 2-3. Change in Output Related to Change in Relative Wage, by Industry, 1990/91-1994 19 2-4. Monthly Wage by Branch, June 1991 and June 1994 20 2-5. Scatter of Hiring Rate to Firing Rate (1991-94 to 1991 Employment) 25 2-6. Log of Average Wage by Groups 29 2-7. Outsiders' Share and Ratio of Layoffs to Separations 33 2-8. Labor Market 41 2-9. Cumulative Density of Wage/Gross Surplus per Worker, 1994 46 2-10. Cumulative Density of Wage/Gross Surplus per Worker, 1993 46 3-1. Change in Average Wages and Cost of Social Benefits, 1990-94 74 4-1. Structure of Liabilities of Industrial Enterprises, 1 April 1994 98 4-2. Trade Credit in Arrears and Inflation in Russia 103 6-1. Structure of Financial Flows by Type of Transfer 174 6-2. Sectoral Distribution of Financial Flows 177 6-3. Sectoral Concentration of Recipients 179 6-4. Concentration of Subsidies among Recipient Enterprises 180 6-5. Concentration of Financial Flows According to Enterprise Size 192 6-6. Labor Productivity 196 Box 4-1. Penalty or Unpaid Interest Included? 94 Foreword This book is the result of a large research program that was launched at the World Bank in collaboration with Russian institutions and individual researchers in a number of countries. The work was generously sup- ported by the Europe and Central Asia Department, the Economic Devel- opment Institute, and the Research Committee of the World Bank. In Russia, the enterprise questionnaire that forms the basis for the research was implemented by the All Russia Centre for Public Opinion Research (VCIOM). Some of the earlier results from this research program were presented initially at a workshop held in Washington, D.C., in M[arch 1995, and then at a larger conference in St. Petersburg, Russia, in June 1995. The latter event was jointly organized with the Ministry of Econ- omy of the Russian Federation and the Leontieff Center in St. Petersburg. Vinod Tlhomas Director Economiic Development Institute xi Acknowledgments The editors would particularly like to thank the following colleagues at the World Bank for their support and encouragement throughout the project: Yukon Huang, Gregory Ingram, Costas Michalopoulos, Pradeep Mitra, Marcelo Selowsky, and Vinod Thomas. In addition, and at various stages of the work, Charles Blitzer, Barry Bosworth, Bingsong Fang, Alan Gelb, Richard Jackman, Martha de Melo, John Nellis, Gerhard Pohl, Randy Ryterman, Paulo Vieira da Cunha, and Wayne Vroman generously gave their time and much-appreciated advice. Stefan Koeberle was actively in- volved in the design of the questionnaire and in the preliminary analysis of results, as was Alexander Morozov in Moscow. At VCIOM the work of administering the questionnaires was ably completed under the direction of Marina Krassilnikova. We particularly thank Sergei Vasiliev and Elvira Nabiullina from the Ministry of Economy and Elena Belova and Irina Karelina from the Leontieff Center for their cooperation and support in the organization of the conference held in St. Peterbeurg, Russia, in June 1995 to present the results of this research program. Caroline McEuen has done a marvelous job in editing the manuscript. John Didier in the Eco- nomic Development Institute, with his customary charm and efficiency, handled the whole publication process. Throughout, administrative sup- port for the project was superbly provided by Yolanda Gedse, Michelle Mancesidor, and Natasha Veligura, with budgetary matters being han- dled by Kathy Hannum and Crummella Medley. Finally, Une Lee was at the heart of the whole endeavor. That the work was completed is in large measure a tribute to her considerable pro- fessional and personal skills. We are indeed very thankful. xii Contributors Gilles Alfandari The World Bank, Washington, D.C. Simon Commander The World Bank, Washington, D.C. Sumana Dhar The World Bank, Washington, D.C. John S. Earle Stanford University, Palo Alto, California, and Central European University, Budapest Saul Estrin London School of Economics Qimiao Fan The World Bank, Washington, D.C. Lev Freinkman The World Bank, Washington, D.C. Une J. Lee The World Bank, Washington, D.C. Larisa L. Leshchenko Central European University, Budapest Andrea Richter Council of Economic Advisers, Washington, D.C. Mark E. Schaffer Centre for Economic Reform and Transformation, Heriot-Watt University, Edinburgh Andrei Tolstopiatenko EDI, The World Bank, Moscow, and Moscow State University Ruslan Yemtsov The World Bank, Washington, D.C. xiii Abbreviations and Acronyms CBR Central Bank of Russia CEE Central and Eastern Europe CIS Commonwealth of Independent States CISAC Center for International Security and Arms Control CMEA Council for Mutual Economic Assistance CPI Consumer price index CSPP Centre for the Study of Public Policy DN De novo DSC Directed state credit EBFs Extra-budgetary funds ESOP Employee stock ownership plan FARP Fund of Workers' Shares GDP Gross domestic product GKI State Property Committee IPI Industrial price index MIC Military-industrial complex MO Managerially owned MoF Ministry of Finance NPE New private enterprise OECD Organization for Economic Cooperation and Development 00 Outsider-owned PCI Producer cost index PE Privatized enterprise PPI Producer price index SMEs Small and medium-size enterprises SO State-owned SOE State-owned enterprise xiv Abbreviations and Acronyms xv VAT Value added tax VCIOM All-Russia Centre for Public Opinion Research VTSIOM Russian Centre for the Study of Public Opinion WO Worker-owned Introduction Simon Commander, Qimiao Fan, and Mark E. Schaffer This book is concerned with a crucial issue of transition: the role and or- ganization of the enterprise sector. This dynamic is of particular interest in the Russian context, where there has been a concerted attempt to change ownership arrangements, and hence corporate governance. To analyze the implications of such changes in a convincing manner requires disaggregated information that covers the essential dimensions of deci- sionmaking within the firm. This volume breaks new ground by present- ing the results of a large survey of firms carried out by the World Bank in mid-1994 that included over 400 industrial firms. Selection of the main sample was random, stratified by both industrial sector and region; an additional forty-odd newly established private firms were added to the sample. Microfirms-employing fewer than fifteen persons-were ex- cluded from the survey (for further details of the survey, see the appen- dix to this volume). Coming in the wake of the mass privatization program, the survey displays a wide range of ownership forms. Of the fi- nal sample of 439 firms, over 60 percent had been privatized, with domi- nant ownership by a variety of combinations (workers, managers, outsiders, and so forth); 25 percent were still state-owned, and 10 percent were de novo private firms. The survey thus enables us to explore the rela- tionship between ownership change and firm behavior. I 2 Enterprise Restructutring anid Econiomic Policy in Russia Shocks and Restructuring Russian industry has faced large adverse shocks that have compounded the secular slowdown in productivity that preceded the breakup of the Soviet Union. Firms have been buffeted by a combination of both aggre- gate demand and supply shocks, including major negative effects associ- ated with decline in Council for Mutual Economic Assistance (CMEA) and intra-former Soviet Union trade. Over time, firms have to varying degrees had to operate as if under a hard budget constraint to respond to the decline in explicit subsidies and the diminution of cheap credits. Fed- eral government transfers declined significantly in real terms between 1992 and 1994, although the profile of local government transfers is less clear. Faced with these shocks, industrial output has fallen dramatically. Official data report industrial production at end-1994 at roughly half the level attained in 1990. It has been argued that official output series are subject to measurement error and tend to overstate the decline in output. Information from the survey, however, yielded roughly comparable numbers, whether output change is measured according to firms' own di- rect estimates of the real change or the nominal values are deflated to ar- rive at an estimate. Furthermore, figures on capacity utilization in the sample also indicate a large contraction since 1990/91, of the order of 30 percent for state firms and 40 percent for privatized firms. The origins of this contraction and the respective time paths do vary somewhat by branch, and perhaps even more so by region. The initial impression is that measured over output, shocks appear to have been distributed in a reasonably uniform manner. The survey allowed a look at the origins of these shocks and the sub- sequent response. It provided an opportunity to distinguish between an impact effect, or negative restructuring, and a longer-term, strategic, or positive restructuring decision. The former summarizes the set of re- sponses that firms have had to make as a result of these shocks, which can be considered a weak measure of restructuring, while the latter pro- vides a stronger measure of restructuring by looking at dynamic choices that relate to changes in trading partners, introduction of new product lines, changes in skill distributions in employment, and so on. The overall story that emerges is that firms have indeed been forced to restructure, but primarily through negative restructuring, generally with reductions in employment, working hours, and wages. Nevertheless, em- Introductihn 3 ployment adjustment in particular has been relatively slow. There is less evidence of more positive restructuring in both state and privatized firms. While nearly two-thirds of the firms included in the sample have experienced major changes in product mix, primarily through introduc- tion of new products and raising the number of product types, and many firms have initiated new trading relationships for both inputs and out- puts, these shifts are not at all tightly or consistently correlated with other restructuring decisions, including those governing employment and wages. Only one-quarter of firms reported that they had phased out pro- duction of any products, a surprisingly low figure, and another indica- tion of inertial behavior. The main exceptions to this picture are the de novo private firms. While still small in aggregate terms, this sector has been growing rapidly, and the de novo firms in the survey reflect this dy- namism in their performance and behavior. Employment Decisions and Worker Compensation Chapters 2 and 3 in this volume use the survey to analyze the aspects of wage and employment decisions of Russian firms. In Eastern Europe the predominant pattern has been that the first, or negative, stage of restruc- turing involves reductions in employment and some initial wage flexibil- ity, because workers have given priority to employment. This appears to be broadly true in the Russian case, but with some important caveats. Russian firms entered the transition with large excess employment. While employment reductions have begun to accelerate in relation to the fall in output, firms have continued to retain excess labor. Althoug:h real wages fell at transition, over time there has been some real wage recov- ery. These combined factors have forced up unit labor costs. The unambi- guous employment bias can be traced not only to insider influence at the level of the firm and the stability of its objectives, including employjment and maximization of worker welfare, but also to the outside environ- ment. With unemployment benefits providing an inadequate fall-back option for separated workers, firms have tended to act with some be- nevolence, retaining workers but adjusting the hours worked and the monetary components of compensation. One result has been a continuing low unemployment rate-2 percent according to registrations data; 5-6 percent in the survey. The pattern of adjustment-rather different jfrom that observed in Eastern Europe-appears to be one of firms partially 4 Enterprise Restrectutring and Econiomic Policy in Russia trading down wages for employment stability. This is clearly consistent with workers placing a high value on employment in relation to wages. But this is only part of the story. First, workers' compensation has become increasingly dominated by the nonmonetary components of compensation, principally the social benefits provided by firms, including some de novo firms. The tax regime, particularly the incidence of the excess wage tax on monetary compen- sation, has likely motivated some substitution of nonmonetary for monetary compensation. Firms-particularly larger firms that have expe- rienced smaller changes to employment and have had a history of pro- viding high benefits-have generally tried to keep up the supply of benefits, thereby anchoring household incomes, while at the same time allowing workers to allocate low effort to work in the firm. One result has been the growing prevalence of multiple jobholding as workers diversify their time allocations. Second, the size of shocks to firm balance sheets has commonly been sufficiently negative to prevent a simple tradeoff. In short, insiders con- tinue to extract rents even if wage levels remain low. Subsidies, tax ar- rears, and other transfers were still a significant factor in allowing rent-taking and in supporting the associated employment bias. Thus, signs of flexibility in the labor market, although real, need be tempered by the realization that for given financial performance, the employment bias in both state and privatized firms appears to remain genuinely large, particularly in the privatized group. As restructuring goes deeper, we can expect significant employment contraction. In short, the picture that has been painted of the Russian path of ad- justment-to take the effects of product market shocks purely through wages, while downplaying involuntary separations-is not fully consis- tent with what emerges on the financing side. Perhaps more significant, the form of adjustment that appears to have dominated in Russia has very ambiguous implications for dynamic restructuring. As chapter 7 demonstrates, the predominant privatization procedures in Russia have given insiders a major, continuing say in firm-level decisions. This insider control has not only ensured an absence of outside resources for financ- ing restructuring costs, but also has enabled the continuing attachment of workers to their original firms. As chapters 2 and 3 demonstrate, this at- tachment is pronounced, in part because relatively stable benefits and a Introdzction 5 rising share of total compensation provide increased incentives for work- ers to stay in their firms. Insofar as firms have continued to receive sig- nificant subsidies or tax breaks to finance benefits, the incentive for firms to reduce benefits provision has been correspondingly small. This im- poses several major costs. First, it tends to force de novo private firms to offer equivalent compensation packages, including benefits, to workers. The likely nontrivial start-up costs associated with provision of such benefits, or with acquiring access to them for workers, may in part ex- plain why formally established de novo firms have not grown faster. Rather, this encourages a phenomenon that has been widely observed in Russia-the informalization of the private sector. Private firms have a clear incentive to externalize the costs of benefits provision by relying on part-time or moonlight labor. Aside from the productivity effects this is likely to cause, such informalization has major fiscal implications because of the widespread tax evasion typical of such informal activities. In the longer run, an evident objective must be to lower attachment and shift the incentives for private sector job creation from informal to formal ac- tivity. To do so will require a continued contraction in the sources of soft financing for firms, including limiting the provision of social benelits in their current form. In sum, while the structure of control in Russian firms and government preferences have come together to limit the growth in open unemployment-at end-1995 unemployment as measured by the la- bor force survey was around 8 percent-it is not evident that this is either sustainable or ultimately desirable, given both the fiscal and restructuring implications that this approach implies. Firms' Budget Constraints The financing environment has obviously been central to firms' restruc- turing decisions. Chapter 4 looks at the problem of "arrears" in Russia, focusing on interenterprise arrears (overdue trade credit), tax arrears, and wage arrears. Arrears to banks, and bank-enterprise relations more gen- erally, are analyzed in chapter 5. In chapter 6 government financial trans- fers to firms are examined. Early in the transition the volume of subsidies remained very large- in 1992, total federal government financial transfers amounted to over 30 percent of gross domestic product (GDP). These fell sharply to around 6- 6 Enterprise Restrnctuiriing and Econ omiic Policy in Russia 7 percent in 1994, and they have become increasingly concentrated, at least by industrial branch, with the coal and agriculture sectors the princi- pal recipients. Fuel sector subsidies appear to have accounted for over 25 percent of all subsidies in 1994. Outside these branches, transfers have also remained highly concentrated, with around 50 percent of total re- ported transfers received by under 2 percent of firms. This means that while a sizable share of firms still receive some form of subsidy from the federal government-the share was around 25 percent in 1994-the aver- age amount of transfer received was small, and declining in real terms. Federal subsidies have continued to be directed toward large firms, when measured in employment, and have likely been related to a combination of employment-stabilizing objectives and compensatory finance for the provision of social services. In general, transfers seem to have been used to finance current operations and losses rather than restructuring. It is likely that the reported numbers understate the volume of trans- fers because they include only federal subsidies. A more comprehensive picture, which accurately measured transfers at a local government level, would indicate not only a far higher level of aggregate subsidy for the economy, but also a tighter link between transfers and employment. While the ultimate objective must be to reduce subsidies, it is evident that this cannot happen instantly. The main challenge in the interim will be to make more transparent the conditions under which transfers would be sanctioned. Insofar as stabilizing employment has been a major trigger for subsidies, it would probably be better to make such transfers explicit as an employment subsidy, with an announced level and time path. Focusing only on subsidies, whether federal or local, misses some cru- cial features. It is important to note that nearly half the firms in the sur- vey had accumulated tax arrears and/or payment arrears (of which more below). By 1994 the stock of tax arrears amounted to the equivalent of 4-5 percent of GDP, with the flow on the order of 2 percent of GDP. The ac- crual of tax arrears parallels developments in Eastern Europe for firms faced with lower explicit subsidies. Such substitution is confirmed by the tendency of recipients of small transfers to have smaller tax arrears than those who did not receive government transfers. What is also striking is the relative concentration of such arrears in financially distressed firms. Such firms comprised roughly 13 percent of the sample, but they ac- counted for close to half of total tax arrears. Tax arrears are a way for such firms to extract subsidies from the state. Introduction 7 As already noted, the knowledge that transfers and tax arrears con- tinue to support a soft budget constraint is important in understanding the employment, wage, and other choices made by firms. What about the relationship between firms and the financial system and firms and their suppliers? Aggregate data suggest very strongly that overdue trade credit ("interenterprise arrears") remained at levels comparable to those of Or- ganization for Economic Cooperation and Development (OECD) econo- mies after the attempt to net out or clear arrears in 1992. The survey indicates that basic credit control mechanisms have been widely used by firms to control their overdue receivables. What is far more worrisome, as already indicated, is nonpayment of taxes. Wage arrears are also a con- cern, not so much in and of themselves, but because they can be used by firms when they lobby the government for subsidies or tax concessions, as occurred on a large scale in 1995. It is also clear that firms have still been able to extract some soft fi- nancing from the banking system. Aggregate data indicate that overdue bank credit started to increase rapidly in 1994, reaching over one-third of total bank credit by mid-1995. The survey evidence suggests that late payments or arrears are mostly short term, that firms frequently fail to re- pay or service their bank debt on time, and that the practice of capitaliz- ing overdue interest and rescheduling principal is widespread-all of which indicates continuing softness in the banking system. The banking system is not completely soft, however. While it is common for firms to be part-owners of banks, this apparently has not translated into easier bank credit. A poor repayment history by firms also has been commonly associated with difficulties in obtaining new credit from the banking sys- tem. Nevertheless, because significant portions of the banking systerrL are undercapitalized and real interest rates have turned significantly positive, the large volumes of overdue bank credit and the continuing softness of the banking sector point to emerging liquidity problems and the risk of associated bank failures. In summary, Russian industrial firms, whether state-owned or privat- ized, have continued to extract soft financing from government, whether at the federal or local level, and from banks through a variety of channels. With respect to government, it appears that industry associations and firms with market power have been able to extract the greatest support. Banks have rising bad debt exposure, and in many cases they have proven unable to undertake effective credit risk assessments or exercise 8 Enterprise Restruicturing and Ecoionomic Policy in Rufssia any effective discipline on firms. While this can partly be traced to the continuing use of directed credits-with allocation decisions in effect taken by government agency-it is also attributable to weaknesses in the banking system itself. The result of this combined softness in the banking system, in tax collection, and in federal and local support has been to weaken the budget constraint facing firms. Direct government transfers from various levels of the fiscal system can still be captured, albeit at de- clining real levels. This declining volume of soft money has partially been offset by higher tax arrears and by the accumulation of bad debts from the banking system. Because Russian industrial firms still do not face hard budget constraints, loss-makers, including chronically distressed firms, have been allowed to survive, effectively sanctioning decisions on current operational costs inconsistent with even a zero profit constraint. Privatization and ]Firm Behavior The question of whether changing ownership had any effects on firm be- havior is considered in all the chapters in this volume, but receives par- ticular attention in chapter 7. Given that most privatization has been done by insiders-workers and managers-with outside stakeholders playing a relatively unimportant role, there are a priori reasons to be pessimistic. Indeed, workers held a dominant ownership position in nearly two- thirds of privatized firms. The evidence presented suggests that insider privatization reflected not only an explicit political choice, but also the de facto importance of workers in decisionmaking at the firm level. While it is true that managers appear to have considerable discretion in decision- making in many firms, they rarely make decisions that are obviously at odds with the perceived interests of inside workers. One result is that ownership changes are generally rather weakly associated with most in- dicators of performance, including sales, wages, and employment. This can, of course, be attributed in part to the short period of time that has passed since privatization. The importance of lags may be partially confirmed, however, by the finding that firms that had already conducted their first shareholders' meeting had significantly larger employment adjustments than those yet to convene that meeting. There was also some evidence that privatization was associated with a lower volume of transactions with the state, including receipt of subsidies, and de novo firms were un- ambiguously more weakly associated with the state in their dealings. Introduiction 9 Of interest is that firms clearly dominated by managers tended to maintain stronger links with the state than those dominated by workers. This could be interpreted as a superior ability of managers to maintain ties to both the state and politicians. Maintaining such ties, however, was not generally associated with lower levels of restructuring. Indeed, manager- dominated firms were likely to do more restructuring than worker-domi- nated firms. While outsider shareholding was surprisingly important- outsiders held a dominant stake in 16 percent of privatized firms-there was also little evidence that these stakes were being turned into more di- rect interference in decisionmaking. There was a largely absent link be- tween outside interest and behavioral variables. These rather weak effects of ownership change in performance vari- ables can in part, as already indicated, be attributable to lags, but they also need to be traced to both the financing environment of firms and the lack of translation of ownership into control. In the latter regard, the emergence of share consolidation and blockholding may begin to acceler- ate this translation. The overall picture of ownership that emerges is that the current share distributions are probably quite transitory and would be subject to major change, which would also depend on how the resid- ual shareholdings of government are resolved. The main exception to this picture has been the strong performance of newly established private firms. In general, the differences between the various ownership categories of state-owned and privatized firms (worker-controlled, manager-controlled, and the like) have remained small compared with the differences between these firms and de novo firms. The performance of the de novo sector is the subject of chapter 8. Aggregate evidence suggests that the de novo sector's presence in manu- facturing is growing-an estimate in mid-1995 puts it at 6 percent of total manufacturing employment in Russia. The de novo firms in the survey were indeed small, but they were growing rapidly, especially in employ- ment. This strong performance by new private firms appears to be driven by factors associated with their ownership rather than their size. Indeed, the low number of small and medium-size enterprises (SMEs) prior to transition might perhaps be expected to lead to rapid growth across the category, and not just of new private SMEs. The survey indicates instead that state-owned and privatized SMEs are, if anything, doing worse than larger state-owned and privatized firms (let alone compared with de nIovo firms). 10 Enterprise Restructuring and Econoomic Policy in Ruissia Prospects and Summary Russian firms have begun to adjust in largely predictable ways. Product lines are being changed, marketing networks are being recast, and changes to employment levels, skill distributions, and relative wages are being made. But the changes in most instances have not progressed very far and remain tnevenly applied. There has been a clear hardening of budget constraints, even if the volumes of subsidies, tax arrears, and soft credit remain substantial. As such, negative restructuring has dominated through most of Russian industry. Net job destruction has accelerated (even though hiring rates continue to be surprisingly high), and worker monetary compensation has remained low. But firms, particularly large firms, still act benevolently, providing employment stability as well as a significant range of nonmonetary social benefits. That this is a feasible strategy can only partially be attributed to wage flexibility. Rather, con- tinuing access to financing outside the firm is important. Such finance comes from the various levels of government, with regional factors as- suming increasing importance, as well as from the banking system. Fur- ther, investment, particularly in machinery and equipment, has declined massively, and there are only very limited indications of firms being able to obtain external financing for new and needed investment. An obsolete capital stock obviously contributes to a further slowdown in productiv- ity. Finally, privatization has yet to show any clear effect in performance. That defensive restructuring has dominated thus far is probably not surprising. Ownership changes have been recent, and there remains con- siderable uncertainty in the policy environment. Insider privatization has not brought new investment resources to bear, and this has commonly been compounded by credit market failures. In addition, transfers to the firm sector appear to have been made in a manner that impedes effective restructuring. This is either because such supports compensate for serv- ices that firms, rather than government, provide, or because they are a short-run response to financial distress used to postpone the required adjustment. While a consistent message emerges that government is likely to re- main unable to play an effective and active role in restructuring, the ab- sence of appropriate management skills and human capital continue to limit the scope and efficacy of restructuring. Nevertheless, there are signs that managers and, in some instances, outside shareholders are consoli- Introduictior 11 dating their stakes, and this may eventually translate into control and im- provements in corporate governance. But it is also possible to see emerg- ing managerial dominance as an outcome consistent with continued rent-taking and exploitation of links to the state. The absence of an effec- tive exit mechanism must continue to affect the efficiency of any disci- pline that should come through consolidation in ownership. In promoting positive restructuring, what the government does with its residual shareholding in the second stage of privatization will be cru- cial. In 1995 an attempt was made to involve the banks explicitly in the management of this residual share. Banks were assigned the dual func- tion of acting as strategic investors in firms, while using government's shareholdings as collateral for lending to government. For a variety of reasons, including a lack of transparency, this proved to be highly prob- lematic. More generally, there are grounds for caution. First, much of the banking system remains not only undercapitalized but also has mnany nonperforming loans to the firm sector. Raising equity exposure could potentially accentuate the existing softness of the financial system and the emerging bad debt problem. Second, mobilizing investment resources from the banking system will obviously depend on banks being able to exercise effective control through the residual shareholding. Given the entrenched power of insiders, particularly managers, in most privatized firms, this control may prove elusive. Pulling the investment into firms that is critical for effective restructuring to proceed continues to run up against the huge concessions that have already been made to insiders, whose rights of control have yet to be effectively diluted. Share consoli- dation and other changes may ultimately facilitate strategic alliances be- tween dominant insider interests and outsiders, but it seems generally unlikely that insider interests can be ignored or overruled. Third, as with the financial-industrial groups that have been formed, there are potential incentive problems associated with closer bank-firm ties. These could well perpetuate close and often undesirable links among firms, banks, and government, and may actually facilitate continuing softness in firms' budget constraints. In this context, closer ties between banks and firms lead to rent-seeking rather than the promotion of improved corporate governance. In conclusion, this book is an attempt to understand a key part oi the unique and exciting process of transformation in Russia. Within a. re- markably short space of time, structural changes to the economy have 12 Enterprise Restrecturintg and Economic Policy in Russia been put in place. The findings of the survey show that in many critical respects, these changes have yet to be completed. This is hardly surpris- ing. While the nature of the privatization in Russia poses a serious chal- lenge to improvements in corporate governance and performance, the chapters in this book also give testimony to the fact that crucial shifts, in- cluding the emergence of a new private sector, are now under way. Part I Employment, Wages, and the Provision of Social Benefits 2 How Russian Firms Make Their Wage and Employment Decisions Simon Commander, Sumana Dhar, and Ruslan Yemtsov Despite large cumulative declines in output, a striking feature of Russian firms has been their continuing unwillingness to shed labor. This is quite evident at an aggregate level from figure 2-1, in which cumulative changes to employment at end-1994 can be seen to have greatly Ilgged those to output, at least for the industrial sector. Given the initial condi- tions-firms entered the transition with large labor hoarding-this appar- ent unwillingness to reduce employment, in contrast with much of the experience in Eastern Europe, begs explanation. One line of argument has been to emphasize the combination of technological factors, benevolence, and continuing access to soft credits, with employment emerging as the chief object of bargaining among firms, the financial system, and govern- ment (as in Commander, McHale, and Yemtsov 1995). Another has been to stress the willingness of workers to trade down wages for employment stability (Layard and Richter 1994). It has also been argued that changes in ownership status-and ultimately in governance-will accelerate We thank Richard Jackman, Saul Estrin, Mark E. Schaffer, and Wayne Vroman for detailed comments on previous drafts of this chapter, as well as the participants at workshops and conferences in Budapest, St. Petersberg, and Washington, D.C. 15 16 Employment, Wages, and the Provision of Social Benefits Figure 2-1. Industrial Output and Employment, 1991-94 December 1990 = 100 95 A Employment 95 80 80 Output 65 65 50 50 35 I, . 35 Jan. July Jan. July Jan. July Jan. July 1991 1991 1992 1992 1993 1993 1994 1994 elimination of any employment bias by raising the relative bargaining power of management and, in some cases, outside investors (Schleifer and Vasiliev 1994). The impact of privatization on these factors is yet to be adequately understood, in part because of the short elapse of time, and in part because of a lack of adequate data. The evidence that we are now able to present, however, suggests that with the massive preponderance of insider privatization, significant subsidies to firms from various levels of government, softness in the banking system, and the continuing reluc- tance of government to sanction high unemployment, the effectiveness of privatization in reducing excess employment has been quite weak. This chapter represents an attempt to understand the factors govern- ing firms' wage and employment decisions at the start of transition and, more recently, in the wake of the mass privatization program. The focus is exclusively on the monetary component of wages; chapter 3 takes a closer look at another important part of compensation, social benefits or in-kind compensation. While these additional considerations do change the overall picture, it is probably appropriate to think of firms and work- Hozw Ruissian Firms Make Their lAWage and Enmployment Decisions 17 ers as primarily concerned with setting monetary compensation and em- ployment in line with their respective preferences, bargaining powers, and budget constraints. That benefits do not appear to have been used as an explicit substitute for cash wages similarly suggests that by focusing on monetary compensation and employment, we are effectively dealing with the primary objects of bargaining. It is important to add, however, that the provision of firm-specific social assets has continued to foster at- tachment, and hence should be seen as one of the impediments to the creation of a better functioning labor market.' This chapter is largely based on a World Bank survey of 439 indusltrial firms. For the bulk of the variables presented there are at least three con- sistent datapoints relating to the pretransition situation in 1990/91, to 1992 and/or 1993, and to mid-1994. The organization of the data clearly pushes us toward exploring the cross-sectional properties, as well as the changes over time that we can isolate. The survey comprises both quanti- tative and qualitative sections that enable us to evaluate not only the evo- lution of financial and real variables, but also the factors internal and external to the firm that have governed decisionmaking. At the same time, by covering the period of 1990 through mid-1994, the dataset picks up a significant segment of firms that have been privatized or have en- tered the process. Indeed, by mid-1994 just over 20 percent of firms in the sample were currently and prospectively state-owned. This rapid evis- ceration of the state sector indicates that, at least in legal status, the per:iod encapsulated by the survey has seen a dramatic transformation. We at- tempt to deal with some of the possible behavioral effects and their lags in this chapter. Branch Evolution: Evidence from Official Data Before looking at the survey results, we will briefly consider the evidence and the story that emerges from official series at branch-level disaggrega- tion. Apart from anything else, this allows us to cross-check the structure of survey responses against information for the industrial sector as a whole. Table 2-1 portrays the structure of employment at two points- 1991 and the first two quarters of 1994-for all industrial branches using Goskomstat and survey data. Several discrepancies can be seen. Capital goods producers in machine-building are significantly overrepresented in the survey, as are firms in the fuel and energy branches. There is signifi- 18 Employnmenit, Wages, anid tlhe Provisioni of Social Benefits Table 2-1. Structure of Industrial Employment (percent) Employmenit by bmnch Goskomstat World Bank Survey Prereform 1994 Prereform 1994 Source (1991) (1-2Q) (1991) (1-2Q) All industry 100.0 100.0 100.0 100.0 Energy 2.9 4.3 6.4 12.2 Fuels 5.1 7.1 10.1 17.7 Ferrous metals 4.7 5.4 1.3 1.2 Nonferrous metals 3.0 3.8 2.9 4.1 Chemical and petrochemical 6.2 7.0 9.1 7.8 Machine-building 45.0 40.7 55.6 44.7 Timber, wood, and paper 8.8 7.9 4.8 3.4 Building materials 4.5 4.7 1.4 1.4 Light 8.4 8.0 6.7 5.0 Food 7.1 8.3 1.0 1.0 Other 4.3 2.6 0.7 1.5 cant underrepresentation in food, building materials, and ferrous metal- lurgy. The main bias that results, however, will undoubtedly come through the overrepresentation of machine-building. Between 1991 and 1994 roughly two-thirds of the employment contraction in Russian indus- try has been concentrated in machine-building. This suggests that the em- ployment results reported below will tend to overstate the adjustment. For the other branches, the change in relative shares in both the Goskom- stat data and the survey are roughly consistent (figure 2-2). We have already signaled the partial adjustment of employment with respect to output. Figure 2-2 goes a bit further by showing the clear corre- lation between the size of shocks to output and those to employment, but it confirms the apparent lag in employment adjustment in relation to out- put. Figure 2-3 uses official data to show that branches with large relative output declines have also tended to experience some deterioration in their relative wage. Figure 2-4, however, also indicates a strong inertial compo- nent in the wage setting. Relative wage levels and ordering at the start of transition have shifted surprisingly little, although there is clear evidence of the emergence of a wider wage range. In short, there is some evidence that the relative performance of branches has had some playback to wages, but by 1994 relative wages had not moved that significantly. How Rtussian Firms Make Their Wage and Emnployment Decisions 19 Figure 2-2. Change in Employment Related to Change in Output, by Industry, 1990/91-1994 Employment change, June 1994 to January 1991 40% 3% Energy 300/- 20% - Fuels 10% B l Chemical/ Nonferrous metals 0% _ petrochemical ClFood -10% IIIron Construction -20% - Light industry ! Industry -30% - Machine tools m Timber -40%l l l l l l l -80% -70% -60% -50/ -40% -30M -20% -10% 0% Output change, 1994/1990 (December 1990 as base) Figure 2-3. Change in Output Related to Change in Relative Wage, by Industry, 1990/91-1994 Nominal relative wage change, June 1994 to January 1991 0.55 Energy- Fuels 0.45 - 0.35 - Nonferrous 0.25 _ Food metals 0.15 _ Construction 0.05 - 'n Iron Chemical/ -0.05 - petrochemical _ - All industry -0.15 - U Timber -0.25 - Machine tools ,- Light industry -0.35 _ ,, , -80% -70% -60%, -50% -40% -30% -20% Output change, 1994 (1-2Q) to 1991 (December 1990 as base) 20 Employment, Wages, and tlze Provision of Social Benefits Figure 2-4. Monthly Wage by Branch, June 1991 and June 1994 Wage in June 1994 (rubles) 450,000 Energy -_ Fuels 400,000 - 350,000 - Nonferrous metals 300,000 - Food 250,000 - All industry Iron and steel 200,000 _ Chemical _~ ~Construction 200,000- 150,000 Timber 100,000 - Machine tools ! Light industry 50,000 - 0 l l l 400 450 500 550 600 650 700 750 Wage in June 1991 (rubles) The very large shocks to output reported in official series and the lagged employment adjustment have obviously reduced labor productiv- ity sharply. This is, however, sensitive to correction for hours adjustment, because significant numbers of Russian firms have placed workers on short time and involuntary leave. In the first quarter of 1994 nearly 6 per- cent of the Russian labor force was subject to a short-time work spell, and an additional 8 percent to involuntary leave. Nevertheless, survey evi- dence has also shown that spells of short-time work have not necessarily been protracted (Commander and Yemstov 1995), and the estimates of labor productivity presented in table 2-2 make no allowance for hours adjustment. Table 2-2 requires some further explanation. In presenting our calculations of unit labor costs and their decomposition, we use a producer-price deflator based on output, rather than the unreliable offi- cial producer price series (Koen 1994). The result is the emergence of a significant deterioration in labor productivity over the period 1992-94. This did not lead to an increase in unit labor costs in 1992, largely because Howv Russian Firns Make Their Wage and Employment Decisions 21 Table 2-2. Decomposition of Unit Labor Costs in Industry, 1992-94 (percent) Aintnial rate of chanige Item 1992 1993 1994 Change in unit labor cost -35.9 7.2 12.1 Change in real consumption wage -28.2 -0.2 -12.4 Change in PCI/PPI -23.8 18.1 17.5 Change in labor productivity -16.1 -10.7 -17.2 From change in Y -19.3 -18.3 -26.7 From change in N 3.2 7.7 9.5 Source: Center for Economic Forecasting and World Bank. real consumption wages fell significantly. Further, in 1992 the wedge of consumption over product prices actually decreased. But in 1993 and 1994 unit labor costs for Russian industry increased as a result of the combined action of the wedge and productivity. In short, two-digit data tell a story of large output declines alongside considerable inertia in employment, but with some evidence of relative wages responding to relative output shock. Declines in real consumption wages and, at least initially, a decrease in the wedge led to a fall in unit labor costs. This has subsequently been reversed. The deterioration in labor productivity over this period would be reduced if hours adjustment were factored into the calculation. Firm Evolution: Survey Evidence Output Firms in the sample have generally experienced large declines in output and capacity utilization. For the total sample, output in constant prices by mid-1994 was barely 35 percent of the 1990 level, and capacity utilization had dropped from over 80 percent to about 50 percent in the same pe- riod. Table 2-3 shows some clear variation across ownership classes. Here the categories are determined by mid-1994 legal status. It can be seen that capacity utilization has fallen sharply and equivalently in both state and 22 Employmienit, Wages, and the Provisiont of Social Benefits Table 2-3. Descriptive Statistics: Mean and Coefficient of Variation for Firms Classyfication by ozwnerslhip, 1994 De novo State-owned Privatized Meall cv 71 Mean cv n1 Mean cT) n 1990 Output (1991) 21.6 223.4 9 8,201.4 712.1 68 1,705.2 635.6 189 Capacity (1991) 80.4 33.9 19 84.9 21.6 85 82.5 19.9 194 Employment 20 217 33 3,810 218 100 2,731 246 263 Sales 19.9 182.4 7 1,327.0 396.2 66 814.3 440.9 166 Wage 72.9 276.1 31 232.1 54.4 88 253.1 92.4 225 1993 Output 378.7 146.8 33 34,790.5 343.0 78 17,149.3 724.5 211 Capacity 74.9 29.9 39 62.7 37.9 86 62.8 37.3 205 Employment 81 130 43 3,260 244 93 1,944 327 240 Sales (1992) 72.3 186.3 25 4,477.8 329.1 75 1,959.3 465.9 181 Wage 45,214.9 99.1 37 34,327.2 90.2 93 35,416.2 70.3 229 1994 (1-2Q) Output 419.8 148.6 35 40,230.7 350.0 83 19,063.7 682.1 220 Capacity 71.3 36.8 39 53.8 48.8 87 49.8 52.8 206 Employment 97 152 43 3,079 251 94 1,878 332 240 Sales 379.4 135.4 38 28,316.4 396.6 82 10,902.8 584.5 215 Wage 186,256.1 61.6 38 159,982.0 63.4 94 154,096.4 56.2 233 Note: Unit: output and sales, million rubles; wage, rubles; capacity, percent. Souirce: World Bank survey. privatized firms. Two-thirds of the latter were involved in the mass pri- vatization program, which occurred mostly in 1993/94. De novo private firms, by contrast, had fairly stable capacity rates. Output losses have been distributed over branches in a manner quite closely replicating ag- gregate data. Branches-such as light industry and capital goods produc- ers-have registered particularly profound drops in output. These output losses have been associated with some degradation in profitability, although this remains far more ambiguous given the inter- action of obscure accounting practices, selective opportunities for firms to exploit market power, and continuing subsidy flows. Indeed, the survey results indicate that between 25 and 30 percent of firms received subsi- dies through the federal budget between 1992 and 1994, although at sig- nificantly lower real levels toward the end of the period. Hotv Riussian Firnms Make Their Wage and Employmient Decisions 23 Employment Employment clearly remains high in relation to output nearly three years after the start of transition. For the full sample, employment at mid-1994 had declined by around 25 percent from 1990 levels. The underlying -large increase in employment for each unit of output provides a simple but striking measure of the continuing employment bias in these firms. The distribution of employment changes is quite revealing. Average employment in de novo private firms has increased more than fourfold, even though the share of de novo private firms in total employment re- mained considerably below 1 percent by mid-1994. Firms privatized by 1994 cut employment by an average of over 30 percent between 1990 and 1994. The average decline in state firms was significantly lower, around 20 percent (table 2-3). While firms with larger negative shocks have in- duced larger employment adjustments, the association has remained quite weak, a feature we explore more systematically below. The laigest firms in the sample-the bulk of which remained in the state sector-have actually increased employment slightly despite large shocks to output! The response on employment is mildly sensitive to market structure. Firms that reported no competition to their main product clearly had lower net changes to employment, and their share of layoffs in total sepa- rations has been lower than in competitive firms. As might be predicted, firms that face significant foreign competition also appear to have experi- enced larger net adjustments to employment, with more layoffs and em- ployment changes that are more sensitive with respect to output than other firms. Nevertheless, the differences across firms classified by mar- ket position is generally not that significant. What is also striking is that involuntary separations, particularly in- volving large layoffs, have been very infrequent. No more than 15 per- cent of the sampled firms reported an involuntary element in total separations exceeding 20 percent. In these cases, layoffs were primarily attributed to financial constraints and lack of demand for firm products; restructurirng and associated changes in product mix were relatively minor explanatory factors. The absence of large-scale layoffs can also be attrib- uted to nontrivial adjustment costs, with compulsory notification and severance. More generally, however, less than one-third of firm manage- ment considered employment reductions to be a high priority, and ex- plicit priority for plant closures and more drastic restructuring measures 24 Em1]ploym1enit, Wages, and the Provision of Social Benefits were accorded even less importance.2 The obvious impression is that Russian firms have shed labor only in extremis, and rarely as part of a conscious restructuring program. Gross Flows The striking differences between Russia and Eastern Europe appear to in- clude not only the small net reduction in employment stocks, but also the high gross flows between firms. Job-to-job transitions have been large, particularly in the major urban labor markets such as Moscow and St. Pe- tersburg (as reported in Commander and others 1993). Unfortunately, the survey does not allow us to get a precise picture of flows, but it does at least permit us to relate hirings and separations that occurred during 1992-94 to the stock of end-period employment. While over 80 percent of firms saw no major hiring in this period-an outcome reasonably com- mon across sectoral categories-it is also clear that incremental hiring has continued.3 Figure 2-5 relates the separations rate to the hiring rate over the period 1991-94. While for most observations the separations rate is significantly superior to that for accessions, it is notable that larger firms-the size of the individual circle scales for employment-tend to be distributed along or below the 45-degree line. While their gross flows re- semble those of other firms, larger firms appear to have far smaller changes in the net. There is also significant churning across the spectrum of firms. There appears to be no predictable relationship between hiring and decisions consistent with restructuring. Thus, relating the change in in- vestment to the hiring rate, we find no predictable association. While firms with positive investment had the highest hiring rate, it was not sig- nificantly different from that of firms where investment had collapsed by more than 50 percent! In short, restructuring decisions, including those on skills, appear not to be the dominant motivation behind hiring. Labor Hoarding In this environment of high initial employment and large gross flows, it is hardly surprising to find that excess employment remains common. As table 2-4 makes plain, over 20 percent of the sample acknowledged that labor hoarding was above 10 percent of their 1994 employment level, and How Ruissian Firms Make Their Wage and Employment Decisions 25 Figure 2-5. Scatter of Hiring Rate to Firing Rate (1991-94 to 1991 Employment) Firing rate, 1991 .921729 - 0 0 0 o .691 - . O . .230 - ' ' 0 y 00 * .46 0 - I;-= I _I0* 0 .240 .481 .721 .96159l Hiring rate, 1991 a further 25 percent reported excess employment in the range of 5-10 per- cent. But while 45 percent reported employment to be roughly at the right level, it is revealing that over half of these firms had significant numbers of workers on short or part time, an indication that their estimates of ex- cess labor may be biased downward. Labor hoarding is clearly negatively associated with changes to out- put, and positively associated with firm employment size (table 2-4). In- deed, relating so-called optimal employment-as judged by firm management for current output-to its actual level, the great bulk. of firms were significantly below the diagonal, and this generally increases with firm size. It is striking, however, that the management of the largest firms perceived employment levels to be less distorted, even though their output shocks have been large. Classifying by ownership status makes clear that excess employment was distributed with reasonable common- ality in both remaining state firms and in firms that have either been pri- vatized or are in the process of privatization. By contrast, over half of the Table 2-4. Firms Classed by Excess Ernployment, 1994 Output Output Ermployment Layoffs/ Part-time/ Unpaid 1994-93 1993-90 1994-90 Mean Capacity separations employment leave Wage Class in 1994 (%) (%) (%) employment (%) (%) (%J (%) (000 rubles) High (> 10%) -41 -23 -30 2,323 46 17 20 10 734 n =88 Moderate (5-10%) -32 -35 -22 3,215 54 7 10 6 696 n = 110 Minor (employment right) -24 -36 -8 1,339 55 7 57 3 745 n = 196 Negative (employment low) -23 -35 -12 752 61 29 16 8 887 n = 35 Sonirce: World Bank survey. How Ruissian Firms Make Their Wage and Employment Decisions 27 private firms believed that their employment levels were correct, and over 20 percent considered employment to be too low. While the great majority of firms with excess labor expected to reduce employment further by early 1995, it is striking that at least one-third of firms with excess labor at mid-1994 projected that the surplus would be either the same or higher in the future. Of interest is that labor-hoarding firms emphasized an expected output and demand recovery as the prin- cipal factor sanctioning hoarding, while of almost equal weight was be- nevolence, with nearly one-third of respondents citing "social and ethical" reasons for not reducing employment. By contrast, worker oppo- sition and employee opinion voiced through share ownership in privat- ized firms were of negligible significance. This bolsters the point that the employment bias is a clear choice of the coalitions that govern firms, and one that has thus far appeared to be weakly disturbed by changes in legal form. Finally, firms with high or moderate labor hoarding appear not to have experienced anything more than a slight deterioration in the wages of their members compared with firms without perceived labor hoarding. Wages Associated with these employment decisions have been those on wages. Aggregate wage series show that real consumption wage levels have re- mained low and fairly stable since 1992, although this ignores the in-kind component of compensation. The survey data clearly support the view that monetary compensation levels are low. Indeed, at least 20 percent of wage observations at mid-1994 fell below the comparable regional pov- erty line. Table 2-5 reports a probit estimation relating a below-pove.rty- line wage to firm and other attributes. Firms in branches with large negative shocks-light industry, machine-building, timber, and erter- prises in the military-industrial complex-tend to pay low wages. This is also true for firms located in the south of the country. Absence of invest- ment and low wages are positively associated, while in terms of owner- ship, low wages appear to be more probable in firms where insiders dominate; the reverse is true for private firms. Neither of the size vari- ables are significant. And a low wage is negatively and significantly asso- ciated with the change in firm-specific revenues, which suggests that wage settlements are predictably linked to firm financial performance. an issue we return to in more detail below. 28 Employment, Wages, and the Provision of Social Benefits Table 2-5. Probit Estimation Relating Below-Poverty-Line Wages to Firm and Other Attributes Dependenit variable Wage belowv the poverty line DlTnln1J varioble Private -0.3783* Insider worker control 0.3652** Big -0.0821 Small 0.1802 Light industry 0.9368*** Machine-building 0.5609** Wood and paper 0.8139** Military-industrial complex 0.5405* Siberia -0.8348** Moscow/St. Petersburg -0.1204 South Russia 0.8585*** No investment 0.5480*** Change in sales revenue -0.2359** Constant -1.5873*** I7 265 Note: Firms with wage under the poverty line 54. Significant at 1%, ***; significant at 5%., ",- significant at 10%, *. Source: World Bank survey. Relating wage changes at the firm level to changes.in employment, we find some evidence for a positive association-firms with relatively large net employment reductions have experienced some deterioration in rela- tive wages, controlling for region. Further, wages that were below the poverty line were generally reported for firms with larger than average employment reductions. Firms with higher shares of layoffs to total sepa- rations were marked by lower wage levels. In addition, using aggregate data to look at regional wage evolution suggests that a conventional in- verse association between regional employment rates and wage changes has emerged. These features would appear to indicate a growing respon- siveness of regional wage setting to a regional activity variable (Com- mander and Yemtsov 1995). Figure 2-6, however, qualifies any presumption that we have been witnessing a sharp increase in wage dispersion as a result, at least in terms of skills.4 The main impression is that with these broad skill catego- ries, relative wages have displayed considerable inertia. If we look at rela- How Ruissian Firms Make Their Wage and Employment Decisions 29 Figure 2-6. Log of Average Wage by Groups (n = 333) Log of nominal wage 6 * 1994 1993 ] 1992 M 1990 (prereform) 5 4 3 1 2 0 _ Firm wage Workers' wage White collar wage Managers' wage tive wages by ownership status, however, we do find a clear improve- ment in relative wages in private firms over the period 1990-94. By 1993/94 the average private money-wage premium over privatized and state firms was between 20 and 35 percent, respectively. Nonetheless, this differential would be largely eliminated if nonmonetary components of compensation were factored in, because private firms supply fewer non- monetary benefits than other firms. Decisionmaking in Firms The preliminary indicators reported above suggest that there are ins titu- tional features, likely associated with the continuing dominance of insid- ers, that appear to support an employment bias with a reasonably common wage outcome. We can think of possible sources of insider influence in firm-level de- cisionmaking in both the depth and degree of influence in the wage and 30 Employment, Wages, and the Provision of Social Beniefits Table 2-6. Relative Influence in Decisionmaking, 1994 (percent) Worker Maniagement Currenit wvages anid wages and Financial operationisa employment employment decisionsb Agent Little High Little High Little High Little High Management/Board of Directors 7 80 11 76 9 78 5 86 Manager-shareholder 13 66 16 65 15 62 11 70 Worker agencies in firm 64 8 46 14 63 9 51 13 Outside trade union 94 1 86 3 93 2 92 2 Worker-shareholder 71 6 68 8 74 5 58 12 Outside individual shareholder 88 5 91 2 91 3 83 8 Outside institutional shareholder 83 5 87 2 85 4 72 10 Local government 81 4 81 3 82 3 77 6 Federal government 74 9 86 3 84 3 77 9 Banks 77 8 92 3 93 3 83 6 Note: Expressed as share of respondents (%o); residual shares ascribed to "moderate influ- ence." a. Sales, production. b. Allocation of profit, investment. Source: World Bank survey. employment setting. For the former, we can assume that employment and wages, given decisions on layoffs, are jointly bargained. We now need to understand more about the degree of insider influence and its distribution over workers and managers. Table 2-6 presents a ranking of influence for different actors over the kinds of decisions made at the firm level. It reflects the responses of man- agers, and as such it may be biased. While managerial discretion in deci- sions on worker employment, operational decisions, and managerial pay and employment clearly prevails,5 workers inside the firm clearly do ex- ert some significant influence on employment, wages, and profit alloca- tion in at least half our sample. Their weight in decisionmaking quite obviously exceeds that of outside shareholders, individual or institu- tional, let alone that of financial institutions and government agencies. How Russian Firms Make Their Wage and Employment Decisions 31 Wage decisions appear to be conditioned primarily by firm-specific fi- nancial variables, liquidity and profitability, but also appear sensitive to constraints imposed by the excess wage tax rule and the need to pay com- petitive wages to maintain attachment. But the most important considera- tion is the explicit association of wages with consumer price chaniges. This effective indexation also partially accounts for the absence of large movements in relative wages. In bargaining, however, it appears that workers' explicit interference in setting wages is largely absent. Around two-thirds of respondents considered worker demands through trade un- ions or collectives to be of no importance in the wage setting. Indeed, de- spite significant union presence, we find a clear absence of militancy; strikes were reported in only four cases, with a threat of industrial action in only twenty-four cases.6 Bargaining appears to be largely cooperative. The presence of a relatively common wage-setting procedure suggests that firm-specific differences in bargaining will largely be reflected in em- ployment. An obvious starting point is to try to control for ownership, given the scale of the privatization that has been completed, and to see whether privatization has been associated with any change in behavicr. Ownership Effects We now ask whether the employment bias has been affected by privatiza- tion. The dominant privatization route in our sample has been through Option 2-where workers and managers received 51 percent of voting equity-but all channels under the mass privatization have effectively yielded the same outcome: insiders have come to control between 51 and 68 percent of shares, levels higher than those assigned by law. Outside shareholding has averaged 21-23 percent, with government retaining the. residual. Government shareholding appears to be more concentrated among the larger firms as measured by employment. These share distri- butions closely reflect the original allocations; by mid-1994 little redistri- bution had occurred. We are interested in separating out the influence of share distributions on labor decisions. We hypothesize that greater outside shareholding could likely translate into larger employment adjustments, including in- voluntary separations, given the initial conditions. By contrast, in firms with delayed privatizations or where the government share has remained large, we could imagine a far weaker effect. Similarly, insider-dominated 32 Employment, Wages, aid the Provision of Social Benrefits Table 2-7. Employment Stability by Ownership Type Total separations, 1991-94, relative to employinent in 1991 Categorn Mean Stanidard devliation Privatizers Option 1 0.55 0.24 Option 2 0.51 0.27 Option 3 0.42 0.19 Lease 0.52 0.25 Other 0.61 0.26 Always private 0.61 0.26 State 0.59 0.24 Source: World Bank suLvey. firms could be expected to give priority to the interests of incumbents with, again, possibly higher employment stability. We would also expect these differences to spill into expected behavior or priorities. The first exercise we undertake is to look at whether share distribu- tions yield different employment outcomes. We construct a simple meas- ure of employment stability. For firms that existed in 1991, we relate total separations in the interval 1991-94 to the initial employment stock to get a crude probability of survival at that firm. While the measure is clearly biased-it excludes any reemployment possibilities-table 2-7 clearly tells us two things. First, share structure is rather uninformative about the employment survival probability. It does appear that Option 1 privatiza- tions-where workers received 25 percent of nonvoting shares for free- have a larger survival probability, as does being attached to a state- owned firm, a result that we could expect. Second, these probabilities are generally quite low, again highlighting the importance of flows. Repeat- ing this exercise and controlling for initial legal status of the firm yields similar results. We now go further, and ask whether share allocations make any dif- ference in the kind of separation found. Our working hypotheses are that firms with larger outside shareholdings will have experienced higher shares of involuntary separations, because outside influence induces re- structuring and amendment of the skill mix. By contrast, firms with de- layed privatization or large government shareholding will be more H-ow Russian Firms Make Their Wage anid Eniployment Decisions 33 Figure 2-7. Outsiders' Share and Ratio of Layoffs to Separations Outside shareholders (O/o) 100-. 75- 50 o 0 0 25 oC 0 7; _ o 25 50 75 100 Layoffs to fires inertial. To test the effect of outside influence, figure 2-7 relates outsiders' shareholding distribution to the share of layoffs in total separations. Each circle represents an observation, and the size of the circle in the figure represents a scaling by 1994 employment. The scatter shows no clear rela- tionship.7 Table 2-8 tries to control for possible delays in privatization, measured in the change in the Board of Directors.8 Privatized firms that have not yet had their first shareholders' meeting not only report a lower separations rate and lower involuntary separations compared with other privatizers, but also greater inelasticity in the change of employment in relation to output over the period 1991-94. There appears to be some evi- dence that changes may be dependent on the timing of the first share- holders' meeting, and hence we should control for lags. The table also reports the result of repeating the same exercise and controlling for a high government shareholding. Firms with high relative government shares have slightly lower involuntary separations, but overall there are no sig- nificant differences. 34 Employmenit, Wages, and the Provision of Social Benefits Table 2-8. Employment Changes: Delayed Privatizers and High Government-Shareholding Firms Compared with Other Privatizers High gov1erntnitt- Delayed Otlzer sliareXloldiiig Otlier Ez>iployn11ent cliange privatizers privatizers firmts privatizers mean values Ratio Layoffs to separations 0.050 0.094 0.090 0.098 Layoffs to employment 0.079 0.133 0.105 0.162 Net employment change -0.158 -0.103 -0.102 -0.100 Employment change to output change 0.069 0.160 0.163 0.142 Source: World Bank survey. Finally, we try to relate strategic preferences to share distributions in an attempt to overcome the problem of lags and to see whether the share- holding structure tells us anything about forward-looking managerial ob- jectives. Table 2-9 presents a correlation matrix for different shareholders and their ranking of employment and wage objectives. The main differ- ence that emerges is that insider shareholding is negatively associated with employment-reducing strategies, as well as with hiring. Outside shareholding appears to be negatively associated with a wage-enhancing strategy. But overall, the size of the coefficients suggests that any predict- able influence from different kinds of shareholders remains weak. This is Table 2-9. Correlation Matrix between Labor Orientation and Shares of Different Actors (n = 227) Shzare Option Governnment Insider Outtsider Reduce employment 0.028 -0.103 0.058 Increase employment 0.051 -0.098 0.090 Reduce wage 0.100 0.068 0.031 Increase wage 0.033 0.045 -0.119 Source: World Bank survey. Howv Rtussian Firms Make Their Wage and Employment Decisions 35 further confirmed by looking at the influence of shareholders on deci- sions concerning management, including the employment and compen- sation of managers. Our simple prior is that outside influence would tend to translate into greater control over managerial appointments and pay. This is not confirmed by the data. Outside influence is very attenuated, and it is no greater in firms that have changed their Board of Directors. Rather, inside managers are easily the most important actors in these de- cisions. Further, managers are clearly converting such dominance in deci- sionmaking into dominance in shareholding. Outside shareholders exert little apparent influence on managerial variables.9 In summary, changes in ownership as measured by share distribu- tions generally suggest little difference across categories. Outside influ- ence is notably absent. Firm Objectives To get a better idea of the degrees of bargaining power, we need :lo un- derstand the objectives of the main parties in the bargain-in short, man- agers and workers. Table 2-10 provides some indication of the weight of specific goals for the current management. It also allows us to contrast current objectives with those before transition in 1990/91. One thing is clear: a history of attaching great weight to an objective is a very good predictor of current objectives, and giving high importance to worker welfare and/or employment is, and was, common. Indeed, by 1994 Table 2-10. Firm Objectives, 1994 and 1990/91 (percent) Of some Unimportant importance Important Obiective: increase/mainitain 1994 1990/91 1994 1990/91 1994 1990/91 Sales 8 15 13 15 79 70 Employment 30 30 33 30 37 40 Worker income/welfare 7 12 23 28 69 60 Profit 7 17 13 22 80 61 Privatization 45 61 17 10 38 29 Shareholder dividends 32 81 29 9 39 10 Note: Expressed as share of respondents for that year. Solurce: World Bank survey. 36 Employmenit, Wages, and tJhe Prov7ision of Social Benefits roughly 70 percent of respondents cited worker welfare as a major objec- tive, a clear increase over 1990/91. Stabilizing or increasing employment was cited as unimportant in only slightly over 25 percent of cases. Even so, increasing profits and sales as objectives had clearly become more im- portant by 1994. At the same time, for managers of profitable firms, allo- cations of profits to productive fixed investments and/or increasing working capital were the main priorities. Raising wages or bonus pay- ments, or actions possibly consistent with short-run decapitalizing behav- ior, were given a minor weight.10 Perhaps most significant is the overlap in objectives. Thus, for firms attaching maximum importance to profits-a ranking of 3-a (3,3) pair with worker incomes occurred in 73 percent of cases, a (3,1) pair in 6 per- cent of cases. For profits and employment, a (3,3) pair was found in 38 percent of cases, a (3,1) pair in 30 percent of instances. When giving prior- ity to sales, (3,1) pairs with employment or worker income were still less frequent. Table 2-11 presents a simple correlation matrix of goals. Giving prior- ity to sales is more tightly correlated with the employment and income objectives than with profits. The latter objective is very weakly correlated with the other goals. In particular, giving priority to profits negatively correlated with giving priority to employment. This is quite under- standable, because jointly maximizing sales and worker welfare would be a goal quite consistent with the former system. Profit maximization is clearly different, but few firms appear to accord priority to profits at the expense of other worker-related interests. The picture emerging from the above discussion is that while revenue, even profit, maximization is an important priority for most Russian Table 2-11. Correlation Matrix among Firm Objectives (n = 418) Inicreasel/mainitain Objective: increase/mnainztaini Sales Emlploynment Worker inconme Profit Sales 1.000 Employment 0.271 1.000 Worker income/welfare 0.197 0.344 1.000 Profit 0.121 -0.031 0.158 1.000 Sou7rce: World Bank survey. How Russian Firms Make Their Wage and Employment Decisio;ns 37 firms, the welfare of workers in both the wage and employment dimen- sions enters quite explicitly into their objectives. Explaining Insider Influence We now turn to look in a bit more detail at the characteristics of firms where worker objectives in income and employment are related to firm-specific attributes. We do so by relating the declared objectives of firms and the consistency of those objectives to firm characteristics. For this exercise, the dependent variable is variously the importance given by each firm to objectives-profit, sales, shareholder dividends, and the like-in 1994 and at the beginning of transition. This allows us to pick up a time dimension as well. We try to ascertain what happens to other objectives when either a high or low importance is attached t:o em- ployment or worker's welfare/income. The estimation is by ordered. logit, where the ordinal values of importance of the objectives are regressed on a vector of firm characteristics. The categorical variables on the right- hand side are the industrial branches (compared with "other industries"), the area including the cities of Moscow and St. Petersburg (compared with the rest of Russia), the ability to generate profit, the existence of price controls/fixed profit margins for the firm, the vintage of the tech- nology used by the firm (compared with machinery older than thirty years), the existence of government financial support in 1994, the share of the labor force organized in new trade unions (compared with firms that do not have any new trade unions), and market dedication (greater than 50 percent) to the state sector both in the acquisition of inputs and the sale of output. Table 2-12 provides the main results for the objectives of employment, worker welfare/income, sales, and profit. The regressions are followed by predictions of the probability associated with each level of importance of the stated objective if the firm simultaneously holds the objective of employment and worker welfare in polar importance (very high or very low). The industrial branch of the firm generally has little effect on the objective of maintaining employment, except in heavy and agricultural machinery. In these branches, the effect is positive and the worker wel- fare objective is also positively affected. These two branches and metal- lurgy give a negative effect on the objective of increasing profit and sales. It is interesting to note that if the firm belongs to the military-industrial 38 Employment, Wages, and the Provision of Social Benefits Table 2-12. Ordered Logit Estimation Relating Firm Objectives to Firm and Other Attributes Dependent variable:firmn objectives Worker Employ- velfare! Independenit variable mient incomite Sales Profit Industrial branch Energy ** Fuel Metallurgy ** Chemicals ** Heavy machinery * ** Machine tools Automobile +* Agricultural machinery +* +* Military-industrial complex ** Wood and paper * Construction material +* Light industry Agroprocessing Profit-maker Price control Government support in 1994 * Moscow/St. Petersburg + +** Market dedication to state sector Inputs + Output +* +* +* _** Vintage of technology > 0-10 years +* +* +* > 10-30 years ** Participation in new trade unions > 0-50% * 50-99% ** +** 100% ** ** ** Categories Observed probability Not important 0.29 0.08 0.08 0.07 Some importance 0.34 0.23 0.13 0.13 Important 0.37 0.70 0.80 0.80 n 377 382 383 380 Predicted probability of objective Sales Profit being important Worker objectives important n= 141 0.80 n 133 0.77 Worker objectives not important n 24 0.74 nt 23 0.80 Note: + , positive coefficient; -, negative coefficient; '**, significant at 1%; **, significant at 5%; ', significant at 10%. Source: World Bank survey. How Ruissian Firms Make Their Wage and Employmient Decisionts 39 complex, the effects on worker welfare, profit, and sales are negative. Re- ceipt of government support in 1994 (which is very closely correlated to receipt of such support in the previous years) has a positive effect on em- ployment, but it seems to be unimportant for the other objectives. The market dedication (of output) to the state sector is positively related to employment, worker welfare, and sales. The location of the firm in or around a metropolis has a positive effect on the objectives of worker wel- fare and sales, as expected, but the profit objective is negatively affected and employment is unaffected. For all the objectives, new technology has a positive, significant effect. We have already noted the coexistence of multiple objectives. We now predict the probability associated with each level of importance of the stated objective of maximization/maintenance of sales and profit if the firm simultaneously considers worker objectives of high/very low im- portance (poles). When the profit objective is ranked high, employment and worker welfare objectives will tend to be of low importance. By con- trast, the predicted probability of holding sales in high importance rises in step with high importance being attached to worker objectives. From this exercise, we can also reasonably conclude that inertia is a very irapor- tant factor for these firms in ordering their priorities. If one attempts to predict future from current behavior, objectives for sales, employment, and worker income are likely to move in tandem. By contrast, where pri- vatization and profit are given high importance, employment and worker income will be given a lower weight. Bargaining in the Firm The discussion thus far has indicated that Russian industrial firms have begun to adjust, but that this process has been halting and generally invari- ant to changes in ownership. Managers appear for the most part to be the dominant players, with outside stakeholders exerting marginal influence. But workers' bargaining power inside firms is far from inconsequential, even if it appears to be largely implicit. A clear employment bias has been maintained, which can be partially explained by the specific objectives of the firms, but also needs to be related to the respective threat points of the main agents in firm bargaining. Given that insiders-managers and workers-clearly must bargain over both wages and employment, we now look more directly at the resulting wage-employment combinations. Our starting assumption is that the manager will try to pursue profits, constrained by the influence of incumbent workers."1 Workers are as- 40 Employmenit, Wages, anid the Provisioni of Social Benefits sumed to act as a collective, maximizing the aggregate utility of its mem- bers. Our framework corresponds to the standard two-party Nash bar- gain, where the players are the firm (manager) and a workers' collective. The objective function for the firm is the Nash product: (2-1) Z=( - C-C) where C = Nl(zv) + ( M - N ) it (zi ). The X is profit and n' is the fall-back that the manager can get in the absence of agreement. C is the total utility of workers, and C* the fall-back for workers. The feasible set for bargainers is given by the value function, V(N) where V' (N) 0 and V" (N) <0 and profit is defined as: it = V(N) - w (N). The workers' function is increasing in both wage and employment, in- dicating the potential for them to select a higher probability of employ- ment for a lower wage. In the event of disagreement, and most generally, we assume that the manager might get some income outside the firm, or even within, but without the cooperation of the workers. This may also be true for the workers; they could continue, for example, to pro- duce, but without the manager. The disagreement point is then given by: d = [ i', 6s + Mu ( w ) ], where 6 is the premium over outside income, and the objective function is: (2-2) Z={ V(N)-zvN- tj {N[u(w)-i(ibfl-6)j. Maximizing this equation in terms of employment and wages, we get the first-order conditions: (2-3) (2-4) 5Zl8N=[Nui'(zu,)] [7r-7u*]+[C-C-] [-N]=O, and hence the expression for the contract curve (2-5) -[u(w)-i(r) ]/[Nu'(wu)I={V'(N)- w]/N, which will be upwvard-sloping in wage-employment space, with the slope given by: Howv Ruissian Firms Make Their Wage and Employment Decisions 41 (2-6) 8zv/[N=[ V" (N)]/[I[it(w)-l(ib) Ii"(w) /[u (w)]2 I- The framework is familiar and flows out of the efficient contracts lit- erature (McDonald and Solow 1981; for an application geared more to transition issues, see also Commander, McHale, and Yemtsov 1995). Looking at figure 2-8, we can imagine that at the start of transition., Rus- sian firms would have been to the right of even the average product curve, with employment at Nt. Subsequent shocks to value added, such as a decline in demand, would have further decreased the marginal value product of labor, shifting the labor demand curve inward. In respo:nse to these negative shocks and to ainounced tightening in the budget con- straint, we can think of firms being forced-at a minimum-to move onto the average product curve. The challenge is to get some idea of the wage- employment pair consistent with zero profits that arises as a result of relative bargaining powers and institutional setting. Given that we are unable to identify directly the wage-employment combinations, an alternative approach is to look at some simple stylized relationships and to explore their time dimensions. Accordingly, we run Figure 2-8. Labor Market W Nt o =f '(N) Marginal value \product curve Contract Average value curve N* N1 N 42 Employment, Wages, and the Prozision of Social Benefits three sets of discrete regressions looking at employment, wages, and the wage markup over the outside opportunity. In particular, we are inter- ested in the responsiveness of these variables to a measure of financial performance-in this case, sales.12 Two-digit producer prices were used as deflators. Table 2-13 reports the results of our basic employment equation. Here we relate the change in the log of employment over two periods- 1994/93 and 1994/90-91-to the change in the log of sales and to a vector of characteristics, including profitability, price controls, firm size, owner- ship at the end of the period, receipt of government transfers, and branch dummies. The coefficients on the sales terms are significant, positive, and small. In the long panel, the employment elasticity was no larger than 0.09, while for the short panel it was 0.03. Being situated in the major ur- ban areas of Moscow and St. Petersburg exerted a positive effect. Gener- ating profit, facing price controls, and the size dummy exerted a small positive but insignificant effect, while receiving government transfers ex- erted a predictably negative influence. We also experimented with a Table 2-13. Regression of the Change in Employment Dependenit variable: chanige in employment (log) Independent variable 1994-90/91 1994-93 Change in sales (log) 1994-90 0.092*** 1994-92 0.031*** Dummy variable Profit-maker 0.002 0.043 Price control 0.165** 0.038** Size (> 3,500) 0.172*** 0.043* Moscow/St. Petersburg 0.102 0.037 Government support in 1994 -0.057 -0.036* Privatized -0.326** -0.122*** Delayed privatizer -0.292*** -0.099*** To be privatized within 18 months -0.209 -0.104** De novo private and industrial branches 0.025 0.155*** ii 196 252 Adjusted R2 0.4 0.27 Note: ***, significant at 1%/o; *, significant at 5%; *, significant at 10%. Soutrce: World Bank survey. How Rtussian Firms Make Their Wage and Enmployent Decisions 43 measure of new unionization, but this was consistently insignificant. Of obvious interest is that in relation to state firms, privatized or prive[tizing firms had negatively signed coefficients on the sales tern, indicating again the apparently weak impact of privatization-or its prospect-on firms' employment behavior. By contrast, de novo private firm status had a positive and significant effect. The branch dummies indicated that em- ployment adjustments have been relatively large in firms in the military- industrial complex and in capital goods and light industry. In short, while we find evidence of some association between changes in employment and sales at the firm level, this association has remained weak. This con- firms the picture that we had assembled from the descriptive statistics. We now turn to the wage setting. The results from the wage equation are reported in table 2-14. Again we are interested in the responsiveness of wages to firm-specific financial characteristics. We therefore include a productivity or sales per worker variable, which we take to be a proxy for the firm's ability to pay. Looking at the results from the long panel--1994 to 1990/91-we can see that the coefficient on the sales variable is again Table 2-14. Regression of the Change in Wage Rate Dependen7t variable, change in wage rate (log) Independent variable 1994-90/91 1994-92 Change in sales (log) 1994-90 0.083*** 1994-92 0.102*** Dummy variable Profit-maker -0.418*** -0.137 Price control -0.237*** 0.169 Size -0.047 -0.381*:S Moscow/St. Petersburg 0.133 0.173* Government support in 1994 -0.226* 0.090 Privatized -1.234*** 0.017 Delayed privatizer -1.220*** 0.158 To be privatized within 18 months -1.122*** -0.140 De novo private and industrial branches -1.021** _0.455* n 172 215 Adjusted R2 0.8 0.34 Note: '**, significant at 1%; **, significant at 5%; ', significant at 10%. Source: World Bank survey. 44 Emiploymetiz, Wages, and the Provisionz of Social Betnefits small, at 0.08. There is some sign of increasing responsiveness over time, however; the coefficient for the 1994/93 estimate was 0.1. In the wage de- termination, being profitable was consistently negatively signed. Over the long panel, firms that received financial transfers from government had lower wage-to-productivity elasticities, but this effect had disap- peared by 1994/93. Being located in Moscow and St. Petersburg exerted a positive effect. It is striking that the ownership dummies indicate that over the long panel, nonstate status at the end of the period had a very significant and negative effect. By 1994/93 this had reversed. The owner- ship dummies were generalLy insignificant, although de novo private firm status continued to imply a lower sensitivity compared with state firms. Branch dummies were generally significant. Although we cannot jointly and directly identify the wage-employ- ment outcomes, these estimations offer a number of hints. Given the in- itial conditions, firms have been weakly responsive in both their wage and employment decisions to firm-specific measures of performance. This refers back to our earlier discussion and the apparent inertia in both employment and wages. On the latter, however, we should note that we are unable to account for wage arrears, which we know to have been an increasing phenomenon through this period. If we were able to factor in wage arrears, it is possible that the true sensitivity would be larger than we report, if indeed arrears were strictly associated with sales. Given our finding that despite substantial labor hoarding through the transition, firms have been reluctant to relate employment to sales changes, and that changes in wages have been quite weakly responsive to changes in productivity, an obvious question we must ask is whether this choice has been consistent with a hard budget constraint. Clearly the broader issue of whether this is a sustainable path of adjustment will de- pend on a range of factors, including the labor share, given the histori- cally low levels of monetary compensation in Russia. To look at how the wage per worker choice relates to firm-level per- formance, we directly observe the ratio of the actual wage to gross sur- plus per worker, as well as in the case of gross surplus per worker evaluated at the reservation wage.13 The reservation wage is assumed to be the branch-area minimum. Table 2-15 indicates that roughly 17 percent of firms had wages per worker in excess of gross surplus in 1993, but that this had declined to under 5 percent in 1994. For the great bulk of firms with positive values, the mean ratio was between 0.3 and 0.4 in 1993, fall- How Ruissian Firms Make Their Wage and Employmenet Decisions 45 Table 2-15. Ratio of Wage to Gross Profit per Worker, 1993 and 1994 1993 1994 n Mean a Mean zI/N Positive values 167 0.31 258 0.13 Negative values 33 -0.70 12 -0.21 il/N Positive values 164 0.42 256 0.12 Negative values 35 -0.71 13 0.14 Source: World Bank survey. ing to 0.13 in 1994. The full distributions are presented in figures 2-9 and 2-10, which report the ratio of the wage to gross surplus per worker evaluated at the reservation wage.14 The overwhelming majority of firms had a ratio below unity, which indicates that wage settlements were clearly constrained by the operating surplus. These ratios, even allowing for tax, are interesting in several respects. Despite a weak association in the changes between a measure of financial performance and wages or employment, few firms show any signs of de- capitalization through the wage setting. Predatory wage setting seerns to be the clear exception. The low responsiveness points instead to a con- tinuing willingness on the part of the insiders who dominate the vast majority of these firms to sanction a continuing low level of monetary compensation. This is quite consistent with aggregate data, whiclh re- vealed the apparent decline in the labor share since the start of transition, as well as the unit labor cost data that we presented in table 2-2. This also seems in keeping with a willingness on the part of insiders to give prior- ity to employment stability. It also implicitly signals a change in the structure of compensation, particularly i-n the larger firms, where the :non- monetary components of compensation have remained large. Finally, we look at how responsive the wage markup has been to firm performance. We ran an OLS regression relating the ratio of the wage to the reservation wage to the difference in the log of sales per worker over the long period 1994-90, as well as 1994-92. We again include brznch dummies and other characteristics (see table 2-16). Using transition prob- abilities for workers extracted from the Russian Longitudinal Monitoring Survey, we construct a reservation wage measure, factoring in the res;pec- 46 Employment, Wages, and the Provision of Social Benzefits Figure 2-9. Cumulative Density of WagelGross Surplus per Worker, 1994 Cumulative density 1~~~~~~~~~~~~~~~~ 0.75 0.5 0.25 0 -5 -2.5 -1 0 1 2.5 5 Wage/gross surplus per worker Figure 2-10. Cumulative Density of Wage/Gross Surplus per Worker, 1993 Cumulative density 1 - CO a 0 0.75 - 0.5 - 0.25 - -5 -2.5 -1 0 1 2.5 5 Wage/gross surplus per worker How Ruissian Firms Make Their Wage and Employment Decisionis 47 Table 2-16. Regression of the Ratio of Wage to Reservation Wage Dependent variable, wage/reservation wage Indepenident variable 1994/90 1994/192 Change in sales per worker (log) 1994-90/91 0.130*** 1994-92 0.127*** Dummy variable Profit-maker 0.399*** 0.417*** Price control 0.095** 0.097** Privatized 0.300* 0.266* Delayed privatizer 0.384* 0.343* Privatizing within 18 months 0.433** 0.379** De novo private and industrial branches 0.882*** 0.869*** n 190 197 Adjusted R 2 0.84 0.82 Note: '**, significant at 1%; -*, significant at 5%; ', significant at 10%. Source: World Bank survey. tive employment and unemployment probabilities.15 The coefficient on the sales term proved positive, very significant, and small: 0.13. Further, there was no change in the size of the coefficient over both the short and long panel. In general, branch dummies were not significant. In owner- ship, the main result is that both privatized and private firms were con- sistently positive and highly significant. Generating profit entered significantly and positively. The main result from this exercise is to con- firm a statistically significant association of the wage markup with changes in firm-level sales. The size of the coefficient is stable over tine. To summarize, in this section we made an attempt to think about how insider influence translates in the setting of wages and employment. We hypothesized that faced with large adverse shocks, Russian firms would have been forced to make adjustments, but that continuing insider bar- gaining power would translate into a wage-employment combination on the zero isoprofit curve, with wages thus set close to average product. Firms in this setting would simply be constrained by a zero profit con- straint, rather than by the labor demand curve. We experimented with some simple estimations relating the change in employment, wages, and the wage markup to the change in sales or productivity across severail pe- riods. A statistically significant relationship between the left-hand side 48 Employment, Wages, anid thie Provisioni of Social Benefits variables and sales or productivity was indeed identified. But the rela- tionship was rather weak for both wages and employment. Nevertheless, looking more closely at the evolution of wages per worker, using either reported or reservation values, we found that these were generally set well within the size of the per worker distributable surplus. This would, of course, be consistent with a maintained employment bias. Conclusions We have tried to get a better idea of the factors influencing the wage and employment decisions of Russian firms. This is of particular interest given the apparent divergence in behavior with regard to Eastern and Central Europe. Russian firms have now begun to cut into the chronic stocks of excess labor that they carried into the transition from the pre- vious system. There is evidence that the larger, mostly privatized, firms have yet to address the problem seriously. In short, privatization, as might be expected given its design and the weight given to insiders, has yet to feed through into substantive changes in employment. Aside from the reluctance to separate, it also appears that many firms, including those with acknowledged excess employment, continue to hire, with such accessions not obviously explained by strategic restructuring choices that require amending the skill mix. We have tried to think about this employment bias in terms of possi- ble insider influence at the firm level. The findings are far from clear-cut. Using qualitative evidence, we note that firm objectives have only par- tially changed. Indeed, the best predictor of current objectives were those preceding transition. Paradoxically, while workers have little explicit in- fluence in decisions, whether at privatized or other firms, and union pres- ence is not an important factor, responses by managers on firm objectives and relative influences in decisionmaking lead us to believe that workers continue to exert a subtle but important influence. Perhaps most impor- tant, the objectives of managers and workers appear in many critical as- pects to coincide. The weights given to worker employment and income by managers are symptomatic of this apparent harmony in interests. Thinking in terms of an underlying bargaining model, we then turned our attention to the identifiable factors governing firm decisions on wages and employment. We implemented simple estimations that looked at the relationship of employment, the wage markup, and financial per- How Russian Firms Make Their Wage and Employment Decisions 49 formance. The story that emerged was broadly consistent with our other results. Simply put, changes in sales or productivity do matter in wage and employment setting and in determining the size of the wage markup over the reservation wage, but the respective elasticities have remained quite small. Despite these small elasticities, firms have generally been quite effectively constrained in their wage setting by their revenues, and they do not appear to have operated as if in the presence of a soft budget constraint. Indeed, the share of firms with wages in excess of gross sur- plus declined significantly between 1993 and 1994. This suggests quite strongly that shocks from the product market have been partially accom- modated as a result of low monetary wage levels. Thus, low responsive- ness of wages to productivity changes has still been consistent with a wage-distributable gross surplus ratio significantly below unity. Insiders appear to have continued to select high employment and low monetary wage levels as the response to a deteriorating product market environ- ment. That this choice has been feasible may partly be attributable to the continuing access to nonmonetary compensation or social benefits. These issues are explored in more detail in chapter 3. References Commander, Simon, and John McHale. 1995. "Worker Influence and Em- ployment Bias in a Transitional Firm." EDI, World Bank, Washing- ton, D.C. Photocopy. Commander, Simon, and Ruslan Yemtsov. 1995. "Characteristics of Rus- sian Unemployment." In World Bank, Ruissia: Poverty Assessment. Washington, D.C. Commander, Simon, L. Liberman, C. Ugaz, and R. Yemtsov. 1993. "The Behavior of Russian Firms in 1992: Evidence from a Survey." World Bank Policy Research Working Paper 1166, Washington, D.C. Commander, Simon, John McHale, and Ruslan Yemtsov. 1995. "Russia." In Simon Commander and Fabrizio Coricelli, eds., Unemployment, Restructutring, and the Labor Market in Eastern Europe and Ruissia. EDI Development Studies. Washington, D.C.: World Bank. Foley, Mark. 1995. "Labor Market Flows in Russia: Evidence from the Russian Longitudinal Monitoring Survey." Europe and Central Asia Department, World Bank, Washington, D.C. Photocopy. 50 Employnienit, Wages, anid the Provision of Social Benefits Koen, Vincent. 1994. "Measuring the Transition: A User's View on Na- tional Accounts in Russia." IMF Working Paper 6, Washington, D.C. Layard, Richard, and Andrea Richter. 1994. "Russian Unemployment." Paper presented at IIASA Conference, Laxenburg, June. McDonald, I., and R. M. Solow. 1981. "Wage Bargaining and Employ- ment." Americani Econiomic Review 71:896-908. Schleifer, Andrei, and Dimitri Vasiliev. 1994. "Management Ownership and the Russian Privatization." Department of Economics, Har- vard University, Cambridge, Mass. Photocopy. Svejnar, Jan. 1986. "Bargaining Power, Fear of Disagreement, and Wage Settlements: Theory and Evidence from U.S. Industry." Economet- rica 54 (5):1055-78. Notes 1. It can also be argued that benefits have tended to attract discrete subsidiza- tion, including those achieved through tax breaks. 2. Indeed, nearly two-fifths of firms attached no importance to labor reduc- tions, and only 20 percent were seriously considering plant or shop closures. 3. No major hirings were reported in 72-89 percent of firms; the lower end was in services. For the bulk of firms-intermediate and consumer goods produc- ers-no major hirings occurred in over 85 percent of cases. 4. These likely understate wage differentiation given nondisclosure about fringe and other payments at the upper end of the distribution. 5. This is in accord with earlier findings based on smaller and less repre- sentative surveys reported in Commander, McHale, and Yemtsov 1995. 6. New trade unions have emerged, particularly in the medium-size and larger firms, and their membership accounted for around 20 percent of current employment. 7. This absent relationship can be replicated in a regression relating layoffs to share allocations. 8. When the Board of Directors is reported to have only four members-the number legally required for participation in the privatization-we can infer that the first shareholders' meeting has not yet occurred and we should not expect major changes. 9. We computed chi-square statistics relating types of shareholders with influ- ence on management. For outside institutional investors, chi-square (87) = 77.6; for workers, chi-square (189) = 185.2; and for managers, chi-square (135) = 89.3. How Russian Firms Make Their Wage and Employment Decisions 51 10. Note that approximately 80 percent of firms classed themselves as nor- mally profitable. Of these firms, increasing productive investment and working capital was cited as the best use of profits in 74 and 65 percent of cases, respec- tively; paying bonuses to workers or managers was cited in 19 and 3 percent, re- spectively. 11. The manager may not, of course, act in the interest of the owner. Incleed, lack of clarity on outside control may facilitate the manager's appropriation of profit, even if this appropriation is unlikely to be indefinite. As such, short-run profit maximization by the manager need not imply an objective of maximizing the value of the firm. 12. Given the discrepancy in some of the datapoints, we use annualized growth rates where appropriate. 13. The gross surplus per worker measure is given by: it = (R - WoL - hI)/L, where R = sales; W0, = wage evaluated at the reservation price or actual wages; L = employment; and H = other costs. This obviously ignores taxes, and hence overestimates the effective "pie" to be distributed. 14. We thank Mark Schaffer for a great deal of help in selecting the apprcpri- ate variables in the dataset for this exercise. 15. A mean wage for the fifteen branches in each of the three geographical re- gions was calculated (WI,a) for each year. For individuals with at least one year of service, the unemployment benefit would comprise 75 percent of their average wage over the previous twelve months for months 1-3 of any unemployment spell, 60 percent for months 4-7; and 45 percent for months 8-12. We assume that the one year of service criterion has been met. By construction, the alternative wage, AW = [probability of being employed * Wf,,] + [probability of being unem- ployed * {(3/12)*(75/100) + (4/12)*(60/100) + (5/12)*(45/100)} * WbJ = [prob- ability of being employed + (probability of being unemployed * 0.5748)]* Wbe. The following transition probabilities, assuming that state and privatized probabilities were the same, were used: Employed Unemployed State 0.915 0.085 Private 0.844 0.156 We thank Mark Foley for providing these transition probabilities from the Rus- sian Longitudinal Monitoring Survey. 3 Social Benefits and the Russian Industrial Firm Simon Commander, Une J. Lee, and Andrei Tolstopiatenko Russian firms have a history of providing a significant range of non- monetary benefits to incumbent workers. This component of total com- pensation was not only quite large, but was also intended to be an important factor in individual agents' employment decisions. A com- pressed wage structure limited the power of wage differentials to guide such decisions. At the same time, an explicit objective of raising the ratio of the nonmonetary component in total compensation was to sponsor at- tachment, primarily through the provision of firm-managed housing. This traditional preference for a low absolute monetary wage component has extended into the transition period. It is likely to have had significant incentive effects, as well as larger implications on the demand side, par- ticularly given that downward adjustments to money wages appear to have been larger than those to nonmonetary compensation in the early part of the transition. In addition, because of concentration in employment, firms were broadly equivalent to local government in functions and tax assignments We thank Lev Freinkman, Mark E. Schaffer, Mark Sundberg, Dusan Vujovic, and Dennis Whittle for very helpful suggestions on earlier drafts of this chapter. 52 Social Benefits and the Rtissian Industrial Firm 53 in a significant number of settings. To a greater extent than in Eastern, and Central Europe, in Russia, firms were the providers of a wide range of services that would usually be provided by municipalities or other branches of government in market economies. Expenditure on social benefits and services by Russian firms amounted to about 4.1 percent of GDP in 1992 and 3.3 percent in 1993 (2.1 percent for industrial firms), and firms are estimated to have contributed at least one-quarter of total ex- penditure on housing, health, education, and cultural services in 1993.1 Spending on social benefits and services was also equivalent to about 14 percent of the enterprises' total wage bill in 1993 for all firms; for indus- trial firms, it accounted for about 20 percent. By comparison, in Poland expenditure on social and housing funds as a percentage of the wage' bill (net wage cost) was no more than 10 percent for industry in 1989 (see Schaffer 1995). It is clear that Russian industry entered the transition with a higher exposure in benefits. Providing social benefits clearly imposes nontrivial costs on firms, particularly when cost recovery remains constrained by low income lev- els and explicit price controls. While explicit pricing caps constrain cost recovery on housing and utilities to no more than 20 percent of cost, esti- mates indicate cost recovery to be significantly less. Although this may have given firms little incentive for greater cost recovery, such costs have also been offset by a combination of tax advantages and compensatory subsidies. Without changing the structure of compensation, the ability of firms to charge for services, and hence shift costs toward market values, will be constrained by the low ratio of cash wages to nonmonetary con- stituents of compensation. This current structure of compensation has negative implications for enterprise restructuring and the growth of the private sector in Russia. A series of policy measures designed to cut the Gordian knot tying firms to benefits have already been implemented, including recent de- crees that have forced firms to divest themselves of their social assets. The arguments for divestiture are several. First, it has been widely argued that the use of benefits to achieve worker attachment now constitutes a second-order problem, given the need for restructuring and associated employment contraction. Tying workers to firms through benefits ntay consequently impede labor reallocation and mobility. Second, freed through institutional change-primarily privatization-from an exoge- nous requirement to provide benefits, firms are likely to limit the provi- 54 Employmiientt, Wages. and the Provision of Social Benefits sion of benefits that are costly or riskier for them. This may, however, have negative welfare implications. Third, firms should concentrate on raising productive efficiency and profits, liberated from a need to provide benefits. If benefits, for example, are a net burden to some firms, this may distort competition. Fourth, multiple provision of benefits through firms in a locality may be inefficient, and these losses could in principle be addressed through consolidation. Fifth, liberalizing wage setting should allow employers to use monetary compensation as the main sorting and motivational mechanism; this would likely be impeded by continuing provision of blanket access to benefits. Sixth, benefits provided by firms could either be devolved to individuals-through privatization of hous- ing, for example-or, when that option is unavailable, through transfer to local governments. While the economic underpinnings of the arguments above are intelli- gible, each has its own problem in the Russian context. First, fiscal rela- tions among federal, oblast, and municipal tiers of the fiscal system have yet to be stabilized. For the bulk of social benefits currently delivered by firms, the issue is less that of possible efficiency gains from decentraliza- tion-regional or spatial differences in customary benefits tend not to be large-than the respective assignment of tax bases and spending. Recent evidence suggests that local administrations have tried to raise their share of existing tax bases relative to the federal government and innovate in local tax collection, often in a highly distortionary way. Aside from the multiple inefficiencies this may induce (a helpful survey of these issues can be found in Tanzi 1995), the central issue is the impact on possible net redistributions within the fiscal system when there is a large asymmetry, the result of relatively revenue-rich regions retaining higher local tax shares, while deficitary regions must continue to rely on federal support in a context of declining aggregate federal tax yields (see Wallich 1994). This problem will be expanded in contexts where a firm or small number of firms effectively comprise the local government tax and institutional base. In these cases, divesting firms of responsibility for social assets and transferring those responsibilities to municipalities does nothing to an- swer the financing question. Second, the current wage-tax regulations, which impose an excess wage tax on pay settlements that exceed six times the minimum wage,2 are designed in a way that could encourage substitution of nonmonetary benefits for cash wages, given that the excess wage tax remains condi- Social Benefits aind the Ruissian Indutstrial Firm 55 tioned on monetary payments. Payroll taxes and deductions comprise up to 40 percent of the wage bill, and there thus appears to be an incentive for firms to increase in-kind compensation. Third, institutional limitations relating to property rights and agency continue to limit the ability to de- volve services away from firms. Fourth, the original objective of attaching labor to firms through benefits, particularly housing, may already be a small impediment to mobility. An earlier survey of firms in the MoScow region found that roughly 40 percent of tenants at end-1992 actually worked outside the firm controlling the housing stock, and these num- bers are replicated in a World Bank survey of twenty-two firms inmple- mented in late 1994.3 That there is little spatial mobility is beyond doubt; that it is attributable to controls on housing is less obvious. This chapter cannot address all of these issues. Rather, we have a more modest objective: to provide an overview of the scale of benefits provided by Russian industrial firms. We look at the scale of provision at the start of transition and the subsequent changes in the provision of benefits through mid-1994. We are also able to pick up the widespread change in title that has occurred since the start of transition. This allows us tc get some idea of the impact, if any, of ownership change on benefits provi- sion. We also begin to look at the way in which shocks to firms have been absorbed. While the main part of this story is in the wage-employment choices that have been made, an important associated question is the ex- tent to which changes to wages and benefits have moved together, or whether benefits and cash wages have been in some measure substitutes. What Benefits Are Provided? Table 3-1 provides a simple listing of the benefits provided by firms be- fore transition and in mid-1994. The numbers are for a balanced panel, with responses in both periods. Several points emerge. First, the great majority of firms provided and continue to provide a wide range of bene- fits. Indeed, only 5 percent of respondents offered no benefits before tran- sition, and this share had shifted only marginally upward, to 7 percent, by mid-1994. Close to 60 percent of firms provided four or more benefits at the end of the period; it is striking that health and childcare facilil.ies continue to be provided by over two-thirds of industrial firms. There has been a slight increase in the number of firms providing no benefits, and a similarly slight decrease in the numbers providing most of the core belle- 56 Employment, Wages, and thte Provision of Social Beinefits Table 3-1. Provision of Benefits, Mid-1994 and 1990/91 (percent) Enterprises Item Mid-1994 1990/91 Benefit Childcare/childcare subsidy 66 79 Healthcare facility 70 71 Food subsidy/cafeteria 78 83 Food and/or consumer goods sold 60 52 Construction of new housing 50 73 Housing/housing subsidy 55 59 Holiday resort/holiday subsidy 45 57 Transportation/transportation subsidy 57 57 Other 21 17 Number of core benefits 0 7 5 1 5 4 2 7 4 3 9 9 4 14 12 5 13 13 6 16 17 7 15 19 8 14 18 >3 72 79 > 4 58 67 Enterprises responding (number) 407 407 Note: Includes only enterprises that responded in both periods. Core benefits exclude "other" benefits. fits. In short, although table 3-1 clearly indicates some decline in the num- ber of benefits provided and in the share of firms offering such benefits, the overall picture is one of significant inertia. The likely high-cost bene- fits-such as healthcare-have remained almost untouched. At this stage, however, we are unable to control for quality of benefits provision where anecdotal evidence suggests degradation. The one exception is in housing construction. Here, we already find a very significant contraction, so that by mid-1994 only half of the firms had a current construction program. Social Benefits and the Ruissian In7dutstrial Firm 57 Firms were also asked about prospective changes in benefits supply over the coming year. About 60 percent of respondents providing at least one of the core benefits indicated no expectation of further reductions. Projected reductions for the 40 percent of enterprises indicating some lessening in benefits were generally in the range of one to two benefits and were concentrated in housing construction and childcare/kindergar- ten facilities. About 33 percent of enterprises currently providing c:hild- care/kindergartens and 35 percent of enterprises providing housing construction expect to cut these benefits. Surprisingly, only about 14 per- cent and 10 percent of enterprises offering healthcare facilities and hous- ing/housing subsidy, respectively, expected to cut these over the next twelve months. This is in contrast to the World Bank survey of twenty- two firms in nine municipalities. Over 50 percent of these firms indicated that as a result of government policy, they were transferring their hous- ing stock to municipalities, with the remainder willing to transfer, but yet to commence negotiations with municipalities. Perhaps more significant, in the responses to a question regarding the burden of providing social benefits, over 50 percent of respondents cited social-cum-ethical reasons for continuing to provide benefits.4 This re- sponse dominated all other responses, with only about 25 percent oiF re- sponses citing attachment of workers as an important factor. While this response could be variously interpreted, it seems to point to an important quality of benevolence or extended social function for firms that is still shared by workers, managers, and local governments in Russia. It cau- tions against treating the Russian firm-state or privatized-as a stand- ard, profit-maximizing entity. It is significant that barely 20 percent of firms considered benefits to be a major financial burden, while less tham 5 percent considered benefits to be a major obstacle to firm restructuring.5 We return to the cost implications of benefits supply below. As to the asset structure associated with benefits, table 3-2 breaks down the kinds of benefits by ownership of assets. What is evident is that in the cases of housing, transportation, and cafeterias, firms were the owners in 80 percent of the cases. In some contrast, municipalities, firm workers, and/or other firms had ownership stakes in health and child- care facilities, as well as holiday homes and the like. There is little dif:Fer- ence among the ownership categories, with a few exceptions. Far fewer de novo firms owned the social assets they provided than other ownership categories, with the exception of transportation. Privatized firms aind 58 Employment, Wages, anid the Provision of Social Benefits Table 3-2. Ownership of Benefits, Mid-1994 (percent) Firms Muniiicipal Employees Othter No respondinig Benefit Firmi gozvernmeneiit offirm11 firms responzse (number) Childcare/childcare subsidy 68 23 2 5 2 279 Healthcare facility 66 26 2 5 1 296 Food subsidy/ cafeteria 87 3 3 5 2 328 Food/consumer goodssold 84 5 2 6 3 252 Construction of new housing 61 8 5 4 2 207 Housing/housing subsidy 87 5 2 4 2 227 Holiday resort/subsidy 69 10 3 12 6 196 Transportation/ trans- portation subsidy 86 9 1 1 2 248 Other 81 6 8 1 4 90 Note: Out of total number of enterprises providing particular benefit. state enterprises were nearly identical in their ownership patterns of the social assets. Municipal governments' exposure wvill certainly have in- creased since mid-1994 given recent government decrees mandating di- vestiture. From the survey we can also see that in cases where firms expect to cut childcare and healthcare facilities, the majority of these firms also indicated that these benefits would be transferred to municipalities. At the time of the survey, however, we find little active disposal or trans- fer of social assets, with only about 4 percent of enterprises that provide at least one of the core benefits indicating recent disposal of their social assets.6 Earlier results from smaller, biased samples indicated that the scale of benefits provision wvas tightly correlated with firm size. This is amply confirmed in table 3-3, where it can be seen that larger firms-particu- larly those with more than 10,000 employees-have far higher exposure to benefits. Indeed, the group of largest firms provided seven or more kinds of benefits. As with other surveys, large firms invariably provided Social Benefits and the Russian Indutstrial Firm 59 Table 3-3. Provision of Benefits and Size of Enterprise, Mid-1994 (percent) L >100, L > 500, L > 1,500, L > Benefit L <100 L <500 L <1,500 L < 10,000 10,000 Childcare/childcare subsidy 31 52 67 90 100 Healthcare facility 35 63 75 86 100 Food subsidy/cafeteria 31 66 90 98 100 Food/consumer goods sold 29 46 68 81 80 Construction of new housing 18 35 53 77 100 Housing/housing subsidy 12 36 62 86 100 Holiday resort/subsidy 28 36 40 62 90 Transportation/ transporta- tion subsidy 46 55 64 60 90 Other 15 22 23 18 40 Number of core benefits 0 38 5 2 0 0 1 7 12 0 0 0 2 12 11 6 2 0 3 12 18 10 2 0 4 10 14 20 10 0 5 9 16 15 11 0 6 6 11 15 20 0 7 4 6 22 27 40 8 1 6 10 29 60 > 3 30 53 82 97 1(0 > 4 20 39 62 87 1(0 Enterprises responding (number) 68 126 81 94 10 Note: Core benefits exclude "other" benefits. housing, health, and childcare facilities. In contrast, over 35 percent of the smallest firms provided no benefits; where benefits were given, the range was far smaller, with housing provided in under 20 percent of cases. Looking a little more closely at the sectoral distribution (see table 3-4), we find a fairly tight link to employment size. Branches where mean firm size is large, such as metallurgy, machine-building, and fuel and energy, have high benefits exposure. The only outlier is automobiles-despite large employment size, benefits levels were close to the mean. Table 3-4. Provision of Benefits by Main Industrial Sectors, Mid-1994 (percent) Machine Non- Chemicall tool Agricili- Food Ferrouts ferrouis petro- Heauy enginieer- teral Other Wood' Construc- Light pro- Benefit Energy Fuel metal metal chemical machinery lug Atito. machinery Defense MB paper tion material indtstry cessing Child care 83 79 75 100 73 72 50 55 93 75 62 53 58 48 47 Healthcare facility 83 86 92 100 77 72 61 82 57 82 64 53 58 59 50 Food subsidy/cafeteria 75 93 75 91 88 90 78 73 100 84 74 66 58 61 65 Food/consumer goods sold 67 64 83 91 54 59 50 73 93 61 53 66 29 45 62 Construction of new housing 67 79 75 91 58 49 33 55 64 54 40 47 45 25 41 Housing/housing subsidy 83 79 50 73 62 64 50 64 64 60 32 60 42 46 29 Holiday subsidy/resort 75 57 67 73 62 51 50 45 50 54 43 28 35 21 32 Transportation/transporta- tion subsidy 75 79 75 45 54 67 56 64 50 58 53 72 52 39 53 Other 25 29 42 0 42 13 28 18 36 15 17 31 16 14 15 Enterprises responding (number) 12 14 12 11 26 39 18 11 14 67 47 32 31 56 34 Benefits (number) 0 0 0 0 0 4 3 6 9 0 4 13 16 13 18 9 1 8 0 8 0 0 3 6 9 0 4 6 6 10 5 6 2 8 7 17 9 12 3 17 0 0 3 4 6 6 13 18 3 0 0 0 0 8 5 11 18 14 6 17 3 16 14 12 4 8 21 0 0 15 28 11 0 14 15 4 6 19 16 18 5 0 0 0 0 8 10 11 0 14 13 21 22 13 11 15 6 17 21 8 27 15 15 22 18 21 22 14 13 3 14 12 7 17 14 33 27 19 23 11 27 14 13 11 19 10 4 6 8 42 36 33 36 19 10 6 18 21 18 9 9 10 5 6 > 3 84 92 74 90 76 86 61 63 84 81 59 69 55 50 57 > 4 76 71 74 90 61 58 50 63 70 66 55 63 36 34 39 Note: Number of benefits excludes "other"; defense sector includes shipbuilding and airplane manufacturing. Social Benefits and the Ruissiani Industrial Finri 61 Table 3-5. Provision of Benefits and Ownership, Mid-1994 (percent) Privatized De novo Benefit SOEs enterprises enterprises Childcare/childcare subsidy 68 69 23 Healthcare facility 75 72 27 Food subsidy/cafeteria 79 81 25 Food/consumer goods sold 66 62 14 Construction of new housing 53 51 11 Housing/housing subsidy 56 58 9 Holiday resort/subsidy 48 45 36 Transportation/transportation subsidy 62 57 48 Other 23 20 18 Number of core benefits 0 5 5 36 1 3 6 9 2 6 7 16 3 7 10 16 4 16 13 14 5 15 11 7 6 23 14 0 7 10 18 2 8 16 16 0 ,3 80 72 23 >4 64 59 9 Enterprises responding (number) 10 92 71 Note: Core benefits exclude "other" benefits. SOE, state-owned enterprise. In the relationship between firm ownership and firm size, there is lit- tle to distinguish state firms from privatized entities (see table 3-5). De novo firms tended to offer a relatively small range of benefits. This is par- tially explained by ownership and size correlation-de novo firms are small in size (we address the question of whether, after controlling for size, de novo firms still offer fewer benefits later in the chapter). What is striking is that de novo firms do generally provide some benefits. Indeed, not only did a majority of de novo firms provide some benefits, but 23 per- cent offered more than four. This is consistent with the evidence from Po- land, and it may indicate that private firms need to offer benefits in order to compete for workers in labor markets that until recently were charac- Table 3-6. Provision of Benefits by Region, Mid-1994 (percent) Number of core benefits Number of Region 0 1 2 3 4 5 6 7 8 > 3 > 4 enterprises Altayskiy 11 33 11 0 0 0 33 11 0 44 44 9 Krasnoyarskiy 16 0 5 5 16 5 26 11 16 74 58 19 Primorskiy 13 0 0 13 7 13 20 13 20 73 66 15 Stavropolskiy 17 13 4 13 0 4 17 13 17 51 51 23 Arkhangelskaya 0 0 25 13 13 25 0 0 25 63 50 8 Vladimirskaya 6 0 6 22 22 11 17 11 6 67 45 18 Vologodskaya 0 0 0 0 14 43 0 29 14 100 86 7 Voronezhskaya 38 8 0 8 15 15 0 8 8 46 31 13 Nizhegorodskaya 4 4 15 8 19 4 8 19 19 69 50 26 Kemerovskaya 29 14 0 0 0 21 7 14 14 56 56 14 Samarskaya 6 6 0 18 18 6 6 18 24 72 54 17 St. Petersburg 10 3 6 13 32 13 10 3 10 68 36 31 Leningradskaya 0 9 9 9 27 9 18 18 0 72 45 11 Moscow 5 9 9 18 16 16 15 9 2 58 42 55 Moscovskaya 0 0 0 18 9 18 18 36 0 88 72 11 Novosibirskaya 5 9 9 5 9 5 27 18 14 73 64 22 Permskaya 8 0 16 0 20 12 20 12 12 76 56 25 Rostovskaya 6 6 6 0 6 12 18 24 24 84 78 17 Saratovskaya 6 0 22 17 11 0 22 11 11 55 44 18 Sverdlovskaya 0 0 10 0 19 14 5 24 29 91 72 21 Smolenskaya 11 11 11 11 0 22 11 22 0 55 55 9 Tyumenskaya 0 0 0 14 0 43 0 14 29 86 86 7 Bashkorstan 5 0 10 14 10 14 24 10 14 72 62 21 Tatarstan 5 0 0 0 5 11 26 16 37 95 90 19 Note: This excludes "other" benefits, Social Benefits and the Rutssian Indutstrial Firm: 63 terized by large gross flows and turnover (for Poland, see Estrin, Schaffer, and Singh 1994; on gross flows in Russia, see Commander, McHale, and Yemtsov 1995). We return to the benefit-to-employment flows relation- ship below. In sum, benefits tend to be heavily concentrated in larger firms. This is true across branches and ownership forms. Given the relatively weak dif- ferentiation in relative wages across firm-size classes, one further implica- tion would be that the nonmonetary share in total worker compensation would be significantly higher in larger firms. It is interesting to note, however, that smaller de novo firms do offer benefits to workers. This may partly be a function of the excess wage tax rule, but it is also probably re- lated to the widespread habit of Russian workers-and hence the expec- tation-of receiving nonmonetary compensation. The attraction of such a compensation scheme would likely be enhanced in a high-inflation envi- ronment where formal indexation schemes have been absent. The survey also allows us to look in more detail at the regional aspect, an element that has been little explored. This is potentially important be- cause we know that given regions have higher employment concentra- tion ratios and degrees of diversity in production. Table 3-6 breaks down the supply of benefits by region. The first point that stands out is that there is significant variation among regions. Refining this to incorporate the question of employment concentration in firms and/or industries, we also differentiate firms and their benefits supply by their setting. These categories include firms in major urban centers, effectively those located in Moscow and St. Petersburg; firms located in a dominant or single-en- terprise setting, as well as in a dominant industry context; and other firms.7 We are also able to classify by oblast centers, other smaller urban areas, and rural contexts. Table 3-7 lists some simple statistics on the share of firms offerirtg given benefits across the different settings. As we might expect, concentration in employment is unambiguously associated with higher shares of benefit provision across all categories. Housing, healthcare, kindergartens, cafeterias, and transportation all have notably larger firm provision in these areas than elsewhere. Inversely, the range of benefits tends to be far smaller in Moscow and St. Petersburg, and this is particularly striking in housing. Firms located in rural areas also offer a low range of benefits, including housing. We now explore the characteristics of firms providing benefits a bit more systematically. We ran two sets of ordered logit estimations looking at both the level of benefits (as measured by the number of benefits sup- 64 Employmentt, Wages, and the Provision of Social Benefits Table 3-7. Provision of Benefits and Firm Setting, Mid-1994 (percent) Domiiitnant Other Urban industry Oblast Othler Rutral Beniefit area towntI Other center towlns area Childcare/childcare subsidy 48 77 66 65 70 52 Healthcare facility 67 77 66 63 73 43 Food subsidy/cafeteria 81 90 71 70 75 57 Food/consumer goods sold 55 75 54 51 56 57 Construction of new housing 28 69 50 49 49 57 Housing/housing subsidy 35 75 53 52 56 43 Holiday resort/subsidy 42 54 39 41 38 33 Transportation / transporta- tion subsidy 42 77 57 55 62 67 Other 19 27 28 30 22 48 Number of core benefits 0 6 0 10 13 7 10 1 6 4 5 4 5 5 2 9 2 8 9 5 24 3 16 8 8 7 9 14 4 22 6 12 12 14 5 5 15 10 12 12 12 10 6 13 23 15 14 19 0 7 7 15 16 15 19 10 8 5 32 14 15 20 24 > 3 62 86 69 68 84 49 > 4 40 80 57 56 70 44 Enterprises responding (number) 88 48 300 151 128 21 Note: Core benefits exclude "other" benefits. plied to workers at the level of the individual firm) and the factors associ- ated with changes in the supply of benefits in the interval from the start of transition through to mid-1994 (a similar exercise is done on Polish data in Estrin, Schaffer, and Singh 1994). For the level of benefits, we relate the number of benefits to size, own- ership, measures of labor influence or power, firm setting, and wages. The results of the regression are reported in table 3-8. In determining the level of benefits, firm size proves to be significant and positive, even con- Social Benefits and the Russian Induistrial Firm 65 Table 3-8. Number of Social Benefits, Ordered Logit Estimation Standard Ben94 Coefficient error t-statistic P> I t 1 [95% Confidence-IIterval] PE 0.0109522 0.2293347 0.048 0.962 -0.4385355 0.4604399 NPE -0.7558396 0.4050822 -1.866 0.062 -1.5497860 0.0381068 lpower 0.4313595 0.2332900 1.849 0.064 -0.0258805 0.8885995 LenIp94 0.8010569 0.0807954 9.915 0.000 0.6427008 0.9,594131 nzew_u 0.2301654 0.2646661 0.870 0.384 -0.2885705 0.74189014 AW 0.5952661 0.1569884 3.792 0.000 0.2875744 0.9029577 profit 0.4337249 0.2921020 1.485 0.138 -0.1387845 1.0062340 donw_iid 0.8425167 0.3106043 2.713 0.007 0.2337434 1.4512900 xwvage -0.0826205 0.2030657 -0.407 0.684 -0.4806220 0.31]53809 cutl 6.125366 1.105278 Ancillary parameters cut2 6.973138 1.116834 cut3 7.691242 1.128282 cut4 8.473939 1.143115 cut5 9.305381 1.160794 cut6 10.107320 1.175523 cut7 10.953360 1.192265 cut8 12.074020 1.214563 2 ~~~~~~2 Note: Log likelihood -627.63754; n 339; chi (9) 207.35; prob. > chi = 0.000; pseudo R2 = 0.1418; ben94 = number of core benefits in 1994; PE = dummy variable = 1 if privatized enterprise, 0 otherwise; NPE = dummy variable = 1 if de niooO private enterprise, 0 otherwise; AW = natural log of average monthly wage for June 1994; profit = dummy variable = I if en- terprise is usually a profit-maker, 0 otherwise; dom_ind = dummy variable = 1 if enterprise located in dominant/one-company or industry town, 0 otherwise; lpower = dummy variable = 1 if employment or welfare of workers is considered to be an important management ob- jective in 1994, 0 otherwise; Lenp94 = natural log of 1994 employment; nezv_u = dummy variable =1 if new union coverage, 0 otherwise; xwvage = dummy variable = 1 if the excess wage tax is important in determining the wage, 0 otherwise. trolling for ownership type. We do not find statistically significant differ- ences between state-owned firms (SOEs) or privatized (PEs) entities in the number of benefits provided. The coefficient on new private firms (NPE) is negative, and is just significant (at the 10 percent level), once we control for firm size. Concentration in employment is positive and statis- tically significant. Firm profitability exerts no significant effect on the number of benefits supplied. We also include two measures of labor power or influence-new un- ion coverage and the importance attached to either employment or worker welfare (wages and benefits) within a given firm. New union cov- 66 Emnploymiienit, Wages, and the Provisionl of Social Beniefits erage proved insignificant, but the second labor power variable was posi- tive and significant at the 10 percent level. We also examined the relation- ship between wages and social benefits. It appears that the wage level and the level of social benefits are positively and significantly associated. This suggests that benefits are not generally treated as a substitute for monetary compensation. Further, there seems to be no correlation be- tween the excess wage tax and the benefits level. In looking at the determinants of change in the level of benefits over the two periods, we included, on the right-hand side, growth in sales and wage growth. The results are given in table 3-9. The coefficients on the Table 3-9. Change in Benefit Levels, Ordered Logit Estimation Stanidard c ben Coefficient error t-statistic P > It I [95% Confide77ce-Intervall PE 0.2942403 0.3368804 0.873 0.382 -0.3660330 0.9545137 NPE 2.1355891 0.7224590 1.240 0.215 -1.2403680 5.5115470 lpower 0.7852497 0.3962209 1.982 0.047 -0.0086709 1.5618280 L_emp94 0.1898186 0.1018642 1.863 0.062 0.0098315 0.3894687 new_ai 0.6292750 0.3638603 1.729 0.084 -0.0838781 1.3424280 csa9094 -0.1602459 0.0791435 -2.025 0.043 -0.3153644 -0.0051274 c.wg9094 -0.2112011 0.2254086 -0.937 0.349 -0.6529938 0.2305916 profit -0.0165803 0.5037573 -0.033 0.974 -1.0039270 0.9707659 doni_ind -0.4213361 0.4199568 -1.003 0.316 -1.2444360 0.4017641 xwvage -0.1805614 0.2987071 -0.604 0.549 -0.7660166 0.4048938 cutI 1.193075 1.010862 Ancillary parameters cut2 3.238544 1.042907 cut3 4.839693 1.092352 2 2 Note: Log likelihood = -199.87718; it = 179; chi (9) = 18.10; prob. > chi = 0.0532; pseudo R = 0.0433; c bem = change in the level of core benefits between 1990 and mid-1994, where 1 = increase, 2 = no change, 3 = small decrease, and 4 = large decrease; PE = dummy vari- able = 1 if privatized enterprise, 0 otherwise; NPE = dummy variable = 1 if de novo private enterprise, 0 otherwise; [pozver = dummy variable = I if employment or welfare of workers is considered to be an important management objective in 1994 and/or prereform, 0 other- wise; L.emp94 = natural log of 1994 employment; newvji = dummy variable = 1 if new union coverage, 0 otherwise; chsa9O94 = real growth in sales, from 1990 to mid-1994, in log form; clhwg9094 = real growth in wages, from 1990 to mid-1994, in log form; profit = dummy vari- able = 1 if enterprise is usually a profit maker, 0 otherwise; domt_id = dummy variable = 1 if enterprise located in dominant/one-company or industry town, 0 otherwise; xw'age = dummy variable = 1 if the excess wage tax is important in determining the wage, 0 other- wise. Social Benefits and the Rnissian Inidntstrial Firm 67 ownership variables are positive, suggesting declines are likely to be greatest in privatized and de novo firms. Neither coefficient, however, is statistically significant. The coefficient on size is positive and just signifi- cant, implying that declines are expected to be greater in larger firnms. The coefficients on both wage variables-wage growth and the importance of the excess wage tax in setting wages in the firm-are insignificant. The two measures of labor influence or power, measured by the weight given to worker welfare and/or employment, as well as new union coverage, are both (just) significant and positive. This suggests that labor power as measured by these two variables, instead of hindering loss of benefits, en- courages greater declines. Employment concentration is not significant. The firm's financial situation-as measured by profitability and sales growth-gives mixed results. The coefficient on firm profitability is nega- tive but insignificant. The coefficient on sales growth, however, is nega- tive and statistically significant. The latter is likely to be a better measure of the firm's financial position. In this section we have surveyed the kinds of benefits provided by firms in Russian industry, both by cross-tabulation and by two sets of simple regressions. The findings are fairly consistent with our priors and have many similarities with the Polish results (see Estrin, Schaffer, and Singh 1994). In general, Russian firms offer more benefits than Plolish firms, but the percentage of firms offering given core benefits is not much greater in Russia. On the number of social benefits, firm size, as measured by employment, followed by average wage are the two most important explanatory variables in both Russia and Poland. Profitability and the ex- cess wage tax are not important in either country, but labor power is just significant in both countries. Concentration in employment, a proxy for a company- or industry-town setting, and unique to the Russian dataset, is also strongly correlated with benefits supply. De novo private firms offer lower levels of benefits, even after controlling for size and other factors, in both Poland and Russia. There does not appear to be a robust d:iffer- ence across other ownership forms. On the change in the level of benefits, we observe that in both Russia and Poland, financial health of the firm is associated with smaller de- clines in benefits-sales is the indicator in Russia, and profits in Poland. De novo firms are associated with increasing the benefits offered in Po- land, even after controlling for size and other factors. In Russia this :rela- tionship does not hold. On the contrary, the coefficient is positive (but not 68 Employment, Wages, and the Provision of Social Bentefits significant), suggesting larger declines in de novo firms than in other en- terprises. We also see a number of other differences between the Russian and Polish results. Size is associated with increases in benefit levels in Russia, but with declines in benefit levels in Poland. Labor power is asso- ciated with smaller decreases in benefit levels in Russia, but not in Po- land. The excess wage tax is associated with smaller declines in benefits in Poland, but not in Russia. Finally, we should highlight that in Russia, despite some reduction in benefits offered by firms over the period 1990/91-1994, most firms have contracted supply in relatively small magnitudes. This suggests that the benefits component of worker com- pensation may have been considerably less flexible than money wages. Costs of Social Benefits As table 3-10 indicates, the gross cost of benefits at mid-1994 averaged about 18 percent of the wage bill, with relatively little variation across the ownership forms. The share does rise, although not monotonically, with firm size and with degree of concentration in employment. There is also fair dispersion across branches. This share is not very different from the 20-25 percent range that emerges from the smaller World Bank survey of firms for 1993 and 1994Q1-3. Indeed, given that the average size of the firms in this latter sample was over 5,000 employees in the survey peri- ods, these shares are very close to those reported for the two largest firm- size groups in the larger survey. These figures are also consistent with aggregate data (Roskomstat data). In 1994 (March), expenditure on social benefits and services by Russian industrial enterprises was about 21 per- cent of the wage bill. Aggregate numbers show some evidence of an in- crease in social benefits expenditure relative to the wage; the change, however, is not that large. In 1991/92, expenditure on social benefits for all enterprises in the national economy averaged about 8 percent of the wage bill, increasing to about 10 percent by mid-1994. Costs data are misleading because firms will tend to underestimate the true costs of benefit provision. This occurs because the prices used to evaluate benefits are based on operational costs, and hence ignore the im- plicit subsidies obtained by using facilities owned by the firm. An obvi- ous example would be inappropriately low attribution of rental costs. But arriving at a better valuation is difficult. One crude way of better estimat- ing the market value of benefits would be to contrast the cost figures for Social Benefits and the Ruissian Induistrial Firm 69 Table 3-10. Cost of Benefits, Mid-1994 (percentage of the wagefund) Category Percenitage Category Percentage Sample average 18 Ownership class Branch SOE 14 Energy 10 PE 21 Fuels 21 NPE 13 Ferrous metallurgy 21 Nonferrous metallurgy 28 Size of enterprise Chemicals 20 L < 100 9 Heavy machinery 17 L > 100, L < 500 14 Machine tools ]5 L > 500, L 1,500 17 Automobiles 1.9 L > 1,500, L < 10,000 25 Tractors/agr. machinery 28 L> 10,000 17 Defense 23 Other machine-building 19 Location Wood and paper 25 Major urban center 17 Construction materials 25 Dominant industry town 21 Light industry 16 Other 17 Food processing 16 Oblast center 16 Other industry 25 Other town 23 Rural 22 Note: These are gross costs. Means are weighted by the wage bill. SOE, state-owned enter- prise; PE, privatized entities; NPE, new private enterprise; L, employment. Defense sector includes shipbuilding and airplane manufacturing. de novo private firms with those for state and privatized firms, in the be- lief that the costs for the de novo private firm will likely lie closer to cur- rent market values. Table 3-11 displays the results of that exercise, and it shows quite strikingly that the average cost of benefits per employee for a comparable basket of benefits rises strongly with the degree to which the firm is private. The average cost of benefits was about 25 percent in re- maining state firms and about 47 percent in privatized firms, relative to de novo firms. Aside from pointing to very significant undervaluation, this suggests at the least that privatization has been accompanied by a shift toward using market values for benefits. Table 3-11 shows that imputing the cost numbers reported for de nova firms to other categories shifts the level of costs quite radically. As such, the gross cost of benefits at mid- 70 Enmployment, Wages, and the Provision of Social Benefits Table 3-11. Adjusted Cost of Benefits, Mid-1994 Average Adjulsted Adjutsted monthly Average Average Cost of average cost of cost of nmonthly wvage beniefits wage benefits Observa- beniefits wvage (% of (% of M% of (% of ,ionis Firna (rubles) (rtbles) TLC) TLC) TLC) TLC) (nuitmber) SOE 16,593 185,906 92 8 74 26 24 PE 31,682 167,358 86 14 71 29 67 NPE 66,795 270,041 85 15 80 20 8 Note: Average cost of benefits and wages are per employee. TLC is total labor costs and includes wages and cost of benefits. See table 3-10 note for definitions. 1994 may have exceeded 30 percent of the wage bill. In addition, using these values shifts the structure of compensation, an issue we return to below. Cost Recovery Given the size of shocks to Russian industrial firms-by mid-1994 firms in the survey were operating at an average of about 50 percent of capacity levels in 1990/91-and the associated deterioration in profitability, we can imagine that firms have been increasingly interested in cost savings. Indeed, we know that cost recovery levels have been historically very low, particularly for housing. What is striking, howATever, is that whether under firm or municipal control, the level of cost recovery for housing has remained very small. While explicit pricing caps constrain cost recov- ery on housing and utilities to no more than 20 percent of cost, survey evidence from the smaller survey of nine cities found a range of no more than 4-16 percent cost recovery at the municipal level in early 1995.8 For firms in the process of divesting their housing, average cost recovery on housing in 1993 and 1994, including maintenance and communal serv- ices, was below 20 percent, but the median was just under 10 percent. More globally, estimated rental and utility charges levied on tenants came to barely 18 percent of financial cost and 12 percent of economic cost at mid-1994.9 With further liberalization of energy prices, World Bank estimates suggest that the cost of operating the housing stock will Social Ben efits and the Russian f Indutstrial Firmfi 71 rise by approximately 50 percent. This would widen the gap without a greater charge-back to tenants.10 As we have seen, the firm response to a deterioration in profitability and low cost recovery can take a number of forms. First, firms have closed down specific operations, either outright or by using previous fa- cilities for commercial or other activities. Further, in 70-75 percent of cases where either childcare or healthcare facilities were cut, municipal governments appear to have taken over or substituted such benefits. Out- right loss of access to services-as indicated in table 3-1-was a less likely outcome in the majority of cases. Second, firms can resort to greater charge-backs for use of facilities, both to incumbent workers and to out- siders using those facilities. The small World Bank survey suggests, how- ever, that this has not been a favored option. Over half the firms surveyed indicated no desire to implement greater cost recovery. This unwilling- ness may partially be explained by explicit local government caps on pricing levels, and hence a weak incentive for cost recovery. It also plays back to the continuing low level of monetary compensation available to workers, and thus their ability to pay for services. While the costs for firms providing benefits are nontrivial, such costs have also been offset by a combination of tax advantages and compensa- tory subsidies. In the former case, prior to the mandated divestiture from mid-1994 onward, firms had been receiving explicit tax relief. For exam- ple, in most settings firms have not only been permitted direct tax credits on housing expenditures of up to 1.5 percent of turnover, but additional, often nontrivial, tax breaks on their profit tax liabilities. As a ccnse- quence, divestiture of housing to municipalities may not have a large di- rect budgetary effect. In addition to explicit offsetting support through tax relief, there is evidence that social benefits costs have also been offset by government transfers (as indicated in chapter 6 of this volume). We might expect the provision of selected benefits, such as housing/housing subsidy and kin- dergartens/childcare facilities, to be related to government financial transfers. We test this by running a regression relating total government transfers per worker to the provision of a subgroup of those benefits vvith merit characteristics and a likelihood of large cost implications-hous- ing/housing subsidy and childcare. We include ownership and branch dummnies and add the cost of social benefits as a percentage of the wage 72 Employmen2t, Wages, and thte Provisiont of Social Beniefits Table 3-12. Financial Transfers, OLS Regression Stanidard Ftron94 Coefficient error I-statistic P > I t 1 [95% Confidence interval] beniefits 0.5431010 0.2765268 1.964 0.055 -0.0123194 1.098521 PE 0.0071642 0.2452120 0.029 0.977 -0.4853587 0.499687 NPE 0.0330417 0.6805420 0.049 0.961 -1.3338670 1.399957 soc_wb -1.5238320 0.8390424 -1.816 0.075 -3.2090990 0.161433 branch2 2.2055760 1.0628610 2.075 0.043 0.0707568 4.340396 branch3 -0.2092506 1.3063920 -0.160 0.873 -2.8332160 2.414715 branch4 -0.0858868 1.1190440 -0.077 0.939 -2.3335520 2.161778 branch5 1.0904950 1.0833670 1.007 0.319 -1.0855110 3.266501 branch6 0.3060733 0.9905473 0.309 0.759 -1.6834990 2.295646 branch7 0.0410337 1.0402360 0.039 0.969 -2.0483410 2.130408 branch8 (dropped) branch9 0.1984083 1.0474870 0.189 0.851 -1.905530 2.302347 branchlO 0.3538287 0.9579377 0.369 0.713 -1.570246 2.277903 branclll2 0.3698282 0.9667561 0.383 0.704 -1.571958 2.311615 branchl2 -0.0598335 1.0626900 -0.056 0.955 -2.194309 2.074642 branch13 0.3123433 1.1309020 0.276 0.784 -1.959139 2.583826 branchl4 0.0739517 0.9997293 0.074 0.941 -1.934064 2.081967 branchl5 0.5222599 1.0231730 0.510 0.612 -1.532844 2.577364 brflnchl6 0.3232304 1.1332190 0.285 0.777 -1.952906 2.599367 constant -0.0631379 0.9505497 -0.066 0.947 -1.972373 1.846097 Note: n 69; F(18,51) = 1.15; Prob. > F = 0.3343; R = 0.2933; = 0.0389; root MSE = 0.90739; Ftran94 = financial transfers per worker, mid-1994; beniefits = dummy variable =1 if enterprise provided kindergartens/childcare facilities and housing/housing subsidy in 1994, 0 otherwise; PE = dummy variable = I if privatized enterprise, 0 otherwise; NPE = dummy variable = 1 if de novo private enterprise, 0 otherwise; soc_wb = cost of social benefits as a percentage of the wage bill; branch' = branch dummy variables. bill. The results of the regression are given in table 3-12. We find that the provision of these benefits is positively related to the level of government transfers. The coefficients on the ownership variables are not all signifi- cant. With the exception of the fuel sector (branch2), which is positively and significantly related to the level of transfers, the coefficients on the branch dummies are all insignificant. The cost of benefits variable is negative and significant at the 10 percent level. The compensation that results, however, appears to be both partial and concentrated among the larger firms, as categorized by employment (also see chapter 6 in this volume). Further, while around 27 percent of Social Benefits and the Ruissian Inddustrial Firm 73 firms in the sample reported receiving government transfers in 1994, over 47 percent reported running tax arrears in the same period. This suggests that transfers tend to arise more ex post than through explicit prior ar- rangement, including compensatory finance. In addition, the continuing concentration of the largest transfers relative to output in firms withi large employment suggests that the decision on transfers may be primarily a function of the employment preference of the government, rather than a set of decisions conditioned on the supply of social benefits. It just so happens that large firms tend to be firms with the greatest exposure to benefits. Structure of Compensation, Benefit Pricing, and Incentives The survey is consistent with aggregate data in showing that since 1990 the monetary component of workers' compensation has generally de- clined in real terms. This is true for privatized as well as state firms. Mak- ing allowance for hours adjustment would only accentuate this decline. Is there evidence that benefits have been used as a counterweight to cash wages? This question is particularly relevant given the bias imparted by the tax regime, with only cash wages subject to the excess wage tax, as well as a rising share of payroll taxes in relation to wage costs. Figure 3-1 relates the change in the cost of benefits per worker to that for mean wages. Most observations fall below the 45-degree line, with a large asymmetry in respective changes to wages and benefits. This would sug- gest that benefits have not generally been explicitly used to offset real wage declines. In general, results from this survey and from the smaller World ]3ank survey of twenty-two firms indicate that wages were generally set with reference to consumer prices, albeit with a lag, while benefits were, in effect, treated by firms as a quasi-fixed cost, and hence not subject to ex- plicit bargaining with workers over the level or quality of services sup- plied. And the latter survey explicitly reveals that changes in the value of benefits received by workers were not reflected in cash wage settlements. That firms have not used benefits as substitutes for wages is a bit mis- leading. Table 3-11 has already indicated that state and privatized firms very significantly underestimate the cost of benefits. Imputing the costs reported by de novo firms for roughly comparable benefits provision has a strong effect on the structure of total compensation. Using these values, 74 Employment, Wages, and the Provision of Social Benefits Figure 3-1. Change in Average Wages and Cost of Social Benefits, 1990-94 Percentage change in cost of social benefits per worker 1,200 - - 1,000 - _ 800 - _ 600 - - 400- - 9 -500 5 0 1,000 1,500 2,000 2,500 -200 Percentage change in average wages Note: Data for 1994 are mid-1994. benefits comprise about one-third of total worker compensation, and this share would obviously rise if only cash wages were subject to downward adjustment through involuntary leave and short-time work, as is likely. Further, it seems probable that even these adjusted cost of benefits num- bers may be underestimates. Changing this structure of compensation is clearly important. This is true for several reasons. First, there is the incentive issue, which we deal with in more detail below. Second, there is the dynamic problem of rais- ing cost recovery in a context where monetary wage levels remain low. It is evident that, in principle, provision of benefits by firms can be offset by adjustment to cash wages-benefits such as canteens, holiday homes, and the like, could be balanced by a lower cash wage-and this would be po- tentially true, even in cases where firms were providers of local public goods. In that context, a tax could be imposed on all users, potential or actual, to cover the costs of such services. But raising cost recovery en- counters the constraint imposed by the level of cash wages. Cash wages were low at the start of the transition, and have generally gone lower. In contrast, the shadow or market value of benefits will generally have in- creased, particularly given the thinness of the markets for alternative pro- Social Benefits and the Ruissian Industrial Firm 75 vision, which will have tended to push prices above their likely steady- state values. Indeed, benefits, whether provided by the firm or municipal- ity, have effectively anchored household incomes. This is partially confirmed by current monetary wage levels: about 20 percent of workers in our survey received wages at or below regionally adjusted poverty lev- els (Commander, Dhar, and Yemtsov 1995). These changes in the struc- ture of compensation may imply that, properly priced, totaTl real compensation has actually increased through the transition, but real dis- posable compensation appears to have fallen. By liberalizing the wage regime-at least partially-cash wages could be expected to act as a conventional, worker-specific incentive. Further, workers' wages would increasingly reflect returns to firm-specific skills. Evidence from our survey suggests, however, that wage differentials have not moved in this way. Region or sector remains a better explana- tory factor, and relative wages have remained very inertial. Among other things, this suggests that any wage premium would come througlh sec- toral or regional attachment. This is potentially important in several re- spects. First, attachment to firms could be predicted to be low, because firm-specific rents will be small. Second, given low mobiLity at a regional level, flows across sectors will depend very much on the degree of sec- toral attachment in Russian industry. While far from conclusive, there is some evidence to suggest a fairly high degree of sectoral attachment, with flows across the same sector dominating among voluntary separations. Transitions will thus tend to be sector- and region-specific, with compen- sation variation within a sector serving as the main motivation for transi- tions. Third, such flows will tend to be motivated primarily by nonmonetary compensation or social benefits when that component of compensation is large. But in effectively reducing the share of monetary wages in total effective compensation, and hence the share of indivictual- specific compensation, the component that is likely to be more evenly dis- tributed-benefits-will have risen. It is quite evident that this creates an adverse incentive problem and will negatively affect the efforts of indi- vidual workers in firms that follow such a compensation scheme. The im- plications for effort are dealt with more formally in the appendix to this chapter. Finally, we should note that reducing cash wages to balance the cost of services would obviously be more difficult in the case of healthcare and childcare facilities, because potential costs will vary widely ac:ross 76 Emnploymenzt, Wages, and t7le Provisioni of Social Benlefits workers and may not be identifiable. To the extent that such risk was identifiable, downward adjustment to cash wages to balance firm provi- sion would only be feasible when there was no other provider and where the worker would otherwise have to pay for access to similar services. Where risks are clearly identifiable through worker characteristics-for instance, provision of childcare to workers' children-we could predict that firms, given greater freedom in decisionmaking, would likely cut back or eliminate such services. Kindergartens are probably the clearest case, and the initially higher incidence of involuntary separations among women may perhaps be linked to an unwillingness to retain or hire staff with a high risk of using costly firm-specific benefits. Curtailing such benefits would thus reflect shifts in the structure of firms' labor demand and, in due course, in the labor supply decisions of women. Indeed, as we have seen above, childcare facilities have been among the main casualties of firm-provided benefits. This outcome is likely to have been accelerated by pricing caps. Conclusions We have demonstrated that Russian industrial firms have, for the most part, continued to supply social benefits to workers. There has been little asset disposal and, more significant, relatively little change in the volume and range of benefits provided. Firms appear to view provision of social benefits as a necessary function, and one consistent with their social-ethical obligations to workers and the community. Ownership change through privatization has yet to affect behavior. We can think of the resulting compensation envelope and the outside alternatives as acting on two sets of decisions: those of firms and those of workers. These may in part overlap, given the apparently strong voice that insiders, including workers, appear to exert in firm decisions on em- ployment and wages. Nevertheless, we can characterize compensation and fallback income as follows. Money wages remain low and have drifted slightly downward in real terms, while their share in total com- pensation has declined. The fall-back given by unemployment benefits remains yet lower, and drops significantly below the subsistence level. From the viewpoint of firms, this outcome promotes what we have characterized as benevolence and its resulting inertia in the employment level. Firms remain reluctant to cast workers into unemployment when Social Beiefits and the Russian Iniduistrial Firm 77 benefits continue to be so low. At the same time, the threat of unemploy- ment induces workers to accept greater wage flexibility, including hours adjustment and temporary layoffs. Low money-wages depress furthter the share of labor costs in total costs to firms. In 1993 and 1994 labor costs av- eraged 13.4 percent of total costs in industry, with payroll taxes and other contributions amounting to an additional 5 percent of costs."1 The low la- bor share, in turn, has sanctioned continued labor hoarding in industrial firms. But it has also impeded steps to greater cost recovery, because firms remain constrained by low-wage regimes and continuing price con- trols. Finally, from the viewpoint of the firm-and particularly large firms-the level of employment, as well as benefits, appears significant in explaining subsidy or transfer decisions by government. This may addi- tionally explain the continuing reluctance to separate. From the perspective of the worker, we can think of her decision pri- marily through the factors pulling her out of a firm, given the reluctance of firms to impose involuntary separations. Unemployment benefits have stayed low, and average private sector wages appear to offer relatively small markups over those paid in state and privatized firms (see Com- mander, Dhar, and Yemtsov 1995). Further, de novo private sector wages incorporate lower social benefits provision, depressing the value of total relative compensation. In sum, even with low money wages in state or privatized firms, workers will have a strong incentive to stay in firms with higher social benefits levels, and thus higher values of total compen- sation; there remain quasi-rents to be extracted within the firm.12 The cor- ollary of this compensation envelope, however, will be that workers will make their effort decisions consistent with worker-specific, as opposed to average, returns to effort. That money wages remain low and declining as a share of total compensation will tend to result in workers allocating their time and effort in order to maximize individual-specific returns. Be- cause firms may not be able to monitor effort adequately and may be pre- pared to accept low effort by workers when faced with large negative demand shocks and consequent low capacity utilization, low effort in pri- mary employment can result, with workers diverting as much of their disposable time as possible to secondary employment. This decision structure relies on workers choosing to stay with their primary employer, mainly as a function of the nonmonetary, or social benefits, component of compensation. We formalize this argument in the appendix. It will thus require that the alternative wage roughly equate the value of total com- 78 Employmenit, Wages, and the Provision of Social Benefits pensation, including secondary income flows and social benefits, for workers to be pulled from their jobs. Equally, quits could be motivated by high relative benefits provision. Several implications flow from our analysis. The first is that there are indeed sound reasons for detaching some benefits provision from firms. Most of the goods provided by firms do not qualify as public goods. Fur- ther, provision of such goods, given the problems in raising cost recov- ery, have clear implications for firm competitiveness. The evidence that they receive but partial compensating transfers from government also im- poses a clear burden on some firms. Moreover, in the new environment firms are unlikely to make optimal decisions on the provision of goods that have merit characteristics, such as healthcare and childcare. In cer- tain cases, we can imagine firms undersupplying services, such as child- care, with nonneutral effects on labor supply. Second, dissociating the supply of such goods from firms will assist in the adjustment of the wage structure and in the fuller monetization of total worker compensation. As matters currently stand, the low level of money wages and their dimin- ished share in total compensation generates adverse incentive effects, raising the incentive for workers to stay in benefits-providing firms, while simultaneously delivering low effort in primary employment. Al- though this assists in maintaining low aggregate unemployment levels, it offers no viable long-run solution and may ultimately impede effective restructuring and the growth of an autonomous private sector. Appendix Social Benefits, Wages, and Effort Motivating workers was a persistent problem for Soviet firms. Because shirking could not be punished by unemployment, thus diluting the ef- fectiveness of any monitoring scheme, firms tried to introduce incentive payments, primarily through piece rates.13 Even this effort was compro- mised by the existence of fairly rigid wage structures, or scales that re- stricted wage differentials. Returns to human capital were often perverse. The monetary component of wages varied relatively little across indi- viduals, controlling for skill, location, and other factors, while nonmone- tary compensation tended to be available to all workers in a firm. Perhaps Social Benefits and the Rtussian Indtustrial Finn 79 not surprisingly, the history of labor productivity in the fonner Soviet Union was not a happy one. The adverse incentive effects flowing out of this compensation struc- ture have been amended in the transition in several respects. Unemploy- ment has been tolerated but remains low, and wage setting has been partially liberalized. The structure of workers' compensation, hovvever, has remained fairly stable. For reasons indicated in the chapter, there has been significant employment inertia in state and privatized firms, along- side a rapid growth in multiple jobholding. Recent survey evidence indi- cates that so-called secondary employment can account for a major share of monetary compensation (see de Melo and Ofer 1994; Commander, McHale, and Yemtsov 1995). The simple model we develop below tries to account for agents' behavior in a world with two basic choices in time and effort allocation. In this world, workers can combine jobs, rather than necessarily select discrete states (for example, state or private employ- ment or unemployment). It can readily be shown that when the relative monetary wage in the state or privatized sector is small, workers will have a strong incentive to reduce effort, subject to a minimum effort requirement, and to allocate as much of their time as possible to secondary work. A Model of Time Allocation across Two Sectors Consider a world in which workers can supply positive hours (h,o,lJ) of work in two sectors: state, ho, and private, hl. The utility function of a rep- resentative worker, neglecting substitution effects, can be written as: (3-1) U ( c,ho,hi ) =u (c)+v ( T-ho-hh ) where c is total compensation, comprising money and in-kind compensa- tion or social benefits, and v(C) is a concave function of leisure. The budget constraint is given by: (3-2) c=s -(ho-hmin)+wo(ho-hmin)+g( ( ) where hmin is the minimum time a worker needs to work in the state sec- tor to get access to social benefits, given as s. 8(ho - hj) 1, hhrnin 0, h < lmin 80 Enmployment, Wages, and the Provision of Social Benefits The term g(hil) is income in the private sector, and this is assumed to be a concave function of the hours, hi, supplied to the private sector. Assum- ing decreasing returns to hours of work in the private sector, we have: g ( h )=go- h, 0<1. Maximizing the utility function (choosing ho,hi), subject to the budget constraint (3-2) and the inequalities, ho 2 0, h, 2 0, we have the following first-order conditions: II' zv0 - v' = 0 It' g' - v' =0 from which we find: 1z= [0 . gTY1- and g(h1) 1 ZVI 1 hi 0 is the average wage in the private sector. To get ho we can assume for simplicity that it(c)='yC, v(X)=0VDx' X@, v1. We then have y 0o - ov ( T - ho - h) 0 , or h, T ( go0 (I .o) It is assumed that all parameters are such that the condition ho 2 hmin is satisfied. It follows from the above that the lower total compensation in the state compared with the private sector, the fewer will be the hours sup- Social Benefits and the Rassiani Indutstrial Firm 81 plied to the former above the minimum requirement, hmi that workers must satisfy in order to maintain access to social benefits. References Aghion, Philippe, Olivier Blanchard, and Robin Burgess. 1994. "The Be- havior of State Firrns in Eastern Europe Pre-Privatization." Eliro- pean Econonmic Reviewv 38 (6):132-49. Boycko, Maxim, and Andrei Schleifer. 1994. "Next Steps in Privatisation: Six Major Challenges." In Ira Lieberman and John Nellis, eds., Ruts- sia: Creating Private Enterprises and Efficient Markets. Washington, D.C.: World Bank. Commander, Simon, and Richard Jackman. 1993. "Providing Social Bene- fits in Russia: Redefining the Roles of Firms and Government." PRE Working Paper 1184, World Bank, Washington, D.C. Commander, Simon, Sumana Dhar, and Ruslan Yemtsov. 1995. "Hlow Russian Firms Make their Wage and Employment Decisions." Pa- per presented at Conference on Economic Policy and Enterprise Restructuring in Russia, St. Petersburg, June. Commander, Simon, John McHale, and Ruslan Yemtsov. 1995. "Russia." In Simon Commander and Fabrizio Coricelli, eds, Unemploynent, Restruicthring, anid the Labor Market in Eastern Eutrope and Russia. EDI Development Studies. Washington, D.C.: World Bank. de Melo, Martha, and Gur Ofer. 1994. Private Service Firmis in a Transitional Econonny. Studies of Economies in Transformation No. 11. Wash- ington, D.C.: World Bank. Estrin, Saul, Mark Schaffer, and Inderjit Singh. 1994. "The Provision of Social Benefits in State-owned, Privatized and Private Firms in Po- land." Paper presented at Workshop on Enterprise Adjustmeni. in Eastern Europe, World Bank, Washington, D.C., September. Freinkman, Lev. 1994. "Note on Social Expenditures." Europe and Cen- tral Asia Department, World Bank, Washington, D.C. Photocopy. Oxenstierna, Suzanne. 1990. FronI Labouir Slhortage to Unemiployment? Thle Soviet Labouir Market in thie 1980's. Stockholm: Elsevier. Schaffer, Mark. 1995. "Contribution to a Roundtable on Social Benefits." Econiomics of Transition Roitnndtable 3 (2):247-50. Starodubrovskaya, Irina. 1995. "Note on Housing Divestiture and Mu- nicipalities." World Bank, Moscow Resident Mission. Photocopy. 82 Employment, Wages, and the Provision of Social Beniefits Tanzi, Vito. 1995. "Fiscal Federalism and Decentralization: A Review of Some Efficiency and Macroeconomic Aspects." Fiscal Affairs De- partment, International Monetary Fund, Washington, D.C. Photo- copy. Wallich, Christine, ed. 1994. Ruissia and the Clhallenge of Fiscal Federalism. Washington, D.C.: World Bank. Notes 1. These numbers are for the entire firm sector except where noted (source: Goskomstat). These figures are likely to underestimate spending on social bene- fits, because they exclude barter transactions, subsidies on goods and services, and investment in infrastructure by firms; see Freinkman 1994. 2. It is currently six times the minimum wage, except for enterprises within the military-industrial complex (eight times the minimum wage). Agricultural en- terprises and those within the food processing branch are exempted. The current limits were adjusted as of 1994 from prior limits of four and six times the mini- mum wage, respectively. 3. Carried out as part of the preparatory work for the project on Social Asset Divestiture. Thanks to Mari Kuraishi and Dennis Whittle for access to this dataset. The earlier results are reported in Commander and Jackman 1993. 4. About 25 percent of firms did cite benefits as an unavoidable burden, possi- bly a grudging equivalent to the social obligation. 5. This is in stark contrast to the general assumption that benefits supply is likely to impede efficiency and deter outside investment-as expressed, for exam- ple, in Aghion, Blanchard, and Burgess 1994 and Boycko and Schleifer 1994. 6. In only three instances did the enterprise indicate that the disposal of the social asset was mandated by government regulation or decree. Government regulations require enterprises to divest social assets to municipalities, but this has yet to be implemented on a large scale. 7. Enterprises were classified as located in a dominant industrial town based on name and location of the enterprises. These enterprises and their location are generally commonly known as such. 8. These figures are based on a note by Irina Starodubrovskaya on the "Pro- gress of Housing Reform," World Bank Resident Mission, Moscow, February, 1995. 9. Estimates taken from the draft of the World Bank, "Russia Country Eco- nomic Memorandum," May 1995. Our thanks to Dennis Whittle for help with the housing numbers. 10. With such uncertainty, it is hardly surprising that housing privatization has remained restricted. The small World Bank survey in nine cities gave a range Social Benefits and the Rtussian Industrial Firnm 83 of 13-50 percent for municipal housing and 16-40 percent, with a mean of 30 per- cent, for remaining firm housing. 11. These comprise pension, medical insurance, social insurance, and employ- ment fund contributions. On average these encompassed just under 40 percent of the wage bill for Russian industry in 1993 and 1994. 12. This may, of course, be compounded by expectations with respect to wealth effects coming from privatization and changes in control structures. Counteracting this may be that access to housing benefits have now been made unequivocally independent of employment status. 13. An overview of compensation schemes and associated labor market poli- cies can be found in Oxenstierna 1990. Part II Financial Aspects of Enterprise Restructuring 4 "Arrears" in the Russian Enterprise Sect,or Gilles Alfandari and Mark E. Schaffer This chapter offers an analysis of payment arrears in the Russian enter- prise sector, using both official (Goskomstat) aggregate data and data from the World Bank survey of 439 large and medium-size Russian in- dustrial enterprises conducted in mid-1994. Enterprise arrears in Russia attracted considerable attention from both analysts and policymakers fol- lowing the rapid growth of trade credit ("interenterprise debt") that en- sued after the start of the Russian reform program and that initially culminated in the so-called "arrears crisis" of 1992. Since then, the levels of both total trade credit and trade credit in arrears have stabilized, and attention and concern is now directed at the other forms of "arrears" of Russian enterprises as well-not just arrears to other firms, but arrears to banks, to the government, and to workers. This chapter will focus primar- This chapter is based on the authors' paper, "On 'Arrears' in Russia," presented at the World Bank/Ministry of Economic Conference on Economic Policy and En- terprise Restructuring, St. Petersburg, June 1995, and at a World Bank workshop in March 1995. We are grateful to Simon Commander, Qimiao Fan, Lev Freink- man, Alan Gelb, Vincent Koen, Olga Shabalina, two referees, and the conference and seminar audiences for helpful comments and suggestions. 87 88 Financial Aspects of Enterprise Restruictiring ily on trade credit arrears, tax arrears, and wage arrears. Fan, Lee, and Schaffer discuss bad bank debts in detail in chapter 5. We begin by distinguishing between two kinds of arrears, which dif- fer by nature (rather than by creditor). The first kind of arrears is that as- sociated with "late payment": firms pay a debt late, but they do eventually pay. In stocks and flows, the stock of arrears from late pay- ment is roughly stable over time because over the medium-term, inflows of arrears (new debts that come due but are not paid) are roughly equal to outflows (the actual payments of debts in arrears). These kinds of ar- rears can act as a short-term "cushion" for firms that have liquidity prob- lems or are suffering from some other kind of "financial stress," and that react to this stress by temporarily delaying payment. They may also re- sult from other causes-for example, customers extracting, ex post, better payment terms from their suppliers. The second kind of arrears is that associated with firms in serious fi- nancial difficulties. The arrears of firms in financial distress can be con- sidered a kind of "bad debt"; most or all of these debts will not be repaid, at least in the short or medium term. We also briefly discuss a third category of arrears, which can be called "strategic arrears," or in Perotti's (1994) terminology, "collusive arrears." Arrears may arise as the result of strategic behavior. For example, firms may expect a general government bailout of arrears; they anticipate this bailout by not paying each other-that is, by running arrears-and in the face of the resulting rapid growth in arrears the government gives in and bails out the firms by clearing the arrears with, for example, an injection of credit. Using this framework we shall try to address a number of issues re- garding "arrears" in Russia, including: * Is there an "arrears crisis" in Russia? Are stocks of arrears continu- ously growing, and if so, which kinds of arrears (trade credit arrears, tax arrears, and so forth) are behaving in this manner? Are arrears really a source of soft budget constraints? If so, which kinds of arrears are in this group? . What causes arrears, and which firms are the most likely to run them? * Have enterprises learned how to deal with bad payers? * What should be the role and contribution of policymakers? "Arrears" in the Ruissian Enterprise Sector 89 Late Payment versus Bad Debts: From Financial Stress to Financial Distress To address the question of what causes arrears, we can distinguish in principle between two kinds of overdue payments: "late payments" and "bad debts." Late payments are overdue payables of enterprises that will be paid in the short term. Bad debts are the arrears of enterprises that cannot pay, or the overdue payables that could not be covered by the en- terprise's surplus in the short or medium term, even after drastic adjust- ments. Late payments are, by definition, paid back. They are one source of liquidity, among others, for the enterprise, and thus can be described as a "cushion" for the firm. The decision to run such arrears can been seen as resulting from intertemporal optimizing behavior. Enterprises de- cide to pay certain debts late because the cost associated with late pay- ment is smaller than the cost required by alternative sources of finance. Bad debts, in contrast, are held by firms in severe financial difficulties. These firms are unable to repay the debts in the short term, and are un- likely ever to repay them in full. In a market economy, these debts would be settled during the processes of reorganization or liquidation. In the former, the debts would be renegotiated as part of the overall settlement between the firm and its creditors; in the latter, debts would be settled as the firm is dissolved, its assets disposed of, and the proceeds used to re- pay debts. In transition economies, problems with the process of bankruptcy (both reorganization and liquidation) mean that bad debts persist be- cause bad debtors persist. While distressed firms may downsize subs tan- tially under pressure, they are not often pursued by their credi tors through the legal system for payment of their late debts. There are a num- ber of reasons for this, including the low liquidation value of the firm, which means creditors can expect a low return to court action (especially by unsecured creditors); the slowness of bankruptcy courts in transition economies; the lack of previous experience with bankruptcy procedures, which means that final outcomes will be difficult to predict; the lack of practical experience in working out bad debts; reluctance of creditors iKes- pecially banks) to pursue bad debtors openly because it signals that the creditors have dubious assets; and so forth (see Gray, Schlorke, and Szanyi 1995 for results of an empirical investigation of the bankruptcy and liquidation process in Hungary). Faced with these obstacles, credi- 90 Finanicial Aspects of Enterprise Restrictutring tors apparently often decide simply not to pursue the bad debtors. More- over, the very large shocks and recessions that have hit transition econo- mies mean that distressed firms, and hence bad debts, are plentiful. Thus there exist large "stock problems," but the mechanisms for sorting out these stock problems are ineffective. One way to see the difference between late payments and bad debts is to take the creditor's point of view. Late payments are credits implicitly accepted by the creditor. In the specific case of trade credit, for example, suppliers must see advantages in continuing to supply firms that pay late-otherwise they would stop deliveries. Of course, these advantages include (the expectation of) eventually being paid. When a firm initially gets into difficulties, it may delay paying its creditors; these creditors may continue to extend new credits (for example, supply on trade credit) if they believe this is a case of late payment and the firm will eventually pay. If the firm's difficulties are severe, however, and it continues to fail to pay, the arrears become bad debts. At this point creditors can stop the flow of new credits; for example, suppliers can simply stop shipping on credit. They cannot, however, easily reduce the stock of credit that has ac- cumulated because of the difficulties in pursuing recalcitrant debtors cited above. While the bad debt stock problem persists, simple market mechanisms can deal with the flow problem. Note that not all of the debts of a firm in financial distress may be "bad." Consider a firm that is in severe difficulties, but has a profitable core after downsizing-that is, it can produce something profitably. It may have debts to suppliers that it cannot and will not repay, and these suppliers will not extend new trade credit, but the firm may have alter- nate suppliers whom it needs-and pays-and these alternate suppliers may choose to grant the firm trade credit. Or, to cite another example: as we shall see below, a response of Russian firms to financial distress is to try to pay suppliers, and instead not pay taxes. Some of the payables to suppliers may be late payments, in that these distressed firms still pay the suppliers, but overdue taxes may be bad debts, because they are unlikely ever to be repaid. Whereas bad debts on the balance sheet are a reflection of a firm's fi- nancial distress, some late payments can be seen as a short-run answer to financial stress. How do we distinguish between "stress" and "distress"? Financial distress is a permanent characteristic: enterprises in this state are chronic loss-makers. These losses may be ongoing operating losses; "Arrears" iiz the Ruissian Enterprise Sector 91 that is, the firm's revenue cannot cover even its basic operating costs (raw materials; labor costs; basic taxes; and, in the medium term, capital costs). Alternately, the firm may be covering its operating costs but failing to cover total costs, including servicing or repaying debts-for example, if the firm has experienced a major loss of markets, has completed downsiz- ing to a "profitable core," but is now highly indebted. Financial stress, in contrast, is a short-term deterioration in prof:its or financial health, resulting from, for example, external shocks that will not last, or problems that can be overcome by marginal adjustment (stopping the least profitable part of production, perhaps temporarily; laying off some workers, again perhaps temporarily; and the like). It is seen as a temporary obstacle that does not threaten the enterprise's existence, butt re- quires additional short- or medium-term liquidity, because adjustment usually takes time. We implement this framework empirically as follows: * Bad debts are associated with chronic loss-making and should be concentrated in enterprises with the poorest performance. Late payments should be correlated with indicators of financial tightness. They should occur when credit is tightening, as a re- sponse to demand shocks, and should be directly related to liquid- ity indicators. In the rest of this chapter we shall bring together empirical evidence that late payments-and not bad debts-constitute the bulk of tiade credit arrears (interenterprise arrears) in Russia. We will also argue that the situation is reversed in the case of tax arrears: that is, tax arrears rather than trade credit arrears are concentrated in financially distressed firms. First, however, we consider definitional issues, and then we put the overall picture for Russia into perspective with the help of some interna- tional comparisons. Definitions and Measurement Table 4-1 presents a simplified balance sheet of a Russian enterprise. In its basic format it is no different from a balance sheet in any other country. On the liabilities side appear the various sorts of debts of the firms. We can group these liabilities according to four different kinds of creditors: (1) suppliers of goods and services received by the firm but not yet paid 92 Financial Aspects of Enterprise Restructutring Table 4-1. Simplified Balance Sheet of a Russian Enterprise Assets Liabilities I. Fixed capital and long-term assets I. Own funds [equity] Nonmaterial assets Equity Fixed capital Reserve fund Long-term financial investments Profit II. Inventories II. Long-term liabilities Long-term bank credit III. Short-term (liquid) assets Long-term loans ("Zaimy") Total debtors ("Debitory") Receivables for goods and services III. Short-term (liquid) liabilities Receivables from the budget Short-term bank credit Receivables from extrabudgetary funds Bank credit for employees Loans to employees Short-term loans ("Zaimy") Accounts with other debtors Total creditors ("Kreditory") Short-term financial investments Payables for goods and services Monetary assets ("Denezhnye sredstva") Payables for wages Cash Payables to the budget Ruble bank deposits Payables to extrabudgetary funds Hard currency bank deposits Advance payments Other monetary assets Other short-term liabilities Losses Balance (I + II + III + Losses) Balance (I + II + III) for (that is, commercial payables or trade credit received); (2) banks and other lenders; (3) the government (tax payables), including extrabudge- tary funds; and (4) workers. On the asset side of the firm's balance sheet appear the physical (fixed capital, inventories) and financial assets of the firm. The latter can also be categorized according to the kind of debtor to the firm. The most impor- tant for the purposes of this chapter are claims on customers for goods and services shipped by the firm but not yet paid for (that is, commercial receivables or trade credit extended). Here we will refer to "arrears" as any overdue liabilities of firms. The qualification "overdue" is very important; simply owing somebody money doesn't put the debt in arrears. We will focus on the following sorts of arrears: "Arrears" in the Rufssian Enterprise Sector 93 * Overdue payables to suppliers (= commercial payables in arrears, = overdue trade credit received) * Overdue receivables from customers (= commercial receivables in arrears, = overdue trade credit extended) * Overdue liabilities to banks (overdue bank debt and/or interest ar- rears) * Overdue payables to the budget and extrabudgetary funds (= tax arrears) * Overdue wages (= wage arrears). Data Issues Goskomstat collects monthly data from enterprises on selected balance sheet items (nominal value of stocks at the start of the month) and distin- guishes between the total stock and the overdue portion, that in arrears. These data are discussed in detail in the appendix to this chapter. Gosk- omstat has collected arrears data from medium-size and large industrial and construction enterprises since early 1992, and since late 1993 from firms in transport and agriculture as well. Because "overdue" is defined by the reporting firm, it is natural to expect that respondents will under- report their overdue payables, and the percentage of payables overdue is indeed usually smaller than the percentage of receivables overdue. The difference, however, is small (about 3-5 percentage points). The 1994 World Bank survey of 439 state-owned, privatized, and de novo industrial firms asked enterprises to provide similar balance sheet data as of 1 April 1994. The survey balance sheet data are better specified than the Goskomstat data in a number of respects (for example, maturity structure of payables, treatment of interest and late penalties). In the im- portant cases of tax arrears and arrears to banks, the coverage of the sur- vey data is superior to that of the Goskomstat data. The survey data include arrears to the budget, as well as to extrabudgetary funds (such as Social Security), while the Goskomstat data collected at the time covered only arrears to the budget. Arrears to banks in the survey include interest arrears as well, while the Goskomstat data cover only overdue principal. The treatment of penalty interest and unpaid interest is important in interpreting empirical estimates when inflation is high (see below) and is summarized in box 4-1 (for more details see the appendix). 94 Financial Aspects of Enterprise Restruicturing Box 4-1. Penalty or Unpaid Interest Included? Goskomstat World Bank survey Total receivables Included Not applicable From customers Not included Included Total payables Included Not applicable To suppliers Not included Included Taxes Included Included Payables to banks Not included Included Some caution is in order regarding the treatment of interest in over- due payables to suppliers. It is likely that debtor firms will not always ac- knowledge penalty interest charged by suppliers by recording it in their balance sheets, and indeed the charging of penalty interest on late pay- ments by customers is not very widespread (see below). This will have the effect of understating the payables in arrears, even in the World Bank survey data. The qualitative section of the survey questionnaire addresses quite specific issues related to arrears that help us understand the payment practices, behavior, and expectations of Russian managers facing overdue payments. A comparison between the survey and Goskomstat data (table 4-2) suggests that the data collected by the survey are fairly representative of aggregate industry in the sense that they match Goskomstat data fairly closely. Most of the differences between the two can probably be ex- plained by differences in coverage and definition. Measitring Arrears First, a common indicator used to measure the relative level of arrears is the "percentage overdue," in which the nominal value of the overdue portion is given as a percentage of the total payable (or receivable) nomi- nal stock. Because payables and receivables are recorded at the time the credit is extended, the high level of inflation in Russia causes the overdue percentage of any payable category to be understated, because it is usu- "Arrears" in the Ruissiani Enterprise Sector 95 Table 4-2. Structure of Liabilities and Receivables, and Portions in Arrears, Survey and Goskomstat Data (percent) Sutrvey Goskomstat data (un weighted average) (aggregate industry) Arrears as Arrears as percentage percentage I April 1994 Structure overdufe Structure overdu,_e Liabilities Payables to suppliers 55a 58a 52b 45b Liabilities to banks 20a 28' 13b 12b Tax payables 15c 46c 14d 44d Other liabilities 9 21 Payables to employees 9 60e n.a. 41e Receivables Total receivables From customers 75 50 From other domestic firms 82 74 n.a. n.a. From budgetary organizations 7 66 In subsidies from government 3 67 From former Soviet Union trade 3 87 From non-former Soviet Union foreign trade 6 62 n.a. Not available. a. Including interest arrears. b. Not including interest arrears. c. All government. d. SociaL Security not included. e. In percentage of monthly wage bill. Source: World Bank survey; Goskomstat. ally older than the nonoverdue portion. For instance, the percentage of trade credit overdue will automatically rise when inflation falls because the downward bias decreases. This can be offset by the practice of charg- ing interest on late payments. Even when interest is charged, Goskom- stat's definitions sometimes exclude interest from the overdue portion 96 Finanicial Aspects of Einterprise Restrutctiriing (see above), but as we have seen, this is less of a problem with the World Bank survey data because the survey's definitions of arrears can include interest. A second convenient measure is a stock/flow ratio: that is, aggregate payables or receivables as a percentage of gross domestic product (GDP), or as a ratio of sales or production. The receivables/sales ratio is particu- larly useful because it gives an approximate average payment period. Similarly, the overdue receivables/sales ratio gives the average delay pe- riod. Again, because trade credit (like inventories) is measured in actual purchase/sales prices, the high level of inflation in Russia since 1992 means some care must be taken when constructing these stock/flow ra- tios.' Since (as we shall see below) average payment periods are usually about two months, our trade credit/sales ratio for Russia is the ratio of end-period trade credit to a two-month moving average of production. The adjustment for the trade credit/GDP ratio is still simpler: we use the ratio of end-month trade credit to annualized monthly GDP (that is, monthly GDP*12). It should be noted here that Russian managers have now had exten- sive experience with inflation, and we would expect the prices they charge customers to include an allowance for anticipated inflation; that is, a charge for the cost of trade credit extended (Koen and Phillips 1993). This practice, like the practice of charging interest on late payments, would mitigate the problem of inflation in measuring the volume of over- due trade credit. Nevertheless, the difficulties caused by inflation should be borne in mind when interpreting the empirical evidence. Estimates of the real net flow of arrears-the difference between real inflows and real outflows-can be calculated from the change in real stocks. Again, some caveats are to be kept in mind. First, the volume of arrears may appear to decrease because some arrears are formally re- scheduled (and hence cease to be overdue). This is particularly a problem with overdue bank credit, because rescheduling of arrears to banks is common in Russia (as evidenced by the firms in the World Bank survey; see chapter 5). Another difficulty arises when write-offs of arrears are granted: we cannot distinguish between outflows from repayment of ar- rears and outflows from write-offs. We expect this to be more of a prob- lem with tax arrears. Finally, because the treatment of interest arrears and penalties affects estimates of the stocks of arrears, it will affect estimates "Arrears" in the Russiani Eiterprise Sector 97 of the net flow as well. As already noted, this is a problem with some categories of Goskomstat data, but not so much with the survey data. Arrears in Perspective: Aggregates and International Comparisons In this section we look in some detail at the scale and recent trends in en- terprise liabilities, and in the overdue portion, arrears. We use both ag- gregate Goskomstat monthly balance sheet data and data from the World Bank survey. We will also make some comparisons between Russia and leading transition and market economies. We begin with a brief look at the balance sheet structure of Russian in- dustrial enterprises on 1 April 1994 as reported by Goskomstat (table 4-2 and figure 4-1). At that date, payables to suppliers accounted for slightly more than half of total liabilities. Bank credit and payables to the budget each account for another 13-14 percent of total liabilities. Wage arrears make up 2 percent of total liabilities, and the remaining 20 percent are "other liabilities." Most important among these miscellaneous items are payables to extrabudgetary funds and interest arrears (unpaid interest) to banks. Penalty interest on late payables to suppliers are also included in "other liabilities." The survey data, and Goskomstat data from early 1995, suggest that payables to extrabudgetary funds amount to about 5 percent of total liabilities; that is, total payables to all of government would amount to perhaps 15-20 percent of total liabilities. The World Bank sur- vey suggests total liabilities to banks, interest arrears included, would come to about 20 percent of total liabilities.2 Table 4-2 offers a breakdown of enterprises' receivables based on the World Bank survey (detailed Goskomstat data are not available). Receiv- ables come mainly from domestic firms (more than 80 percent of total re- ceivables). Foreign firms and all of government each contribute 10 percent in the structure of receivables. Trade credit extended to custom- ers is greater than trade credit received from suppliers, meaning that in- dustrial enterprises are net trade creditors to the rest of the economy (and abroad). According to the Goskomstat data, as of 1 April 1994, 45 percent of payables to customers, 44 percent of payables to the budget, and 8 per- cent of bank credit were overdue, but there are definitional prob:Lems 98 Financial Aspects of Enterprise Restructutring Figure 4-1. Structure of Liabilities of Industrial Enterprises, 1 April 1994 (percenitage of total liabilities) Total payables to suppliers 51.8% Overdue to suppliers 23.4% Total payables to budget 13.5% Wage arrears 1.9% Overdue to the budget 5.9% Other liabilities Total bank Overdue credit bank credit 13.3% 1.6% with the first and third of these figures. Because interest arrears are po- tentially large in magnitude relative to the actual principal of the loan in a high-inflation, high nominal interest rate environment, the Goskomstat figure for overdue bank credit is particularly badly affected by the failure to include interest arrears. The World Bank survey and Central Bank of Russia (CBR) data suggest that a more accurate figure for the first half of 1994 would be that 20-30 percent of total liabilities to banks are overdue (see chapter 5 for further discussion). The single largest category of arrears is overdue trade credit received (overdue payables to suppliers), which accounts for fully one-quarter of total liabilities in the second quarter of 1994. The World Bank survey sug- "Arrears" in the Russiani Enterprise Sector 99 gests that this Goskomstat figure may be an understatement, perhaps again because of the treatment of interest/penalties, although this may also indicate that the firms in the World Bank survey have more arrears to suppliers than the average Russian industrial firm.3 Total arrears to all government, banks (correcting for the treatment of interest arrears)., and workers (wage arrears) would each amount to about 5-7 percent of total liabilities on that date. We note, however, that while wage arrears in Rus- sia are small in relation to total liabilities and trade credit arrears, they are substantial compared with the aggregate wage bill. According to Gosk- omstat, on 1 April 1994, wage arrears in industrial firms amounted to close to one-half of the monthly total wage bill in industry (that is, the equivalent of about two weeks of wages). In addition, wage arrears are fairly common: 30 percent of firms in the World Bank survey have them. We will briefly review the main patterns in arrears in Commonwealth of Independent States (CIS) trade with Russia. We are very limited here by the data available on this issue, because Goskomstat statistics do not cover the trade and services sector, and a considerable portion of export transactions may go through intermediaries (such as trade firms) that are not covered by Goskomstat data. The total reported overdue trade credit with the CIS in the Goskomstat aggregate data accounts for around 1 per- cent of total trade credit in arrears. From the World Bank survey (see ta- bles 4-2 and 4-3), which covers industry only (and very possibly undersamples fuel and energy firms with very large CIS receivables), we indeed learn that overdue receivables from CIS trade should instead ac- count for perhaps 5 percent of total overdue receivables. Time Trends Goskomstat data show that since late 1993 the volume of overdue pay- ables to the budget, overdue bank credit, and wage arrears have in- creased substantially, while trade credit in arrears (overdue payables to suppliers and overdue receivables from customers) fluctuated rather than increased (table 4-4 and figure 4-2). Figure 4-2 presents monthly data on overdue trade credit of Russian industrial enterprises, measured in months (overdue receivables as a per- centage of monthly sales). The pattern over time is the same as for total trade credit; indeed, the variation in total trade credit is driven mostly by the overdue portion.4 Overdue trade credit increased in early 1992 (the Table 4-3a. Balance Sheet of Russian Industrial Sectors, 1 January 1995, General Items and Selected Assets General items Selected assets (billion rubles) Average Cash Delivered Employment monthly wagea Receivables Overdue and production (thousand (thousand Total Overdue from receivables bank 7 January 7995 (billion rubles) workers) rubles) receivables receivables customers from customers deposits Total industry 55,728 14,497 379 106,288 57,735 85,855 48,921 7,086 Electric power 8,775 584 694 19,495 12,837 16,985 11,044 663 Total f'uel 10,895 966 876 27,231 17,481 21,807 14,682 1,882 Petroleum extraction 5,486 240 815 12,714 8,194 10,389 6,728 891 Petroleum refining 2,421 118 766 6,252 3,549 4,694 3,038 641 Gas 1,274 58 2,485 4,283 3,253 3,497 2,845 101 Coal 1,453 518 766 3,628 2,343 2,917 1,932 150 Ferrous metallurgy 4,812 772 376 10,748 5,796 8,697 4,908 417 Nonferrous metallurgy 3,763 543 688 5,189 2,546 3,674 1,602 443 Chemical and petrochemical 4,146 1,005 317 8,527 4,276 6,606 3,665 627 Machine-building 9,962 5,962 279 20,591 8,854 16,625 7,786 1,344 Forest, woodworking, 2,199 1,198 312 2,884 1,268 2,151 1,079 291 and paper Construction materials 1,588 678 337 2,223 1,149 1,892 1,061 166 Light industry 1,407 1,185 188 1,883 779 1,487 663 163 Food industry 5,984 1,173 419 4,695 1,713 3,585 1,470 718 Other industries 2,197 431 2,822 1,036 2,346 961 372 Agriculture 3,716 6,659 195 4,059 1,767 3,183 1,488 457 Transport 13,417 3,607 527 23,465 12,150 19,253 10,904 1,717 Construction 7,909 5,549 483 16,646 8,744 14,660 7,968 959 TOTAL 80,770 30,312 354 150,458 80,396 122,951 69,281 10,219 a. December 1994. Soturce: Goskomstat. Table 4-3b. Balance Sheet of Russian Industrial Sectors, 1 January 1995, Selected Liabilities (billion rubles) Ozverdue Payables to Ovierdue Overdue Payable payablesfrom the budget payables Total Ozverdue I January 1995 Total payables payables from suppliers suppliers (taxes dlue) to the budget Wage arrears bank credit bank credit Total industry 124,109 66,596 73,194 41,451 23,016 14,330 2,170 28,835 3,485 Electric power 18,259 10,365 14,236 8,341 1,681 923 126 1,415 51 Total fuel 34,073 21,984 17,402 11,361 10,292 7,657 663 1,673 70 Petroleum extraction 16,796 11,027 7,056 4,336 7,019 5,150 353 850 37 Petroleum refining 7,017 3,715 4,382 2,942 6,009 325 1 436 0 Gas 4,500 3,596 2,792 2,211 1,349 1,230 41 104 7 Coal 5,335 3,448 2,824 1,691 1,297 945 264 230 26 Ferrous metallurgy 11,183 6,267 7,073 3,863 957 494 71 2,061 15 Nonferrous metallurgy 7,604 3,445 4,147 1,983 1,165 641 128 2,248 302 Chemical and petro- chemical 8,888 5,134 6,812 4,380 581 174 44 2,022 201 Machine-building 26,2911 1,842 12,611 6,479 5,024 2,943 703 12,263 1,421 Forest, woodworking, and paper 4,146 2,261 2,206 1,267 917 573 169 1,275 366 Construction materials 2,917 1,347 1,945 984 427 201 73 522 61 Light industry 2,141 941 1,266 585 431 214 49 1,384 325 Food industry 5,305 1,865 2,941 1,234 1,252 435 105 3,153 568 Other industries 3,302 1,145 2,555 974 289 75 39 819 105 Agriculture 8,259 4,148 5,104 2,717 1,205 603 1,301 7,534 1,480 Transport 25,354 11,616 16,965 8,088 3,704 2,180 n.a. 1,188 82 Construction 16,291 7,990 9,164 4,524 4,081 2,215 729 1,996 298 TOTAL 174,013 90,350 104,427 56,780 32,006 19,328 39,553 5,345 n.a. Not available. Source: Goskomstat. 102 Financial Aspects of Enzterprise Restructuirintg Table 4-4. Arrears in Russian Industry since 1992 Percenitage of total liqutid assets or current liabilities Assets in arrears Liabilities in arrears From To To Total cius- Total suippli- the Bank Date receivables ton ers payables ers butdget debt Wages 1 October 1993 27 24 24 18 3 1 1 1 January 1994 32 28 30 21 5 1 1 1 April 1994 34 31 34 23 6 2 2 1 July 1994 35 31 38 25 8 2 2 1 October 1994 37 32 41 27 8 2 2 1 January 1995 37 31 43 26 9 2 1 1 April 1995 33 28 41 22 9 2 1 1 July 1995 32 27 40 20 10 2 1 Percenitage of asset or liability overdue 1 April 1992 37 n.a. 1 July 1992 58 1 October 1992 62 1 January 1993 46 42 4a 1 April 1993 40 38 6a 1 July 1993 42 36 12 a 1 October 1993 38 42 29 37 22 7 l1a 1 January 1994 43 45 36 41 40 8 13a 1 April 1994 46 50 40 45 44 12 41a 1 July 1994 49 54 46 49 54 12 55a I October 1994 53 57 51 56 58 14 54a I January 1995 54 57 54 57 62 12 36a 1 April 1995 50 53 51 51 12 45a 1 July 1995 48 52 49 47 14 40a n.a. Not available. Note: Liquid assets = total receivables + financial investments + cash. Current liabilities = total payables + bank debt + borrowings ("zaimy"). a. In percentage of average monthly wage. Souirce: Goskomstat. "arrears crisis"), fell in late 1992 (CBR mutual debt-clearing), was roughly flat from late 1992 to late 1993, increased again in early 1994, peaked at about 1.4 months in mid-1994, and fell again in late 1994 to about 1 month. The pattern of overdue trade credit measured in percentage over- due is similar. The increase in the proportion of overdue payables was more rapid for tax payables (reaching 62 percent overdue for total industry on 1 "Arrears" in the Russian Enzterprise Sector 103 Figure 4-2. Trade Credit in Arrears and Inflation in Russia Percentage of industrial output Consumer price index monthly inflation (%) 160 Consumer price index 140 one-month lag . 10 (inverted right-hand scale) 120 1 100 - . 20 80 ' 25 60 f30 Overdue commercial receivables in "Y of 40 industrial output 35 (left-hand scale) 20 I , I , I I I , , , E I I ] I I I ., o March Sept. March Sept. March Sept. March 1992 1992 1993 1993 1994 1994 1995 January 1995) than for any other category of liabilities. The share of ar- rears to the budget in total liabilities of industrial enterprises increased from about 3 percent in September 1993 to about 9 percent at the end of 1994. Overdue bank credit also grew substantially: according to Goskom- stat, the percentage of bank credit overdue grew from about 6 percent in late 1993 to about 12-13 percent in late 1994. CBR data on overdue bank credit show a similar trend at a higher level, resembling that found in the World Bank survey, probably because the CBR data include interest ar- rears. At the end of 1994, according to the CBR, overdue bank credit amounted to about one-third of the total credit stock (see chapter 5 for further discussion). International Comparisons Table 4-5 presents trade credit extended and trade credit received for the entire enterprise sector as a percentage of annualized GDP in selected Western countries, in three leading transition countries (Poland, Hun- gary, and the Czech Republic), and in Russia.5 It has been noted (Begg and Portes 1993 is an early example) that trade credit is a normal featLre Table 4-5. Trade Credit and Overdue Trade Credit in Western and Transition Economies Total trade credit Overdue trade credit Payment period Percentage of Overdue period Percentage of (months) annualized GDP Percent overdue (months) annualized GDP Country Receivables Payables Receivables Payables Receivables Payables Western countries Austria 2.1 Canada 16 14 Denmark 1.6 40 0.7 Finland 1.8 20 23 45 0.8 9 France 3.5 38 35 44 1.6 17 Germany 1.6 38 0.6 Ireland 2.0 50 1.0 Italy 3.0 33 1.0 Japan 59 45 Netherlands 1.7 42 0.7 Norway 1.6 38 0.6 Sweden 1.6 21 20 38 0.6 8 Switzerland 2.0 50 1.0 United Kingdom 2.6 20 19 62 1.6 12 United States 17 14 Transition countries Czech Republic 1994 2.5 50 37 0.9 18 Hungary 1988 1.4 1.2 37 28 1989 1.3 1.2 35 27 1990 1.5 1.3 36 29 1991 1.7 1.5 35 30 47 0.8 17 1992 27 21 Poland 1988 1.4 1.3 30 21 1989 1.3 1.3 24 18 1990 1.2 1.2 20 14 1991 1.5 1.7 22 19 1992 1.3 1.6 19 17 1993 1.4 1.7 19 16 51 40 0.7 0.7 10 6 Russia 1990 0.6 0.5 10 5 1991 0.6 0.7 12 7 1992 2.5 22 46 1.1 10 1993 1.6 15 44 39 0.7 7 1994 1.4 2.1 17 14 56 54 0.8 1.1 9 8 Note: Receivables: trade credit extended, commercial receivables, = receivables from customers. Payables: trade credit received, = commercial payables, = payables to suppliers. Payment period: directly measured in survey data; estimated from receivables/sales ratio (receivables), payables/purchases ratio (payables) for balance sheet data. Percent overdue: directly calculated for balance sheet data; calculated from ratio of overdue trade credit months/total trade credit months for survey data. In percentage of GDP: measured as a percentage of annualized end-period (last qutarter or month) GDP for transi- tion countries; annual GDP used for Westem countries. Figures for overdue trade credit as percentage of GDP are estimated from total trade credit as a percentage of GDP and percentage of trade credit overdue, except for the Czech Reputblic, Hungary, and Russia, where the volume of overdue trade credit is directly observed. Western figures are for 1985-90. Austria: balance sheet estimates; industry only. Canada: balanice sheet estimates. Denmark: survey estimates. Firdand: total trade credit in percentage of GDP, balance sheet estimates; percent overdue, survey estimates. France: total trade credit in percent of GDP, balance sheet estimates; percent overdue, survey estimates. Germany: survey estimates. Ireland: survey estimates. Italy: survey estimates. Japan: balance sheet estimates. Netherlands: survey estimates. Norway: survey estimates. Swe- den: total trade credit in percentage of GDP, balance sheet estimates; percentage overdue, survey estimates. Switzerland: survey estimates. United Kingdom: total trade credit in percentage of GDP, balanice sheet estimates; percentage overdue, survey estimates. United States: balance sheet estimates. Poland: figures for percentage overdue from survey of 200 Polish manufacturing firms, Belka, Estrin, Schaffer, and Singh (1995). Russia: percentage overdue end-92 is for industry and construction only; overdue in percentage of GDP derives from this figure, applied to estimate of total trade credit. Percentage overdue end-93 and end-94 is for industry, agriculture, transport, and con- struction only. Overdue in percentage of GDP derives from this figure applied to estimate of total trade credit. End-94 figure for total trade credit received is an estimate cal- culated by scaling up the reported figure for industry, agricultutre, transport, and construction. Source: Data derives either from balanice sheet figures or survey questions. Intrum Justitia, in Chittenden and others (1993) (Western survey data); OECD; UK CSO; Fan and Schaffer (1994); Bonin and Schaffer (1995); Czech Statistical Office, "Economic Results of Nonfinancial Firms, 4th quarter, 1994"; and authors' estimates. 106 Financial Aspects of Enterprise Restriucturing of a market economy and could be expected to emerge in transition economies, and this is indeed what we find. The comparisons in table 4-5 show that pretransition levels of trade credit in Russia were very low by both transition-country and Western standards. More important, levels of trade credit in Russia, the Czech Republic, Hungary, and Poland in the transition period have been unexceptional by Western standards. Indeed, total trade credit in Russia since 1992, at the equivalent of about 15 percent of GDP, has been on the lower end of the scale seen in Western countries. Measuring total trade credit by an average payment period leads to the same conclusions: table 4-5 shows that average payment periods in Russia prior to the transition were very short by Western European standards (less than one month, compared with 1.5-3 months), and dur- ing the transition have not been unusual by Western standards (on the or- der of 2-3 months). That overdue trade credit is also a normal feature of a market econ- omy is less well known. Table 4-5 presents figures for overdue trade credit-measured in months (average overdue period), as a percentage of total trade credit, and as a percentage of GDP-for a number of Western countries, Poland, Hungary, the Czech Republic, and Russia. These fig- ures show that overdue trade credit is endemic in developed market economies as well as in transition countries. The percentage of trade credit overdue in the eleven Western countries for which we have data ranges between 30 percent and 60 percent-the same range observed in Russia. Measured in months, overdue trade credit in these countries ranges between 0.6 and 1.6 months; again, about what we observe in Rus- sia. As a percentage of GDP, overdue trade credit in Russia has been rela- tively low compared with that found in Western countries: between 7 and 10 percent of GDP in Russia, compared with 10-18 percent in the Western countries. In short, the volumes of both total trade credit and trade credit in arrears in Russia, however measured, are roughly average by both Western and Eastern European standards. While having a high level of arrears is quite common for enterprises in Western countries, one needs to take into account the relatively poor worth of Russian liabilities. One would expect a higher proportion of bad debts in Russia than in market economies; that is, firms in financial dis- tress account for a larger fraction of trade credit arrears than in Western countries. We shall see below that using one measure of "financial dis- "Arrears" in the Russian Enterprise Sector 107 tress" (the firm is a chronic loss-maker), perhaps 10-20 percent of total payables to suppliers, and of arrears to suppliers, is in firms that are fi- nancially distressed. This would suggest that if we were to exclude these items on the grounds that they are "bad debts" and unlikely to be col- lected, the volume of trade credit in Russia would be low-to-average by international standards. In contrast to trade credit, tax arrears numbers are much higher for Russia than in Western economies, but they are still similar to levels ob- served in Central and Eastern European (CEE) economies. Using Gosk- omstat and other data, we estimate that at the end of 1993 total tax arrears of the enterprise sector to all government (budget plus ex- trabudgetary funds) amounted to about 2 percent of GDP.6 By the end of 1994, this figure had risen to about 4-5 percent of GDP. By comparison, it has been estimated that at the end of 1993 the stock of tax arrears (includ- ing late penalties and interest) in Hungary and Poland amounted to about 10 percent of GDP, and in the Czech Republic and Slovakia, about 5 percent of GDP (Schaffer 1995). Using the reported Goskomstat figures, we estimate the real flow of tax arrears to all government to have been about 2 percent of GDP in 1994. Allowing for write-offs and rescheduling would increase this figure, but probably not much. Write-offs of tax arrears were unlikely to be very common. Reschedulings have the effect of reclassifying taxes from over- due to not overdue; that the volume of nonoverdue tax payables was, if anything, falling suggests that reschedulings were not taking place on a large scale. Reschedulings became a much bigger problem in 1995 when the government, in response to lobbying by firms with wage arrears, in- troduced a scheme that allowed firms to defer payment of taxes. Our estimate of the flow of tax arrears in Russia in 1994-about 2 per- cent of GDP annually-is of about the same magnitude as the flow of tax arrears observed in the Czech Republic, Hungary, and Poland (see Schaf- fer 1995). The sitLation in Western countries is quite different. In the West, we expect stocks of overdue taxes to be roughly stable over time, as new overdue taxes are added (inflow) and existing overdue taxes are either paid or written off as uncollectible (outflow). We can use the vol- ume of write-offs in Western countries to estimate the flow of uncollect- ible tax arrears. In New Zealand, for instance, the volume of write-offs of uncollected and uncollectible taxes in recent years amounted to less than one-half of one percent of GDP.7 108 Financial Aspects of Enterprise Restrictutring We have shown that the level of trade credit in arrears in Russia; which at first glance seems high, has stayed within the range observed in Western countries. Both tax arrears and wage arrears are much more worrisome for the economy, because both can be seen as ways for firms to capture subsidies from the government, as we shall argue below. Macroeconomic Policy, Liquidity, and Financial Distress Having considered first the aggregate evidence, we now take up our dis- tinction between late payments and bad debts (and strategic arrears as well). We begin with three examples, each relying on three different data sources: macroeconomic policy in Russia since 1992, using aggregate time-series data; the correlation of liquidity and arrears, using sectoral time-series panel data; and the concentration of arrears in financially dis- tressed firms, using the World Bank enterprise survey. Macroeconiomic Policy and Trade Credit Arrears Figure 4-2 shows a very close inverse relationship between consumer price index (CPI) intlation, lagged one month, and the ratio of overdue com- mercial receivables in industry to monthly industrial sales (that is, receiv- ables in arrears, measured in months of payment delay) between early 1992 and mid-1995. When inflation is high or rising in Russia, trade credit ar- rears are low or falLing; the reverse is true when inflation is low or falling. The developments in the first half of 1992-the so-called arrears cri- sis-can probably be interpreted in part as an example of strategic or col- lusive arrears (for a detailed theoretical analysis of collusive arrears, see Perotti 1994). The Gaidar government had launched its first stabilization program at the start of the year, and during the first half of 1992 it strug- gled to keep to a strict monetary and fiscal policy, despite the pressures brought by the industrial lobby (that is, enterprise managers), among other disruptions (see Gaidar 1995). In the first half of 1993, unpaid pay- ments of enterprises grew dramatically.8 This growth of arrears may be attributed partly to "payments gridlock" (Ickes and Ryterman 1992)-a lack of liquidity combined with an inflexible payments system9-but an- ecdotal evidence suggests that firms also deliberately withheld payments to suppliers in expectation of the failure of the stabilization program. These expectations were fulfilled in mid-1992, when the stock of unpaid "Arrears" in [lie Russiani Enzterprise Sector 109 payments was frozen and then cleared (with the help of an injection of credit) by the CBR, the rules governing unpaid payments were changed, and monetary and fiscal policy was loosened.10 There has been no repetition of the arrears crisis, however. Moreover, the level of trade credit arrears has fluctuated within a fairly na:rrow range (in months of sales, between 0.5 and 1.5 months). Thus, over a pe- riod of 40+ months, most arrears are getting paid in the end. For the 1992-95 period, we offer the interpretation that falling inflation is associ- ated with increasing arrears because both are being driven by tightening liquidity conditions; hence the very close inverse correlation between in- flation and arrears in figure 4-2. A tightening of the money supply at the macroeconomic level (reduction in the volume of money emission, tight- ening of credit, rise in the refinancing interest rate, and the like) brings in- flation under control. Macroeconomic studies on inflation in Russia show that a tightening of CBR credit leads to a corresponding decrease in the rate of inflation four months later as measured by the CPI (for example, see Koen and Marrese 1995). We would expect that when the cost of ob- taining bank credit increases, borrowers (purchasers) will be motivated to obtain additional liquidity elsewhere, particularly by increasing their use of trade credit received by delaying payment to suppliers. Disinflal:ion also depresses demand, and again we would expect firms to respond to decreases in demand and the consequent losses in liquidity by delaying payment to suppliers. As we shall see below, this is consistent with our correlation analysis of the World Bank survey: arrears are correlated wTith measures of declines in demand. Arrears and Liquidity We now examine the relationship between arrears and liquidity using Goskomstat sectoral panel data. We ask whether changes in levels of ar- rears over time are correlated with changes in sectoral liquidity over tirne. We use as our measure of liquidity the ratio of money holdings (cash + bank deposits) to total liabilities. We have available monthly data on ar- rears and liquidity for twenty-odd sectors for the period 1 October 1993 to 1 January 1995, except for the data on wage arrears, which start in April 1994. We are interested in whether arrears of different sorts are correlated with liquidity. We investigate this by exploiting the panel nature of the 110 Financial Aspects of Enterprise Restnrcturing sectoral dataset. We calculate partial correlation coefficients between the variables of interest, controlling for industrial sectors (that is, including 1/0 industry dummy variables)." The effect of this procedure is to allow the "benchmark" levels of arrears (the intercept in the regression) to be different for individual sectors (that is, we allow for sector-specific "fixed effects"), and therefore we capture correlation based on co-movements of the variables of interest (arrears and liquidity) over time. Controlling for sectoral variation across industries is particularly important here, because a sector may tend to be highly liquid for what are basically structural rea- sons-petroleum extraction is the main example (see table 4-3). By hold- ing the industrial sector constant (that is, allowing for sector-specific intercepts to capture structural differences in liquidity across sectors), we look only at the relationship between the change in money holdings and the change in arrears. The partial correlation coefficients are normalized such that 1 indicates perfect correlation; -1, perfect negative correlation; and 0, no correlation. We present two sets of correlations of arrears and liquidity: where ar- rears are measured as the percentage overdue and where arrears are a percentage of total liabilities. The results are presented in table 4-6, and are quite clear. Changes in money holdings are negatively correlated with changes in arrears; when arrears are increasing, money holdings are decreasing. The correlations are highly significant-in excess of the 1 per- cent level-in all cases. We interpret these findings in terms of "late pay- ments": arrears can be seen as a "cushion" in periods of tight liquidity. Concentration of Arrears in Financially Distressed Firms We first need a satisfactory indicator of financial distress. We use the firm's response to the question of whether the firm is "usually a profit- maker." Most of the firms in the survey (86 percent of the total sample) reported that they were usually profit-makers, about the same figure that Goskomstat recorded for the entire population of industrial enterprises in 1993. This does not indicate, however, that most firms are financially se- cure. It is difficult to make (nominal) losses in a high-inflation environ- ment because of the upward bias caused by paper capital gains, the result of historical cost accounting. In periods of high inflation, historical cost accounting causes paper capital gains on inputs-costs are calculated at purchase prices, and between the time the inputs are purchased and the 'Arrears" in the Ruissian Enterprise Sector 111 Table 4-6. Partial Correlations of Assets and Liabilities in Arrears with Money Holdings Arrears measutred as Correlation Arrears measutred as Correlation percentage overdue With liqtuidity percenitage of total liabilities with liquidity Total debtors -0.58 Total debtors -0 42 (receivables) (0.00) (receivables) (0.00) [4311 [4311 Commercial receivables -0.58 Commercial receivables -0.37 (0.00) (0.(0) [4311 [4'-1] Total creditors -0.56 Total creditors -0.57 (payables) (0.00) (payables) (0.C00) [431] [431] Commercial payables -0.47 Comrnmercial payables -0.45 (0.00) (OCO) [431] [431] Tax arrears -0.56 Tax arrears -.58 (0.00) (0.00) [431] [431] Bank credit -0.24 Bank credit -0.17 (0.00) (0.03) [424] [424] Nonbank loans -0.09 Nonbank loans -0.12 (0.07) (0.02) [409] [409] Overdue wages -0.66 Overdue wages -0.30 (as percentage of (0.00) (0.0() monthly wage bill) [621 [24:1] Note: Period included in the table extends from 1 October 1993 to 1 January 1995 (except wage arrears, which begins on 1 April 1994). Industrial sector held constant; approximnately twenty sectors (regressions include twenty-odd industry dummies; see text). Results are: partial correlation; P-values in parentheses (0.05 = significant at 5 percent level); number of observations in squared brackets. Liquidity = (cash + bank deposits)!/total liabilities. Source: Goskomstat monthly sectoral data. time the output utilizing these inputs is sold, the price level has in- creased. This can cause serious cash flow problems for firms because pa- per capital gains are part of the profit tax base,12 a quite common problem in economies experiencing inflation. When asked what have been the most important problems faced by the firm since January 1992, 58 percent of firms answered "poor financial situation of the firm"; this was ex- 112 Finatcial Aspects of Entterprise Restructiliring ceeded only by "increase in price of inputs" (83 percent of firms)'3 and "high interest rates" (76 percent of firms). "Lack of domestic demand," for example, was marked as very important by fewer firms (53 percent). To be "usually a loss-maker" in an inflationary environment such as Rus- sia's suggests quite severe financial difficulties, and it is therefore a rea- sonable indicator of financial distress. We would like to kniow how extensively arrears are concentrated in fi- nancially distressed firms. The practical problem we face is that the World Bank survey is very heterogeneous in size of firm, with reported employment, for example, ranging from 10 persons to 100,000 persons. Calculating the concentration of arrears in the relatively small number of financially distressed firms can be indicative, but it is of limited reliability because the results are not robust to outliers; including or excluding a sin- gle (large) firm can change the results significantly. We therefore adopt the following, more robust, but admittedly ad hoc, approach. The idea is to address the size-heterogeneity problem by normalizing arrears through some size variable. Having done this for each firm, we then calculate the mean level of (normalized) arrears for the financially distressed sample and for the nondistressed sample. From these two means, and from the numbers of distressed and nondistressed firms in the sample, we can calculate the implied proportions of arrears that dis- tressed and nondistressed firms would represent were they all of uniform size. A statistical test of the difference of the two means is our test of whether the concentration is statistically significant. For example, we could use sales to normalize overdue commercial payables. If we find that overdue commercial payables for each unit of sales are twice as high in distressed firms as in nondistressed firms, and 10 percent of firms are financially distressed, then the predicted concentration of arrears in dis- tressed firms would be [2*10/(2*10 + 90)) = 18 percent. Some further remarks are needed on how this is implemented. For ro- bustness we use two different size variables, sales and employment, and do everything twice. We consider only firms with nonzero arrears (in contrast to the analysis in the next section). The distributions of the ar- rears/sales and arrears/employment ratios are very skewed, with many small values and a smaller number of very large values. Inspection showed that the distribution of log(arrears/sales) and log(arrears/em- ployment) were roughlv normally distributed, and so means were calcu- lated using the log values. We calculated medians as well as means, and "Arrears" in the Russiani Enterprise Sector 113 repeated the estimates of concentration using medians in place of means as a robustness check. For each category of arrears we therefore have four sets of results: normalize by sales and use means; normalize by sales and use medians; normalize by employment and use means; and normalize by employment and use medians. The differences between the means in the distressed and nondistressed groups are tested statistically using a t- test; the medians are tested using the Wilcoxon rank-sum test. The im- plied concentrations of arrears we calculate are essentially what we would expect to observe in a population with the observed statistical fea- tures of the two subsamples (means, standard deviations, and the pcpu- lation divided into distressed and nondistressed in the same proportions as we observe in the survey data). The results are presented in table 4-7. We report not the tested log means, but the calculated ratios these imply.14 We begin with the finding that total commercial payables are not significantly concentrated in finan- cially distressed firms. Testing among firms with nonzero commercial payables shows that the mean commercial payables/sales ratio is not sig- nificantly higher in financially distressed firms, regardless of whether means or medians are tested. That is, the implied concentration of com- mercial payables in distressed firms (14-19 percent) is not significantly different from the proportion of distressed firms in the sample (12 per- cent). The same holds true if we normalize by employment instead of sales; indeed, if anything, this set of results suggests distressed firms hold fewer commercial payables than average. The results for overdue commercial payables provide some evidence of weak concentration in distressed firms. Of the four sets of results, one shows the level of overdue commercial payables to be significantly higher in distressed firms, at the 1 percent level, with an implied concen- tration of 21 percent of overdue commercial payables in distressed firms (compared with the 10 percent distressed firms in the sample). The re- maining three results suggest no statistically significant concentration of arrears to suppliers in distressed firms. These findings for commercial payables and overdue commercial pay- ables should be interpreted with some care because of problems with in- flation effects. As noted earlier, debtor firms may not be calculating the penalty interest they owe to suppliers, even when suppliers are chargin,g such interest. Existing stock problems would therefore be steadily re- duced by inflation. Note, however, that these low figures also suggest no Table 4-7. Concentration of Liabilities and Arrears Nondistressed Distressed Significance of firms firms difference between Predicted Percentage Mean Percentage Mean nondistressedl concentration in Liability n of sample Median of sample Median distressedfirms distressedfirms Total commercial payables CP/sales (percent) 217 87.8 12.2 Mean 11.9 14.1 14.1 Median 13.4 23.0 19.3 CP/L (million rubles) 270 88.1 11.9 Mean 0.643 0.357 6.9 Median 0.716 0.467 8.1 Overdue commercial payables ODCP/sales (percent) 178 89.9 10.1 Mean 9.0 16.7 17.3 Median 10.4 25.0 21.2 ODCP/L (million rubles) 176 90.3 9.7 Mean 0.442 0.581 12.3 Median 0.550 0.748 12.7 Arrears to banks ODB3C/sales in percent 55 78.2 21.8 Mean 4.42 5.28 25.0 Median 3.96 7.64 35.0 ODBC/L (million rubles) 54 77.8 22.2 Mean 0.169 0.214 26.5 Median 0.249 0.116 11.8 Tax arrears ODIX/sales in percent 111 86.5 13.5 Mean 3.11 15.32 43.5 Median 2.95 13.59 41.9 ODTX/L (million rubles) 109 87.2 12.8 Mean 0.134 0.661 42.1 Median 0.172 0.582 33.3 Wage arrears ODWG/sales in percent 93 86.0 14.0 Mean 3.30 3.72 15.5 Median 3.53 3.75 14.7 ODWG/L (million rubles) 91 86.8 13.2 Mean 0.107 0.137 16.2 Median 0.118 0.141 15.4 Note: Significance levels: = 5 percent level; 1 percent level; 0. 1 percent level; no star not significant at 5 percent level. Source: World Bank survey. 116 Financial Aspects of Enterprise Restructutring particular ongoing flow problem (since large recent and current flows would create large current stocks). Neither wage arrears nor arrears to banks are significantly concen- trated in financially distressed firms. The results of the statistical tests are not significant, and the implied concentrations differ little from the actual proportions of distressed firms in the samples. The results for arrears to banks should be treated with some caution, however, because of the small numbers involved. Although the predicted concentration of arrears to banks is fairly high, the sample differences that served as the basis for these calculations are not statistically significant. The only kind of arrears that are clearly concentrated in financially distressed firms are tax arrears. The statistical tests are uniformly highly significant: all are in excess of the 0.1 percent significance level. The im- plied degrees of concentration are also striking. Distressed firms make up 13 percent of the sample, but we estimate they would account for be- tween one-third and close to one-half of all tax arrears. We note, however, that the degree of concentration of tax arrears in distressed firms is not as high in Hungary and Poland, where it has been estimated that two-thirds to three-quarters of tax arrears are found in these firms (see Schaffer 1995). This would suggest that problems with late payment of taxes and poor tax discipline are more widespread in Russia than in these two lead- ing transition countries. We will return to these findings later in the chapter. Here we summa- rize only insofar as to note that the evidence suggests only small portions of total commercial payables, overdue commercial payables, wage ar- rears, and arrears to banks would appear to be concentrated in financially distressed firms. The bulk of these arrears would seem to be late pay- ments rather than bad debts. The case of tax arrears is different, and we will return to this later. Microevidence on Arrears We turn now to an examination of the correlates of arrears at the firm level, based on the World Bank survey. We will interpret these correla- tions in terms of arrears as late payments and arrears as bad debts. The methodology we employ is to calculate both simple correlations between pairs of variables and correlations controlling for firm characteristics: the firm's size (measured by log employment), the kind of location (na- "Arrears" in the Ruissian Enterprise Sector 117 tional urban center, oblast capital, other city, or rural), and industry (fif- teen sectors). We measure arrears in two ways. First, we normalize by sales; this gives us a measure of the volume of arrears relative to the size of the firm's turnover. Second, we calculate a 1/0 variable based on whether the firm has a "significant" amount of a category of arrears or not; "signifi- cant" means in excess of 2 percent of annualized sales. Each pair comparison therefore generates four correlations: (1) A simple correlation between the volume of a kind of arrears (nor- malized by sales) and another variable (2) A simple correlation between the presence (1/0) of a significant volume (> 2 percent of sales) of a kind of arrears and another variable (3) A correlation between the volume of a kind of arrears (normalized by sales) and another variable, holding constant a set of firm charac- teristics (size, location, industry) (4) A correlation between the presence (1/0) of a significant volume (> 2 percent of sales) of a kind of arrears and another variable, holding constant a set of firm characteristics (size, location, industry). The correlations are presented in tabular form in tables 4-8-4-11. A ++ indicates a positive correlation that is significant at the 1 percent level; a + means significant at the 5 percent level; 0 means not significant; - means a negative correlation at the 5 percent level; and - - means significant at the 1 percent level. A fuller explanation of abbreviations and methodology accompanies the tables. Our first finding is that arrears are, not surprisingly, highly correlated with arrears. Table 4-8 presents correlations between arrears, again both simple and partial correlations, and again both for the volume of arrears and the presence of arrears. Nearly all arrears are correlated with nearly all other kinds of arrears, however measured. This finding suggests that different kinds of arrears are (often) codetermined by underlying factors. In what follows we try to track down some of these factors. We begin with some negative results. First, ownership shows rela- tively little correlation with any sort of arrears, whether or not we control for other firm characteristics (table 4-9). The main exception is that de noVO (newly established) firms tend to hold fewer payables in arrears in the simple correlations. De novo firms are also small, however, and thus when 118 Finanzcial Aspects of Enterprise Restructutring Table 4-8. Correlations of Arrears with Arrears ODCP ODBC ODTX ODWG ODBC 0 0 ++ ODTX -f-+ ++ ~ ~ + ODWG ++ ++ + ++ ++ ++ + ODWG ++ ++ ++ ++ ++ + ODRC ++ 0 ++ ++ ++ + ++ ++ ++ 0 ++ ++ ++ + + ++ n.a. Not available (that is, estimation failed). Note: ODCP: overdue commercial payables. ODBC: overdue bank credit. ODIX: overdue tax, Social Security, and so forth, payables. ODWG: overdue wages. ODRC: overdue receiv- ables. ++ = positively correlated, significant at 1 percent level. + positively correlated, sig- nificant at 5 percent level. 0 = not correlated at 5 percent level. - Negatively correlated, significant at 5 percent level. - - Negatively correlated, significant at 1 percent level. For each pair of variables, four correlations are reported: Line 1: Simple correlation with volume of arrears. Correlation coefficient (both volume measures) or t-test (if comparison variable a 1/0 variable). Arrears variable: arrears normalized by sales. Line 2: Simple correlation with presence of significant arrears. t-test by presence of arrears (if volume measure), chi (if 1/0 variable). Arrears variable: presence of significant arrears ( 2 percent of sales), 1/0 variable. Line 3: Correlation with volume of arrears, controlling for a set of firm characteristics. Partial correlation coefficient of comparison variable when also controlling for size (log employment), location (four city dummies), and industry (fifteen dummies). Arrears variable: arrears normalized by sales. Line 4: Correlation with presence of significant arrears, controlling for a set of firm characteristics. Significance of comparison variable in a logit regression also including size (log employ- ment), location (four city dummies), and industry (fifteen dummies). Arrears variable: presence of significant arrears (2 percent of sales), 1/0 variable. Source: World Bank survey. "Arrears" in the Ruissiani Enterprise Sector 119 Table 4-9. Simple and Partial Correlations between Arrears and Firm Characteristics ODCP ODBC ODTX ODWG 0DRC Size-log employment ++ 0 0 0 ++ ++ 0 ++ 0 ++ ++ + 0 0 C ++ + ++ 0 +F Military-industrial complex ++ 0 0 0 ++ firm (1/0) 0 0 0 ++ ++ o o 0 0 0 O O 0 + + Monopolist (1/0) ++ 0 0 0 0 o o 0 0 0 o 0 0 0 0 o o 0 0 0 State-owned (1/0) 0 0 0 0 - o o 0 0 0 o 0 0 0 + o o 0 0 0 Privatized (1/0) 0 0 0 0 0 +4- 0 ++ 0 0 o 0 0 0 0 o 0 0 0 0 De nzovo, newly established -- 0 - - 0 private firm (I/0) --- -- o o 0 0 0 o o 0 0 0 Major city (I/O) -- -- - 0 (Moscow, St. Petersburg) -- - -- --- Oblast capital (I/O) 0 0 0 0 0 (Major cities excluded) 0 0 0 0 0 o 0 0 0 0 o o 0 0 0 City, not oblast capital (I1/0) 0 + + + 0 +4 ++ ++ + o ++ +4- + 0 Rural (I/O) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 + 0 0 0 0 NVote: See note to table 4-8. Source: World Bank survey. 120 Financial Aspects of Enterprise Restructutrinig Table 4-10. Arrears, Demand, and Response to Change in Demand ODCP ODBC ODTX ODWG ODRC Log change in output,a 1993-94 0 0 - -- 0 -- O -- -- 0 o 0 0 -- 0 -- 0 0 - 0 Log change in output,' 1993-94 - 0 0 - 0 (alternate measure) -- -- -- 0 0 o 0 0 0 o -- 0 0 0 Log capacity utilization, 1994 -- 0 0 -- 0 -- O O -- -- -- 0 0 -- 0 -- 0 0 - - Log change in capacity - - 0 0 - - 0 utilization, 1991-94 -- 0 0 -- -- 0 0 -- 0 - 0 0 -- 0 Share of sales going to FSU + 0 + ++ 0 market in 1990 + 0 0 0 0 o o + 0 0 o 0 0 0 0 Log change in employment, 0 0 0 - 0 1993-94 -- - -- -- 0 o o 0 0 0 -- 0 - - 0 Log change in employment, 0 0 0 0 0 1991-94 0 0 0 - 0 -- 0 0 0 0 - 0 0 0 0 Log change in wage, - 0 0 0 0 1993-94 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Log change in wage, -- 0 0 - 0 1990-94 0 - 0 - 0 - 0 0 0 - 0 0 0 0 0 Note: See note to table 4-8. a. Change in real output reported directly by firms. Alternate measure of change in real output derives from deflation of reported nominal output figures. Sou(rce: World Bank survey. "Arrears" in the Russian Enterprise Sector 121 Table 4-11. Arrears and Financial Indicators ODCP ODBC ODTX ODWG OORC "Financial distress" (1/0) + 0 ++ 0 0 (Usually a loss-maker) 0 + ++ 0 0 ++ + ++ 0 0 O ++ + 0 0 "Bank credit easy to obtain" (1/0) 0 0 -- 0 0 - - -- 0 0 - 0 - 0 0 o 0 -- 0 0 Log wage - - - --- 0 O -- O --0 - _ _ 0 - - O O O - -- Note: See note to table 4-8. Source: World Bank survey. we control for size in the partial correlations, this negative correlation dis- appears. Second, partial correlations between arrears and industry (not reported) show, rather surprisingly, very little systematic correlation be- tween arrears of any sort and the industrial sector in which the firm oper- ates. Third, membership in the military-industrial complex is clearly associated only with overdue receivables, even after controlling for size, industry, and location. This probably reflects payment delays by the gov- ernment in its military procurement. Fourth, and somewhat surprisingly, "being monopolist" (as described by the firm itself) is not clearly corre- lated with any sort of arrears. One might have suspected that monopo- lists, in a strong position with their customers, would have fewer overdue receivables, or that they would have larger overdue taxes, perhaps being in a strong position in relation to the government, but this does emerge from the data. Fifth, size (as measured by employment) is clearly corre- lated with arrears, especially overdue trade credit (both arrears to suppli- ers and arrears from customers). To check our hypothesis on financial stress, in table 4-10 we present correlations between arrears and a variety of variables meant to capture both the degree to which firms were hit by the loss of demand (or by their inability to meet demand)-change in output, capacity utilization, share of sales going to the markets of the former Soviet Union in 1990-and the 122 Financial Aspects of Eniterprise Restructuirinig extent of their response to this loss through shedding labor and restrain- ing wages. The correlations are fairly strong evidence that overdue pay- ables are associated with a fall in demand and (sometimes weakly) with the firm's response in its adjustment of employment and wages. What is somewhat surprising is that the correlation between overdue receivables and the demand and demand-response variables is much weaker than the correlations for overdue payables, even though overdue receivables are closely correlated with overdue payables, as we have seen. This is consistent with the view that overdue receivables generate overdue pay- ables, and declines in demand generate payments difficulties and hence overdue payables, but that losses in demand are not systematically asso- ciated with increases in overdue receivables because suppliers experienc- ing losses in demand do not keep shipping to customers that do not pay (more about this below). Additional evidence in favor of the "financial stress" interpretation is the very strong correlation observed with location (table 4-9). Firms in Moscow and St. Petersburg have fewer and less frequent arrears to all categories of creditors, oblast capitals (excluding Moscow and St. Peters- burg) are average, and cities that are not oblast capitals tend to have firms with significant arrears to all kinds of creditors. It is likely that there is a connection here with local demand and economic prosperity. Another piece of evidence that the bulk of overdue trade credit should be categorized as late payments rather than bad debts comes from the term structure of overdue trade credit. The World Bank survey data on the term structure of overdue payables are presented in table 4-12: the bulk of overdue liabilities are short term, although a significant portion of payables are overdue by more than three months. Aggregate Goskomstat data for overdue payables to suppliers show that in the year preceding the survey, the percentage overdue had been relatively stable (40-50 per- cent), while the total volume of overdue payables had increased. This suggests that the large amount of short-term overdue payables to suppli- ers in the surveyed firms is not primarily the result of the early-1994 in- crease in the total volume of overdue payables; the main reason is probably that firms consistently pay each other late. The low figure for very old (over one year) overdue payables to suppliers would indicate that most of these payables are eventually paid, although a lack of infla- tion indexing (firms failing to charge, or to record, late penalties), as noted, also contributes to this low figure. Interrepublican arrears (of Rus- "Arrears" in the Russian Einterprise Sector 123 Table 4-12. Term Structure of Arrears (percent) Term strctutre of overduie liabilities and receivab,es Overdute Overfue Not Overdute less 3 months- more than Itemi overdue than 3 monzths I year I yeatr Liabilities Payables to suppliers 42 30 23 4 Liabilities to banks 72 17 9 1 Payables to all government 54 31 13 1 Payables to employees 52 42 5 0 Receivables From domestic firms 26 27 32 5 From budgetary organizations 34 27 29 2 In state subsidies 33 22 14 5 From former Soviet Union trade 13 16 29 31 From non-former Soviet Union foreign trade 38 13 30 3 Source: World Bank survey. sian enterprises with the other former Soviet republics), as shown by the survey, are not large and are rather old. For the "typical" industrial firm, arrears from the CIS are more a stock than a flow problem, because most of the older receivables in arrears are probably uncollectible. Goskomstat aggregate data show, by contrast, a steady increase in the percentage of bank credit, and especially tax payables, overdue; it is here that we ob- serve flow problems. The picture is similar if we look at the proportion of firms in the sam- ple that have some overdue payables of a given age (table 4-13). This ap- proach has the advantage of being unaffected by the presence or absence of inflation indexing. Although most firms have some overdue payables to suppliers, only 14 percent have payables to suppliers that are overdlue by a year or more. The number of firms more than three months in ar- rears to banks is fairly small. Tax arrears are fairly common: over one- third of the sample is behind in taxes, and 14 percent of the sample have taxes overdue by more than three months. We note, however, that the 124 Financial Aspects of Enterprise Restructuzring Table 4-13. Frequency of Occurrence of Arrears (percentage of sample) With arrears at With arrears at least 3 months least 1 year Category With arrears overdue overdue Liabilities To suppliers 58 41 15 To banks 18 6 1 To all government (taxes) 36 14 3 To employees (wage arrears) 30 5 0 Receivables From domestic firms 73 53 25 From budgetary organizations 16 9 1 In government subsidies 8 3 1 From former Soviet Union trade 23 19 12 From non-former Soviet Union foreign trade 15 9 3 Source: World Bank surey. small number of firms with long-term arrears to banks and the tax authorities may in part reflect recent growth in these items (rather than the likelihood of eventual repayment), as will be suggested below. Fi- nally, wage arrears are fairly common-30 percent of firms have them- but they are very short term in nature. Only 5 percent of the sample had wage arrears overdue by more than three months, and none had wage ar- rears over a year old. This is to be expected, because the relevant time ho- rizon will be different for workers than for other creditors: beyond several months of delay in receiving wages, employees wvould probably protest or leave their jobs. We return now to our analysis of correlation. Table 4-11 presents cor- relations between arrears and some indicators of the financial status of the firm (including our financial distress dummy variable). We include, somewhat unconventionally, the wage level. If we think of these firms as insider-controlled, then labor will capture some (or all) of the firm's profit, and the wage will be a function of profit. It is more reasonable to argue that this wvill be captured with the correlations that control for firm characteristics-these include industry dummies, and hence control somewhat for wage variation by industry. As expected, difficulties in ob- "Arrears" in the Russian Enterprise Sector 125 taining bank credit and low wages are associated with arrears. Our finan- cial distress dummy is somewhat correlated with overdue commercial payables, more so with overdue bank credit, and highly correlated' with overdue taxes.15 Financial distress thus measured is not correlated at all with wage arrears. Nevertheless, we note from the table that wage a:rrears are strongly negatively correlated with the wage level., Low-wage firms are also wage-arrears firms, even when we control for variation by industry. Large-scale wage arrears distinguish Russia from the leading transi- tion countries of Central and Eastern Europe. In our view, the real issue here is not so much whether they are correlated with financial distress or not, but rather in the use made by firm managers of weak workers to lobby the authorities and extract subsidies. Wage arrears are in some sense an accounting fiction; in principle, managers always have the choice between paying a low wage (promptly) or essentially promising to pay higher wages with money the firm doesn't have (or the managers won't admit to having)-that is, to accumulate wage arrears. We suspect managers promise wages in excess of the cash they have available so as to turn workers' protests toward the government authorities in order tc ob- tain financial assistance. The case of the so-called "30:70 rule" provides a good illustration of this. In early 1994, following an increase in wage arrears (see table 4-4) and lobbying by firms, the government introduced a scheme airned at providing relief to firms with wage arrears: firms with wage and tax ar- rears could legally defer payment of part of their taxes and use the money to pay wages instead.16 The taxes deferred were exempt from interest and late penalties, which meant (given Russian inflation rates) a substantial real subsidy. The scheme at first was temporary and applied only to firms in selected industrial branches, but by the end of 1994 it had been made indefinite and extended to all firms. In 1995, "30:70 rule" tax deferrals were utilized by a large number of firms and accounted for most of the growth of tax arrears in that year. Use of the scheme apparently was not limited to firms in serious financial distress, and the incentives provided by the scheme were perverse. It seems that firms could obtain a tax defer- ral simply by generating wage arrears; that is, by choosing a wage thiat was sufficiently high to preclude payment in full to employees. It is also worth noting that the initial list of industrial branches approved for use of the scheme included several that were, if anything, relatively healthy, such as the oil extraction and gas branches. 126 Finanicial Aspects of Eniterprse Restricturing We now resume the line of investigation we began with our results on concentration of arrears-whether enterprises in financial distress behave differently regarding arrears. Managers were asked to identify the most important causes of their overdue payables: the answers are summarized in table 4-14 for the sample as a whole, and separately for financially dis- tressed (chronic loss-making) firms. When the answers of the financially distressed group differ in a statistically significant manner from the rest of tlhe sample, this is marked in the table. The most commonly noted reasons for overdue payables-cited as im- portant by all but about one-third of the sample-are the need to pay taxes first, the financial difficulties of the firm, and overdue receivables. The differences between the nondistressed and distressed groups are in- teresting. The chronic loss-makers are no more likely than others to cite overdue receivables as a reason to delay payment to their suppliers. Not surprisingly, they are considerably more likely to blame the financial dif- ficulties of the firm, trouble obtaining bank credit, and pressure to pay wages as the reasons for not paying their suppliers instead. Note that Table 4-14. Cited Causes of Overdue Payables (percent) Fintancially distressed CaUse Firms citing cause firms citi71g caulse Overdue receivables 32 29 Slow payment clearing 20 16 Bank credit expensive or hard to get 18 27 (+) Result of business strategy 2 2 Financial difficulties of the firm 33 53 (++) Must pay wages first 20 30 (f ) Must pay banks first 14 11 Must pay taxes first 36 34 Other 3 2 Not applicable (no overdue payables) 24 15(-) Hard to say 3 6 Note: Response to question: "What are the most important causes of your overdue pay- ables?" 2 Statistical test is chi . Significance results: ++, statistically greater than rest of sample at the I percent significance level; +, statistically greater than rest of sample at the 5 percent significance level; -, statistically smaller than rest of sample at the 5 percent significance level; - -, statistically smaller than rest of sample at the 1 percent significance level. Source: World Bank survey. "Arrears" in the Ruissian Enterprise Sector 127 Table 4-15. Payment Priorities and Financial Distress: Ranking of Payment Obligations in Order of Urgency Average raink Fiinancially distressedfirns Creditor (total sample) Average ranzk Ordered logit results Suppliers 3.1 2.8 + Banks 2.5 2.5 0 Government (taxes) 1.5 1.9 Employees 2.7 2.6 0 Note: Ranking: 1 = highest; 2 = next highest, and so forth. Ordered logit results: ordered logit of payment rank on financial distress (1/0), size (log employment), citv type and in- dustry dummy variables. Significance of coefficient on financial distress is reported in table: ++, statistically higher ranking than rest of sample at the 1 percent significance level; +, sta- tistically higher ranking than rest of sample at the 5 percent significance level; 0, no differ- ence at 5 percent significance level; -, statistically lower ranking than rest of sample at: the 5 percent significance level; - -, statistically lower ranking than rest of sample at the 1 percent significance level. Source: World Bank survey. these financially distressed firms are not more likely to blame the pres- sure to pay taxes; we will return to this shortly. The responses of managers, and our findings on correlations and -con- centration above, suggest payment priorities by firms as a line of inquiry. How do managers rank their payment obligations in order of urgenicy? They were asked this question explicitly. The answers are summarized in table 4-15. Again, the financially distressed group is considered sepa- rately. Paying suppliers is ranked the least important, which is consistent with the aggregate and survey data on the frequency with which fi:rms pay each other late. Paying taxes gets the highest ranking. In the micddle are the payment obligations of firms to their employees and to the banks, with about equal ranking. The effects of financial distress on the firm's perceived urgency of payments is to lower the priority given to paying taxes, and to raise the priority of paying suppliers. The statistical significance of these findings, reported in table 4-15, is quite robust: we report the significance of the 11/0 financial distress dummy variable in an ordered logit where payment priority is the endogenous variable, but ordered logit equations of pay- ment priorities on the financial distress variable, a size variable, and city and industry dummies yielded virtually identical qualitative results. Our 128 Finanicial Aspects of Enlterprise Restrutctitrng payment priority results are consistent with our findings that financial distress is highly correlated with tax arrears (table 4-11), and with our finding that tax arrears are much more highly concentrated in financially distressed firms than are other kinds of arrears (table 4-6). The explanation of this finding is probably that firms in financial dis- tress have little choice. If they don't pay their suppliers, they will cease to be able to operate because the suppliers will refuse to supply (as shown in the following section). If they don't pay their taxes, they may get into trouble, but the tax authorities may be unwilling to take drastic measures to collect the tax arrears, and in the meantime the firm will continue to operate. We note that Belka and others (1995) found a similar effect of fi- nancial distress on the rankings by Polish firms of their payment priori- ties, with the difference that Polish firms in financial distress also raise the priority of paying their workers. (The average rankings given by the Polish firms also differed from the Russian survey. The average Polish firm gave considerably higher priority to paying workers than the aver- age Russian firm.) The finding that financial distress does not raise the priority of paying workers is consistent with our earlier finding that fi- nancial distress is not correlated with the presence or the volume of wage arrears. That said, we note that financially distressed firms did cite the importance of paying workers significantly more frequently than did nondistressed firms as a reason for overdue payables. We would have ex- pected to find that financial distress would raise the priority of paying workers-firms cainot survive indefinitely without paying their workers. The Responses of Enterprises and Policy Implications In this last section we turn to the supply side of the trade credit market. We have argued so far that most arrears can be considered late payments, with the major exception of tax arrears. Late payments exist because as a liquidity cushion, they are cheaper than, for example, bank finance, but they are possible only because creditors accept late payments. In a situ- ation with no monitoring, or loose monitoring, late payments would turn into bad debts: arrears would not eventually be paid back. But we have already seen some evidence that this is not the case. Cushions are cush- ions and not bad debts because creditors monitor their debtors. Reinforc- ing this last point is the main goal of this section. We wili also discuss the motives for allowing late payment under effective monitoring: that is, "Arrears" in the Russian Eniterprise Sector 129 why creditors have an interest in allowing late payments. The section, and the chapter, concludes with a discussion of policy recommendations. As we saw earlier in the chapter, the total volume of trade credit in Russia in the transition period has remained generally within the levels commonly observed in Western economies. One of the reasons for t-his is that Russian firms have learned to apply fairly basic credit control nmeth- ods to collect on their debts and to keep their overdue receivables under control. Some case study evidence (for example, Fan and Schaffer 1994; Singh and Gelb 1994) suggests this lesson was learned fairly quickly by firms, within the first year of transition. The evidence from the World Bank survey demonstrates that by mid-1994, the importance of ciredit controls was widely appreciated, and basic credit control methods wvere in general use. When managers were asked to identify the most important features of their management strategy in the area of financial management, the sin- gle most common action listed, by a wide margin, was "reducing out- standing receivables." Close to two-thirds of the sample said this was "very important"; all other measures'7 were listed as "very important" by less than half the firms. The importance of obtaining trade credit was also widely appreciated; 43 percent of firms said lengthening the period for payables was "very important." The methods firms use to control their overdue receivables (table 4-16) are fairly commonsense measures, and are similar to those found in surveys of Polish and Hungarian firms (Belka and others 1995; Bonin and Schaffer 1995). Few firms (10 percent) said they had no problems with overdue receivables; virtually all of the remainder used at least one method, and on average they used between two and three. The most common method, used by a majority of firms, was to require pay- ment in advance or cash on delivery from new customers. The second most common action was to apply the same requirement to established customers; Russian firms are apparently wary of extending trade credit, even to long-term customers. In both cases, partial or full payment in advance was the main requirement; only about 10 percent of firms asked for cash on delivery. The other main strategies used were to refuse to supply a customer behind in its payments until repayment had been arranged, and "informal methods," both used by about 40 percent of firms. Relatively few firms (15 percent) charged interest on their overdue receivables. This is somewhat surprising given that a majority of firms in 130 Financial Aspects of Enterprise Restruicturinig Table 4-16. Methods Used to Control Overdue Receivables (percent) Method Firnms 7vwhere meth1od is uised Partial or full payment in advance or cash on delivery, from new customers 55 Partial or full payment in advance or cash on delivery, from established customers 44 Charge interest on overdue receivables 15 Refuse to supply if customer is behind on payments until old debt is paid in part or in full 41 Informal methods (phone, letter, . .) 38 Legal action (lawyer, court system) 27 Sell overdue receivables on the debt market < I Other methods 3 No special methods used 2 Not applicable-no problems with overdue receivable 10 Hard to say 3 Average number of methods used (excluding not applicable) 2.6 Response rate 439/439 (100%) Note: Response to question: "What methods do you use to control the level of your receiv- ables? (Tick most important)." Sozurce: World Bank survey. the Polish and Hungarian surveys utilized this practice."8 Finally, about one-quarter reported resorting to the court system to collect on overdue receivables. These findings suggest Russian firms have learned the importance of credit control and are imposing hard budget constraints on each other, mostly through very basic methods. In this way, financial discipline is be- ing imposed on firms by the market. The number of firms that acknowledge resorting to the law to collect on overdue debts is surprisingly high, given the perceived problems with the court system. The managers in the survey were asked what they saw as the main obstacles to filing for the insolvency of a debtor who had not paid them. The answers are reported in table 4-17, and suggest that in Russia, managers are aware of how to use the bankruptcy framework and appear to be able to make economic decisions about whether or not it pays to use it (for a discussion of bankruptcy frameworks in transition "Arrears" in the Russian Enterprise Sector 131 Table 4-17. Obstacles to Pursuing Debtors (percent) Obstacle Firms citinlg the obsta2cle The probability of repayment is low because the firm is highly indebted to other creditors or because other creditors have priority. 47 The situation in the economy is complicated and we have to support our partners. 38 Russian courts are weak and incompetent and court procedures won't produce the desired outcome. 26 The bargaining power of the firm vs. the debtor means recovery of the debt is more likely outside the court system, by informal means. 15 The firm has no other customers, so bankrupting the debtor would be costly. 15 No major obstacles to filing. 10 Hard to say. 13 Note: Response to the question: "What are the main obstacles to filing for the insolvency of a debtor who hasn't paid you?" Source: World Bank survey. economies, see Baer and Gray 1995). Only 10 percent of managers said they saw no major obstacles to filing for the itnsolvency of a debtor. The most common obstacle, cited by 47 percent of firms, was that the prob- ability of repayment was too low, either because the debtor was highly indebted to other creditors or because other creditors had priority irt re- payment. A quarter of the sample said that the weaknesses of the Russian court system were an important deterrent. The ability of the firm to use its bargaining power outside the court system to collect from the debtor, and the fear that bankrupting the debtor would cost the firm a major cus- tomer, were both cited by 15 percent of firms. Curiously, the second most commonly cited "obstacle" was the statement that in the current eco- nomic environment, firms needed to support their customers and not bankrupt them.19 It is unclear how we should interpret this response--it may be a softer variant of the statement that bankrupting a debtor can mean losing a customer, or evidence that Russian managers are prepa.red to act altruistically. We also note that bankruptcy law reform is seen as important, but not especially so, by the managers surveyed. When asked 132 Financial Aspects of Enterprise Restricturing what legal areas needed reform, bankruptcy law was in eighth place (the clear leaders were taxation and banking regulations). We have presented clear evidence that Russian firms have imple- mented basic credit control mechanisms themselves, that trade credit flow problems are not great, and that trade credit arrears in Russia are not large by either Western or transition country standards. This suggests that basic market-oriented behavior by Russian firms has been effective in containing the volume of trade credit arrears, and to speak of an "inter- enterprise arrears crisis" exaggerates the problem. The most useful policy measures regarding trade credit arrears would be those that introduced or improved existing market mechanisms. One avenue would be to facilitate or provide incentives for the tradability of trade credit, as argued by Rostowski (1994a, b). For example, in Poland a firm may freely offset a payable of another firm against the cost of goods purchased from that firm. A customer that has taken delivery of coal from a coal mine, for example, may purchase a payable of the coal mine (at a discount, from someone who is having trouble collecting the debt from the coal mine), and use that payable to pay for the coal. There is consequently an active, albeit small, market for trade credit in Poland. Another avenue would be to introduce legal and institutional reforms that improve the incentives for creditors to pursue debtors who do not pay. In contrast, forcing firms to "do something" about their arrears with- out changing their incentives is liable to be ineffective. Setting up a cen- tralized scheme for clearing or managing trade credit arrears, such as a government agency that is to identify firms for special treatment, is un- likely to be as effective as decentralized, market-oriented reforms. Such schemes can also backfire if "special treatment" is at all generous; firms will lobby for special treatment and will stop paying their suppliers if the prospect of special treatment is attractive enough. The same problem ap- plies to generalized bailout schemes (such as the mutual arrears-clearing operation in 1992). Such schemes send precisely the wrong message to firms-that "it doesn't matter if you can't pay your supplier, because if you can't, the government will do it for you." Moreover, it may encour- age the formation of "collusive arrears" by providing firms with a coordi- nation mechanism. Wage arrears, we have seen, are commonly found in low-wage firms and in firms hit by losses of demand, but are not systematically found in firms in financial distress. Our view of wage arrears is that to some extent "Arrears" in the Rlussian Enterprise Sector 133 they represent the promise by managers to pay workers with money the firm doesn't have. Why managers make these promises is an interesting question. One reason may be that the existence of wage arrears in a firm makes is easier for the managers to lobby the government for subsidies, cheap credits, or other financial assistance. Another reason may be the relative power of managers in relation to workers, which the standard view suggests is greater in Russia than in Central and Eastern Eutope. We have seen, for example, that the relative priority managers place on paying workers is lower in Russia than in Poland, and wage arrears are rarely observed in CEE countries. In any case, we do not see a need for policies to deal with wage arrears in firms. Rewarding unrealistic prom- ises by managers to their workers only damages incentives. The so-called 30:70 rule is an example of how a policy that attempted to "solve" the wage arrears problem ended up creating new problems, in this case a large increase in tax arrears. In our view, tax arrears and arrears to banks are much more pressing problems for policymakers. (Policy regarding the banking sector is dis- cussed in the chapter 5.) With respect to tax arrears, much of the problem appears to be limited to financially distressed firms that do not have the money to pay. It is tempting to allow such firms to continue to operate. If such a firm is receiving no other subsidies, it must be covering the costs of its inputs, and hence is generating positive value added. Reallocation of the labor and assets of the firm following its liquidation would likely be slow. There is a danger, however, that poor tax discipline will spread from these firms to financially healthy firms that can but will not pay, and it is important for nonpayment of taxes to be an unattractive option for healthy firms. The most useful effect of liquidating firms with la:rge tax arrears would be to send the right signals to firms that are considering nonpayment or late payment of taxes. To the extent these are distressed firms, the liquidation value of the firm, and hence the recovery rate on the tax arrears, is likely to be low. If the survival of these enterprises is sEen by the govermment as vital for either political or economic reasons, then subsidies should be made explicit and written in the budget expenditures plan, as argued by Alfandari, Fan, and Freinkman in chapter 6. Schemes that allow firms to defer tax payments, such as the 30:70 rule, should be avoided. They encourage poor tax payment discipline by firms in general and are difficult to limit to firms whose survival the government sees as crucial. 134 Financial Aspects of Enterprise Restruicturing Appendix: Definitions and Coverage of Goskomstat Data We refer the reader to table 4-1, which presents a simplified Russian bal- ance sheet. Between October 1993 and February 1995, the Goskomstat monthly bal- ance sheet data are as follows. (All stock items are for the 1st of the month.) Assets Total debtors ("debitorskaia zadolzheninost"'), both total and overdue. This con- sists of the "total debtors" portion of the balance sheet, pluts advance pay- ments. Interest and late penalties charged by enterprises to their customers appear in the "accounts with other debtors" line in section III of the assets side of the balance sheet, and are therefore incltuded in this "total debtors" figure. Also included are recalculations of tax payables, which may be positive or negative depending, for example, on whether actual taxes paid (as based on the tax payment schedule set out by the tax collection authority) are greater or less than the actual tax liability sub- sequently incurred. Receivables from customers for goods and services ("zadolzhennost' pokut- patelei za tovary, rabotjy, utsluigi"), both total and overdute. This is a subset of "total debtors"-it accounts in aggregate for about 80 percent of the lat- ter-and corresponds to the "receivables for goods and services" line of the balance sheet. Receivables for goods and services are recorded in transactions prices ("v tsenach realizatsii") rather than wholesale or pro- ducer prices, and in the great majority of enterprises are recorded at the time of shipment. Interest and penalties charged by firms to their custom- ers for late payment are not included in this item, but rather in the figure for "total debtors" (see above). Monetary assets ("denezhnye sredstva"). This consists of the monetary as- sets portion of the balance sheet. It includes hard currency holdings. Financial investments ("finansovye vlozheniia"). This is composed of short-term financial investments (from section III of the asset side balance sheet) and long-terni financial assets (from section I of asset side the bal- ance sheet). Liabilities Total creditors ("kreditorskaia zadolzlhennost"'), botlh total and overduie. This consists of the "total creditors" portion of the balance sheet, plus advance "Arrears" in the Russian Enterprise Sector 135 payments. (Note that both bank credit and loans are excluded.) Most in- terest and late penalties charged by creditors to the firm appear in the "accounts with other creditors" line in section III of the liabilities side of the balance sheet, and are therefore included in this "total creditors" fig- ure. In particular, late penalties charged by suppliers and unpaid inte:rest payments to banks are included in this figure for "total creditors." Payables to suppliers for goods and services ("zadolzhennost' postavshchikam za tovary, raboty, uslugi"), both total and overdue. This is a subset of "total creditors"-it accounts in aggregate for about 60 percent of the latter- and corresponds to the "payables for goods and services" line of the bal- ance sheet. Penalties for late payment charged by suppliers are not included, but do appear in the figure for "total creditors" (see above). Payables to the budget ("zadolzhennost' v biudzhet"), both total and overdue. This is also a subset of "total creditors"; at the end of 1993, it made up about 15 percent of total creditors. It is the same as the corresponding line on the balance sheet, and so does not include payments to extrabudgetary funds. The figure for payables to the budget includes interest and penal- ties for late payment of taxes. Bank credit ("zadolzhennost' po kreditam bankov"), both total and overlue. This is composed of short-term bank credit (from section III of the liabil- ity side of the balance sheet) plus long-term bank credit (from section II of the liability side of the balance sheet). It does not include "bank credit for employees" (see section III of the liabilities side) nor interest arrears. "Loans" ("zadolzhennost' po poluchennym zaimamn"), both total and over- due. This is composed of short-term loans (from section III of the liability side of the balance sheet) plus long-term loans (from section II of the li- ability side of the balance sheet). Miscellaneous Total delivered productioni, works anid services ("ob'em otgruzhennoi prod uktsii, vypolnennykh rabot, okazannykh usluig") to the start of the nmonth. At the same time firms report balance sheet data for the 1st of the month, they report the preceding month's delivered production. "Delivered production" is measured in transaction prices ("v tsenach realizatsii"); that is, the prices at which goods are actually sold. This means that value added tax (VAT), excise taxes, etc. are included, and product subsidies are not included. By contrast, the figures Goskomstat usually reports for industrial product:ion are in producer prices, meaning that they are net of VAT and excise talxes and gross of subsidies. This difference is moderately important in aggre- 136 Financial Aspects of Enterprise Restrtucturing gate-on average, delivered production in industry is about 15-20 per- cent greater than the usual industrial production figures-and very im- portant for sectors that pay high excise taxes (for example, petroleum extraction, where delivered production is roughly double production at producer prices). References Abel, Istvan, and Pierre L. Siklos. 1993. "Constraints on Enterprise Li- quidity and their Impact on the Monetary Sector in Formerly Cen- trally Planned Economies." CEPR Discussion Paper No. 841, London. Baer, Herbert L., and Cheryl W. Gray. 1995. "Debt as a Control Device in Transitional Economies: The Experiences of Hungary and Poland." World Bank, Policy Research Working Paper No. 1480, Washing- ton, D.C. Begg, David, and Richard Portes. 1993. "Enterprise Debt and Financial Restructuring in Central and Eastern Europe." Eutropeani Economic Review 37: 396-407. Belka, Marek, Saul Estrin, Mark E. Schaffer, and I. J. Singh. 1995. "Enter- prise Adjustment in Poland: Evidence from a Survey of 200 Pri- vate, Privatized, and State-Owned Firms." CEP Discussion Paper No. 233, London School of Economics. Bigman, David, and Sergio Pereira Leite. 1993. "Enterprise Arrears in Russia: Causes and Policy Options." IMF Working Paper No. 61, Washington, D.C. Bonin, John P., and Mark E. Schaffer. 1995. "Banks, Firms, Bad Debts and Bankruptcy in Hungary 1991-94." Centre for Economic Perform- ance Discussion Paper No. 234, London School of Economics. Chittenden, Francis, Anthony Kennon, Suneil Mahindru, and Richard Bragg. 1993. "Payment Practices, Legislation and their Effect on SMEs: A Comparative Study." National Westminster Bank, Man- chester, U.K. Condon, Timothy, and S. Ramachandran. 1995. "Cash Constraints and Credit Corsets: The Chimera of Interenterprise Credit." World Bank, Private Sector Development Department Note No. 41, Washington, D.C. "Arrears" in the Ruissian Eniterprise Sector 137 Fan, Qimiao, and Mark E. Schaffer. 1994. "Governmental Financial Trans- fers and Enterprise Adjustments in Russia, with Comparisons to Central and Eastern Europe." Econ7omi1ics of Tranisition 2 (2): 151-88. Gaidar, Yegor. 1995. "Russian Reform." In Yegor Gaidar and Karl Otto Pohl, Ruissian Reform/International Moniey. Cambridge, Mass.: MIT Press. Gomulka, Stanislaw. 1994. "The Financial Situation of Enterprises and Its Impact on Monetary and Fiscal Policies, Poland 1992-1993." Eco- nomics of Transition 2 (2): 189-208. Gray, Cheryl, Sabine Schlorke, and Mikl6s Szanyi. 1995. "Hungary's Bankruptcy Experience, 1992-93." World Bank Policy Research Working Paper No. 1510, Washington, D.C. Ickes, Barry W., and Randi Ryterman. 1992. "Interenterprise Arrears and Financial Underdevelopment in Russia." World Bank Policy Re- search Department, Washington, D.C. Photocopy. Koen, Vincent, and Michael Marrese. 1993. "Stabilization and Structural Change in Russia, 1992-1994." In Road Maps of the Transition IMF Occasional Paper No. 127. Washington, D.C. Koen, Vincent, and Steven Phillips. 1993. Price Liberalization in Ruissia: Be- havior of Prices, Household Incomes, and Consumption During the First Year. IMF Occasional Paper 104. Washington, D.C. Mitchell, Janet. 1995. "Strategic Creditor Passivity in Economies in Transi- tion." Working Paper No. 442, Department of Economics, Cornell University, Ithaca, N.Y. Perotti, Enrico C. 1994. "Collusive Arrears in Transition Economies." Fi- nancial Markets Group Discussion Paper No. 198, London School of Economics. Rostowski, Jacek. 1994a. "Comment." In Gerard Caprio, David FoLkerts- Landau, and Timothy D. Lane, eds., "Building Sound Finance in Emerging Market Economies." IMF and The World Bank, Wash- ington, D.C. . 1994b. "Interenterprise Arrears in Post-Communist Economies." IMF Working Paper No. 43. Washington, D.C. Schaffer, Mark E. 1995. "Government Subsidies to Enterprises in Central and Eastern Europe: Budgetary Subsidies and Tax Arrears." In David M. G. Newbery, ed., Tax and Benefit Reform in Central and Eastern Europe. London: CEPR. 138 Financial Aspects of Enterprise Restrtctutring Singh, I. J., and Alan H. Gelb. 1994. "Public Enterprise Reforns in Transi- tional Economies." World Bank, Policy Research Department, Washington, D.C. Photocopy. Smith, Jennifer C. 1995. "Interenterprise Debt and Monetary Policy in the UK." Bank of England, London. Photocopy. Sunley, Emil M., John Norregaard, James A. Daniel, Victoria P. Summers, and Philippe H. Le Houerou. 1995. Rtussianl Federation: Tax ReformT and Development. Washington, D.C.: Fiscal Affairs Department, In- ternational Monetary Fund. Notes 1. For example, the Russian receivables/sales ratio in industry in 1992 would be biased seriously upward if we used 1992 nominal industrial production and end-year trade credit because trade credit would be largely in late-1992 prices. 2. This is also supported by CBR data on overdue bank credit. Unlike the Goskomstat data, the CBR data include interest arrears. In the CBR data, overdue bank credit as a percentage of total bank credit is considerably higher than in the Goskomstat data, and closer to the percentage overdue in the World Bank survey data. See chapter 5 in this volume for more details. 3. Thus the survey shows that only 24 percent of firms do not have any over- due payables: the same category of firms totalled 47 percent on the same date, according to Goskomstat data. This difference could also be caused by under- representation of small firms in the survey (see the appendix to this volume); as we shall see below, firm size is positively correlated with arrears. 4. This is perhaps not surprising, since nonoverdue trade credit will be defined primarily by standard payment terms (for example, "payrment due in 30 days"). 5. The data for Russia derive from Goskomstat. For more details on data sources and estimates, see the notes to the table. 6. The estimate is derived starting with Goskomstat data on tax arrears for firms in industry, construction, agriculture, and transport only, and then scaling up, using Goskomstat data on total receivables/liabilities of the entire enterprise sector. 7. We are grateful to Robin Adair of the OECD for information on tax arrears in New Zealand. 8. These arrears were those in the so-called "kartoteka 2." They were techni- cally unpaid payment demands that had arrived at the bank accounts of firms to be paid, but because firms did not have sufficient funds in their bank accounts, the payment demands were queued in the special "second file" until sufficient funds arrived. "Arrears" in the Russian Enterprise Sector 139 9. The queuing in "kartoteka 2" (see previous note) was first-in-first-out, meaning firms, in principle, did not have a choice in deciding which unpaid pay- ment demands to pay first. 10. That is, the "kartoteka 2" files were frozen, and the rules were changed so that in most circumstances when a firm did not have sufficient funds to pay a payment order when it arrived at the firms' bank, the payment order woulcd not be queued, but would instead be returned to the originator. 11. The effect is the same as regressing percentage overdue of arrear Y on per- centage overdue of arrear X plus a full set of industry dummy variables. 12. The firms in the survey clearly understood this problem. Most of the firms identified problems with their accounting systems, and when asked which were the most serious, the most common response was understatement of costs and overstatement of profits resulting from historical cost accounting (67 percent of firms identifying accounting problems). "Unclear or inconsistent accounting rules" (56 percent) and "depreciation charges insufficient to replace fixed capital" (29 percent) were less frequently mentioned. 13. The high level of importance attached to input prices may reflect a genleral concern with inflation, as well as particular concem with the relative prices of inputs. 14. That is, we test log(arrears/sales). The mean m is m = mean(log(ar- rears/sales)). We report er', the estimated arrears/sales ratio. The Wilcoxon rank- sum test for medians is scale-independent, and thus testing the ratio or the log ratio yields the same results. 15. Note that the samples on which the correlations are based include firms with no arrears, unlike the analysis of concentration of arrears earlier, in which firms with zero arrears were excluded. 16. Initially, a firm with tax arrears could use 50 percent of the funds available in its bank account to pay wages instead of taxes; this was later reduced to 30 per- cent (hence the term "30:70 rule"). For more on the 30:70 rule, see Sunley and oth- ers 1995. 17. Reducing, rescheduling, or obtaining new loans; lengthening the period for payables; changing bank connections; seeking foreign investors or partners; or "other." 18. This does not, of course, necessarily mean that Russian firms are ignoring inflation. As noted earlier, Russian managers have a lot of experience with Sinfla- tion, and we would expect the price charged to customers to include a component for anticipated inflation. Nevertheless, the infrequency with which interest is charged does suggest that we need to be careful about inflation effects when in- terpreting the data on commercial payables. 19. This option was not included in the original draft survey, and was possi- bly inserted by the local survey team or volunteered by the firms. 5 Firms, Banks, and Credit in Russia Qimiao Fan, Une J. Lee, and Mark E. Schaffer The relations between enterprises and banks have been at the center of both enterprise reforms and financial sector reforms in many transition economies. On the one hand, financial sector reforms have been ham- pered by bad loans and the lack of adjustment of the enterprise sector. On the other hand, the lack of enterprise restructuring has been accommo- dated by the softness of the financial sector. It has become increasingly clear that for enterprise reform and financial sector reform to succeed, they have to be dealt with simultaneously and jointly. Financial sector reforms were introduced in Russia in the late 1980s with the restructuring of the monobank into a two-tiered system, with a Central Bank responsible for monetary and credit policy and regulatory functions and a second tier of five specialized banks that took over lend- ing and deposit-taking functions. The commercial banking sector has evolved rapidly since then, more so in Russia than in most Central and Eastern European (CEE) economies, with the fragmentation of the spe- We are very grateful to Anders Aslund, Barry Bosworth, Simon Commander, Richard Jackman, and the participants at the June 1995 St. Petersburg conference and the March 1995 Washington, D.C., workshop for helpful suggestions and comments. 140 Firms, Banks, and Credit in Ruissia 141 cialized banks and the emergence of increasing numbers of independent, but primarily small and poorly capitalized, commercial banks.1 As of mid-1995, there were approximately 2,500 commercial banks registered in Russia. Many of these banks originated as spin-offs of the former specialized state banks. This group includes some of Russia's larg- est banks, with a broad shareholder base that includes the state, which has retained partial ownership through the Central Bank of Russia (CBR), GKI (the state property committee), or other agencies. The majority of commercial banks in Russia, however, are new, or so-called "zero" banks, created from scratch by their shareholders. In most cases the shareholders were public or semi-public institutions and enterprises that wanted a fi- nancial arm to facilitate their credit needs, as well as to perform a number of their treasury functions, and thus the banks they founded are some- times also known as agent banks. Ownership of these new banks tends to be concentrated, often with only a few shareholders. With privatization of the enterprise sector, bank privatization for all categories of banks has en- sued, although the state does retain shares in some cases; for example, the state owns substantial shares of Sberbank and Vneshtorgbank through the CBR (current law forbids CBR ownership of commercial banks). Early on in the reform process, government-directed credits domi- nated lending. A large share of the available financial resources went to targeted enterprises and regions through CBR and federal and local gov- ernment credit programs at subsidized rates below Central Bank discount rates, which were themselves negative in real terms until late 1993. The commercial banks, particularly the former state banks, were used to channel the bulk of government-directed credits. In this process they played a passive role; the credits were not only directed but also funded by government or CBR sources rather than commercial banks' own re- sources. Total directed credits have been reduced significantly since 1992 but remain substantial: from about 48 percent of total bank credit in 1992 to about 29 percent in 1993 and 23 percent in 1994 (Koen and Marrese 1995, p. 58). Many of the remaining programs are dominated by directed credit for agriculture and the northern territories. The ownership structure of the majority of new commercial banks also promoted the practice of connected, or insider, lending. Such lending practices acted to segment the market, leaving many new enterprises with extremely limited access to financial resources. Lending rates have also been known to vary depending on a number of factors, including the bank-enterprise relationship. 142 Financial Aspects of Enterprise Restructuhrintg A repercussion of commercial banks' lending practices-highly con- centrated loan portfolios as well as connected lending-is their greater exposure to risk. This is complicated by the rising portion of commercial bank funds composed of time and savings deposits of individuals com- pared with the funds raised from deposits of enterprises and organiza- tions. Moreover, commercial banks, in theory, bear the credit risk on government-directed credits to enterprises.2 This is an added source of risk given that such loans are largely based on the recipients' bargaining or political power rather than on financial or economic considerations. More generally, there has been a dramatic increase in overdue bank credit since 1992. As of mid-1995, overdue bank credit amounted to the equiva- lent of over one-third of total bank credit, or about 3 percent of annual- ized GDP (more about this below). The freezing of the interbank credit market in Russia in mid-1995 also demonstrated the fragility of the com- mercial banking sector. Among this large universe of commercial banks in Russia, however, there is a growing population-about 50-100-of emerging "good" or "real" commercial banking institutions. This group is made up of both new banks and spin-offs of the former specialized state banks, and it in- cludes some of the largest private independent banks in Russia. These banks are making efforts to strengthen their balance sheets and banking skills, including reducing their connected-party lending and single-bor- rower exposure. These banks tend to lend to new private entities and are, in general, involved in profitable and sophisticated banking activities. The emerging characteristics of the Russian banking system highlight the intertwined relations between enterprises and banks. We now turn to what the survey tells us about this relationship. General Features At the time of the survey (mid-1994), the majority of firms held some bank credit, but a large number of firms did not: 53 percent of the total sample had some debt to banks. The large number of firms without bank debt reflects the continuing presence of nonbank sources of credit. There is relatively little variation by ownership category; 41 percent of the de novo firms had some bank debt, even though they are new and small and might be expected to hold significantly less bank debt than established, larger firms. Firms, Banks, and Credit in Ruissia 143 A separate question, with a higher (nearly 100 percent) response rate, asks if firms have received bank loans in the past two years. Here the number of bank loan recipients is rather high, at 81 percent. The differ- ence is probably attributable to firms that once held bank debt but at the time of the survey did not; some may be "between loans." In any case, the pattern by ownership type is similar: 62 percent of de novo firms had re- ceived bank loans in the past two years, which is lower than the figure for the total sample, but not by very much. Firms that do have loans, not just from banks but from other sources as well, also provided some information about their two largest loans. Conmmercial banks are listed as the sources for about two-thirds of all loans; the second most common source is the CBR, providing about 13 percent of loans. Other firms, oblast governments, and the Ministry of Fi- nance (MoF) each account for no more than 5 percent of loans, withi the remaining sources given as "other" or not available. There is no apparent correlation between the source of the loan and the firm's ownership type (state-owned, privatized, de novo). Aggregate data from both the CBR and Goskomstat show that most bank loans in Russia are short-term, and this is the pattern we see in our data as well. One-third of loans were for three months or less, 60 percent were for six months or less, and over 80 percent were for one year or less. We note here that the high rates of inflation seen in Russia have the effect of making bank loans even more short-term in practice. Inflation has the effect of front-loading the repayment schedule of a bank loan, because the real value of the principal will depreciate over the term of the loan. The length of the loan period is, of course, related to the use ol the loan. When firms were asked about the use of their largest loans, half of the sample said the loan was for operating (or "working") capital. About 15 percent of loans were for productive fixed investment, and 7-8 percent were conversion loans (nearly all of these were held by military-indus- trial complex, or MIC, firms). Very few loans were for nonproductive in- vestment (that is, in social assets). The remainder were for miscellaneous, or "other," uses. The most interesting feature of the use of loans with re- spect to ownership is the much larger share of loans to de novo firms (over 40 percent) that are funding productive fixed investment. Virtually all loans, regardless of source, were collateralized. Between 40 and 50 percent of loans were collateralized by inventories, another quarter used machinery and equipment as collateral, and receivables and 144 Financial Aspects of Eniterprise Restrictuiring Table 5-1. Ownership Cross-holdings in Banks and Firms (percentage of cases) Major Minior Not a Not Category shareliolder shiareholder shlareholder applicable Unknowwn Firm a shareholder in the lending bank? Loan 1(209 cases) 11 29 47 10 2 Loan 2 (105 cases) 10 27 46 14 3 Bank a shareholder in the borrowing firm? Loan 1(209 cases) < 1 3 85 8 3 Loan 2 (105 cases) 2 4 78 12 4 Note: Figures may not sum to 100 percent because of rounding. buildings and land were each used in another 5-10 percent of cases. The value of collateral as a percentage of the loan was about 120 percent on average, but the variation by loan and firm was substantial. This may re- flect large variations in the relationship between book and market value of assets more than anything else. We should not place too much empha- sis on the use of collateral, however, because of the practical (legal) diffi- culties facing a creditor that wants to take possession of collateralized assets. Finally, firms provided information on cross-holdings between them- selves and the banks that made the loans, summarized in table 5-1. It is common, but not standard, for firms to hold shares in the bank making the loan, but not to be dominant shareholders. Firms were minor share- holders in close to 30 percent of cases, but major shareholders in only about 10 percent. In close to half of the cases reported, firms were not shareholders at all in the lending bank, and in about 10-15 percent of cases the question was not applicable or no answer was available. "Agent banks," banks set up (and owned) by firms to handle their lending and little else, do not seem to play a very sizable role in the credit allocation process. De novo firms were rather less likely to own shares in their banks; such ownership was found in only a few cases. We note that in a majority of the loans that listed the CBR, the oblast government, or the MoF as the source that is, for a majority of cases of directed state credits (DSCs)-firms noted that the shareholding question Firms, Banzks, and Credit in Russia 145 applied, and in most cases the answer was "not a shareholder." Firms probably answered this question because the DSCs were charneled through or administered by a commercial bank. Nevertheless, it is some- what surprising that firms were rarely shareholders in the commercial bank acting as the channel in these cases (and only a few reported that they were majority shareholders). This would suggest that agent banks are not, or are no longer, a very important channel for DSCs. So far the picture is one of relatively few (subsidized) state credits. This is consistent with a separate question in the survey, albeit one with a relatively low response rate: on average, firms estimate that 85 percent of their loans from commercial banks (and, from what we have seen, firms probably included DSCs channeled through commercial banks as well) are on commercial terms, meaning at or above the CBR's discount rate. Although firms are often shareholders in their banks, banks are rarely shareholders in the client firms; firms reported lending banks as share- holders in only about 5 percent of cases. It is rare for firms to be partially owned by any banks, let alone by banks that lend to them. Elsewhere in the questionnaire, when firms were asked about their ownership struc- ture, only 7 percent of firms reported that banks owned any of their shares. The average shareholding of banks in these cases was 13 percent. Shareholding by nonbank domestic financial institutions (for example, in- vestment funds), some of which may be controlled or owned by banks, is somewhat more common-25 percent of firms reported such instit-ations holding some of their shares. The average shareholding of the instit-utions in these cases was 15 percent.3 The Distribution of Bank Debt and the "Bad Debt" Problem It is common for the banking system in transition countries to become saddled with "bad debts"-loans to firms that are partially collectible at best, and should at some point be qualified, written down, workEcd out, and so forth. In this section we first use official aggregate data and then data from the enterprise survey to try to make a first estimate of the size of the Russian bad debt problem. Table 5-2 presents the available official data on overdue bank debt. The data come from two sources, the CBR and Goskomstat. The CBR's data on overdue bank debt derive from the lenders-the banks-and cover the entire credit stock. Two CBR data series are presented in ta- Table 5-2. Bank Credit and Overdue Bank Credit in Russia, 1990-95 CBR dataa Goskomstat datab Total bank Overdue Total Overdue bank credit, bank bank Overdue bank credit Total bank Overdue, percentage Annualized credit, percentage percentage credit, credit, Overdue, (R bn) credit of total bank credit monthly ofannualized GDP of annual- covered covered percentage Date Old series New series (R bn) Old series New series GDP (R bn) Old series New series ized GDP sectors sectors of total 01.01.90 0.9 211 0.4 573 0.2 36.9 n.a. n.a. n.a. 01.01.91 4.5 178 2.5 612 0.7 29.1 n.a. n.a. n.a. 01.01.92 4.5 439 1.0 1,253 0.4 35.1 n.a. n.a. n.a. 01.07.92 47 1,393 3.3 14,784 0.3 9.4 n.a. n.a. n.a. 01.01.93 140 140 5,102 2.7 2.7 39,876 0.4 0.4 12.8 n.a. n.a. n.a. 01.04.93 248 248 8,370 3.0 3.0 73,200 0.3 0.3 11.4 n.a. n.a. n.a. 01.07.93 337 1,490 15,773 2.1 9.4 145,200 0.2 1.0 10.9 n.a. n.a. n.a. 01.10.93 654 2,158 23,417 2.8 9.2 205,200 0.3 1.1 11.4 422 6,497 6.5 01.01.94 1,609 3,623 30,019 5.4 12.1 366,000 0.4 1.0 8.2 997 10,443 9.5 01.04.94 6,507 38,888 16.7 434,400 1.5 9.0 1,774 13,676 13.0 01.07.94 11,590 52,940 21.9 576,000 2.0 9.2 2,811 23,184 12.1 01.10.94 17,382 69,865 24.9 673,200 2.6 10.4 4,535 33,117 13.7 01.01.95 26,007 83,561 31.1 1,010,400 2.6 8.3 5,345 39,554 13.5 01.04.95 41,043 105,951 38.7 1,129,200 3.6 9.4 7,486 48,442 15.5 01.07.95 39,218 112,877 34.7 1,670,400 2.3 6.8 8,642 52,629 16.4 01.10.95 42,348 126,518 33.5 2,064,000 2.1 6.1 8,821 65,001 13.6 n.a. Not available. Note: CBR "old series" is prior to revision in January 1994; CBR "new series" is following revision in January 1994 (sec text). CBR overdue bank credit, new series, report- edly includes arrears on interest payments. Goskomstat overdue bank credit does not include arrears on interest payments. Annualized GDP is 12*GDP in preceding month, except for 01.01.90, which is 1989 annual GDP. a. Coverage: entire economy. b. Coverage: industry, agriculture, transport, contstruction; reporting enterprises only. SouLrce: CBR, Goskomstat. Firms, Banks, and Credit in Ruissia 147 ble 5-2. The "old series" refers to data published prior to January 1994 in the CBR's Butlletin of Bank Statistics; in that month the CBR revised up- ward the figures for overdue bank credit very substantially. We have no information on the reasons for the revisions, which extended retrospec- tively to mid-1993; early 1993 figures were not revised, but it is unclear if these figures are actually compatible with the revised numbers. The CBR reportedly defines "overdue bank credit" in its "new series" to include interest arrears on overdue credit; we have no information on the treat- ment of interest arrears in the "old series." Goskomstat's data on overdue bank credit derive from the borrow- ers-enterprises-and cover reporting firms in industry, agriculture, transportation, and construction only. Based on figures for total bank credit, Goskomstat's coverage amounts to one-third to one-half that of the CBR's data. Overdue bank credit in the Goskomstat data is defined as overdue by the reporting firm, and it does not include (uncapitalized/un- rescheduled) interest arrears. Comparing the CBR and Goskomstat data on overdue bank credilt, we find that the percentage of total bank credit represented by this category is consistently higher in the CBR's (new) data. At the time our survey was conducted (mid-1994), for example, overdue bank credit amounted to over 20 percent of total bank credit according to the CBR, but only 12 per- cent according to Goskomstat. The difference is probably largely ex- plained by the treatment of interest arrears. In a period of high inflation, a large portion of the total liability associated with an unserviced and over- due bank loan will be in the form of interest arrears. The Goskomstat data on overdue bank credit are biased downward, probably badly, because interest arrears are not included. The CBR and Goskomstat data are, how- ever, in rough agreement concerning recent trends-both show a signifi- cant increase in the volume of overdue bank credit. In the case of the CBR data, the increase is particularly dramatic: overdue bank credit rises from 9 percent of total credit in late 1993 to a remarkable 37 percent in mid- 1995. Because the total credit stock in Russia is relatively small (most of it was wiped out in early 1992 by inflation and it has stayed low silnce), however, the volume of overdue bank credit is not huge; it amounted to about 3 percent of annualized GDP in mid-1995. How do these figures compare with those from other transition coun- tries? In Hungary, at the end of 1992 (after about two-thirds of the bad debt problem had emerged), overdue bank credit, including interest ar- rears, amounted to about 25-30 percent of total bank credit, or about 7 148 Financial Aspects of Enterprise Restructutring percent of GDP (see Bonin and Schaffer 1995). Goskomstat's data on overdue bank credit are comparable to those collected by the Polish Cen- tral Statistical Office; both derive from enterprise balance sheets, and nei- ther includes interest arrears. At the end of 1992, after most of the Polish bad debt problem had emerged, 9 percent of credits and loans to firms in industry, transport, and construction were overdue (Poland 1993, p. 162). This last figure badly underestimates the volume of bad bank debt in Po- land, probably for reasons similar to those producing lower Goskomstat figures (notably the failure to include interest arrears). Estimates of the to- tal Polish bad debt problem at this time range from 20-40 percent of total bank credit, or perhaps 3-6 percent of GDP. Based on these figures, the current Russian bad debt problem would appear to be roughly compara- ble in scale to that in Hungary or Poland. These estimates and comparisons are of limited reliability, however, because of a fundamental problem with the definition of "bad" or "over- due" bank credit. It is common in transition countries for banks to roll over bad debts by rescheduling principal payments and capitalizing in- terest arrears. Once this is done, a debt is no longer formally "overdue," even if it is still genuinely "bad." Reported figures on overdue bank credit will depend in part on how frequent this practice is: the more com- mon it is for firms in financial difficulties to get their arrears to banks re- scheduled and rolled over, the more these numbers will understate the true scale of the bad debt problem. At the same time, a healthy firm with a loan from a "soft" bank may not put a high priority on prompt pay- ment and may regularly pay late-but still pay. For this reason, figures on overdue bank credit can (in principle) also overstate the volume of genuinely bad debt. Even if both factors are operating, however, a signifi- cant volume of overdue bank credit indicates softness in the banking sys- tem. Similar problems of interpretation arise when examining trends in overdue bank credit. Our survey indicates that the practice of rolling over overdue bank credit is very common indeed in Russia (see table 5-3). Close to half of the firms in the survey report they have had trouble repaying or servicing a bank loan in the previous two years. In nearly all cases, the outcome was either capitalization of interest or rescheduling of principal.4 We note that seizure of collateral, or indeed any legal action, did not occur in a single case. Collateralization may be very common, but it is apparently very ineffective. Firms, Banks, anid Credit in Riussia 149 Table 5-3. Bad Debt and the Rollover Problem: "Have You, in the Past Two Years, Failed to Repay or Service a Bank Debt on Time?" Response Numtber offirms (total = 439) Yes 203 What happened Capitalization of interest 101 Rescheduling of principal 85 Write-off of part of principal or interest due 3 Legal action 0 Other 12 Charged fines or penalty interest 10 Not available 3 No, always repaid or serviced on time 153 No, not applicable, never received a bank loan 80 Not available 3 Note: One firm is double-counted and reported both capitalization and rescheduling. Nevertheless, data on overdue bank credit can be informative, and we now consider the information supplied by the firms in the survey on their overdue liabilities to banks (table 5-4). About one-third of the firms in our sample have some overdue liabilities to banks. According to Goskomstat, an average of between 10 percent and 15 percent of bank debt held by in- dustrial firms was overdue in 1994. In our sample the figure is consider- ably higher, at 28 percent, and is much closer to the CBR figure for overdue credit in mid-1994 (22 percent). Again, the most likely reason for the difference is that Goskomstat's data do not include interest arrears, whereas the firms in our sample do.5 The term structure of overdue bank liabilities shows that most over- due bank debt is short-term; the bulk of arrears to banks are overdue less than three months, and little is overdue more than one year (table 5-4). This reflects the rolling-over practice noted above, practiced either be- cause of financial distress (a firm gets into financial difficulties and is im- able to service its bank debt; after a time the debt is rescheduled and ceases to be overdue) or lax payment practices with respect to banks (a firm may regularly pay its bank late, but pays eventually).6 In either case, this is further evidence of softness in the Russian banking system. It is of interest here to compare these figures to those from a World Bank survey 150 Financial Aspects of Enterprise Restructiuring Table 5-4. Term Structure of Overdue Liabilities to Banks (unweighted means, in percentage) Russia survey Poland survey Statuts (mid-1994) (late 1993) Not overdue 72 85 Overdue 28 15 Overdue less than 3 months 18 2 Overdue 3-12 months 9 5 Overdue more than 1 year 1 9 Total 100 100 Note: Figures may not sum to 100 percent because of rounding. Firms with zero bank debt are excluded. of 208 Polish manufacturing firms conducted in late 1993 (Belka and oth- ers 1995), also presented in table 5-4. In the Polish survey we observe a somewhat lower proportion of overdue bank debt, only 15 percent, com- pared with 28 percent for our Russian firms. More important and more striking is the difference in term structure. Most arrears to banks in the Polish survey were overdue more than one year, and very few were over- due less than three months. In Poland, the bad debt problem is a "stock problem"; the flow of new overdue bank debt is small. In Russia, we ob- serve both a stock problem and an ongoing flow problem. A different approach to the bad debt issue that sidesteps these issues of classification of "overdue" or "bad" bank debt looks at the concentra- tion of total bank debt in "bad" firms. The idea is that if a firm is in severe financial difficulties and is unlikely to ever repay its bank debt, then we may reasonably regard the firm's bank debt as bad, regardless of how much of it is overdue, how much has been rescheduled, and so forth. We use as our set of "bad" those reporting "usually being loss-making." These financially distressed firms amount to about 10-15 percent of the total survey. (Because of the upward bias to profits caused by inflation and historical cost accounting,7 it was something of an achievement to "usually make losses" in Russia in 1994.) We want to calculate the fraction of total bank debt held by these firms and see if they account for much more of total bank debt than their numbers warrant. As explained in chapter 4, however, this comparison is complicated by the great size heterogeneity of the firms in the sample, Firms, Banks, and Credit in Russia 151 Table 5-5. Concentration of Bank Debt in Financially Distressed F:irms Signif- icance of difference Predicted between concentra- Nondistressed firmtis Distressedfirms nondis- tion in Banik credit Percentage Mean Percentage Mean tressedl distressed (normalized) n of sample Median of sample Median distressed firms BC/sales (percent) 182 83.5 16.5 Mean 6.27 15.1 ** 32.2 Median 6.03 22.0 ** 41.9 BC/L (million rubles) 177 83.6 16.4 Mean 0.297 0.391 20.5 Median 0.347 0.500 22.0 iote: Significance levels: * = 5 percent level; I = 1 percent level; *** = 0.1 percent level. No star = not significant at 5 percent level. Means: t-tests of log values. Medians: Wilcoxon rank-sum tests of log values. and so we adopt the procedure used in that chapter. We normalize bank credit by a size variable (for example, sales), and then calculate the con- centration of normalized bank credit in financially distressed firms. As a check for the robustness of our results, we calculate four sets of results: first using the mean of normalized bank credit, and then again using the median; we do this first normalizing by sales and then normalizing by employment. The statistical tests of whether bank debt is concentrated in financially distressed firms are tests of the significance of the difference between the mean (median) levels of normalized bank credit in the dlis- tressed and nondistressed samples. The analysis above (including the sta- tistical tests) is limited to firms with some bank debt; firms with zero bank debt are excluded (for more details, see chapter 4). The results are presented in table 5-5. We find some evidence that bank debt is concentrated in financially distressed firms. These firms make up 16 percent of all firms with bank credit. When we normalize bank credit by sales, we find that these firms would account for 32-42 percent of total bank credit: that is, double their 152 Financial Aspects of Enterprise Restrtcturinig weight or more. The large holdings of bank credit in these firms in rela- tion to their sales is statistically significant at the 1 percent level, both for means and medians. When we normalize by employment rather than sales, however, we do not find any concentration of bank debt in dis- tressed firms, and the bank credit/employment ratio in these firms does not differ in a statistically significant manner from nondistressed firms. We note here that Alfandari and Schaffer report (see chapter 4) that what is significantly concentrated in these distressed firms is tax arrears. We note also that they present results showing that financial distress in the surveyed firms demonstrates a statistically significant correlation with both the presence and the volume of overdue bank debt. This last finding suggests that data on overdue bank debt may indeed be a useful indica- tor of the scale and trends in the bad debt problem in Russia, despite the rollover problem. We note that studies for other transition countries (Bonin and Schaffer 1995 for Hungary; Gomulka 1994 for Poland; World Bank 1993a for Ro- mania) have found high degrees of concentration of bank debt in "bad" firms. Our results on concentration are not as strong, and this leads us to ask whether one reason for this may be that perhaps a large bad debt problem is still to emerge fully in Russia, and that the frequency of roll- overs and the like noted above reflects both a general problem of a soft banking system and truly bad debt held by distressed firms. One reason that the bad debt problem may not yet have fully emerged is that subsi- dized credit with highly negative real interest rates in the past may cause the volume of total bank debt in financially distressed firms to grow less quickly, or even to fall in real terms. This would suggest that the full bad debt problem may emerge only after subsidized credits have been phased out for some time and real interest rates have risen. Because overdue bank debt may be a useful indicator of genuinely bad debt, it is of interest to look at the characteristics of firms that hold this debt. We do this by estimating a series of logit regressions in which the dependent variable is a 1/0 variable, = 1 if the firm has a "significant" volume of overdue bank debt (defined as greater than 2 percent of an- nualized sales), = 0 if not; about 12 percent of the sample has a significant volume of overdue bank debt thus defined. We run two logits for each characteristic we investigate: in the first the only independent variable is the characteristic of interest; in the second we re-run the regression, add- Firnms, Banks, and Credit in Rnssia 153 Table 5-6. Characteristics of Firms with Significant Amounts of Overdue Bank Credit Itenm Correlation Size (log employment) Financial distress (yes/no) ("Usually a loss-maker") 4 Overdue payables to suppliers as percenitage of sales ++ Overdue tax payables as percentage of sales + Arrears to workers (wage arrears) as percentage of sales Change in real sales 1993-94.H1 (deflated nominal sales) - -- Shareholder in bank (yes/no) 0 0 Major shareholder in bank (yes/no) 0 0 Nate: Significant amounts of overdue bank credit: in excess of 2 percent of annuclized sales. Factor is associated firm holding (+) / not holding (-) overdue bank credit. Results come in pairs. The first result is the significance level of the independent variable in a simple logit procedure with the presence of bank credit as the dependent variable. The second result is the significance level of the independent variable after adding indatstry dummies and a size variable (log employment). ++ = Significant at 1 percent level, positive coefficient. + = Significant at 5 percent level, positive coefficient. 0 = Statistically insignifi- cant at 5 percent level. - = Significant at 5 percent level, negative coefficient. - - = Significant at 1 percent level, negative coefficient. A small number of outliers with very high overdue pavables of different categories were excluded from the tests involving these variables. ing industry dummies and a size variable (log employment). The resualts are reported in table 5-6. The picture that emerges is one of large, financially troubled finns. Firms with significant amotmts of overdue bank credit tend to be large, to usually be loss-makers, to have arrears toward other creditors (suppliers, the tax collector, and their own workers), and to have experienced recent output falls that were larger than average. Shareholding in the lending bank does not appear to be correlated with holding overdue bank debt. Variation across branches (results not reported here) also shows no corre- lation; these firms are roughly evenly scattered across industrial sectors. 154 Financial Aspects of Enterprise Restructhring Bank Credit Supply and Demand Firms in the survey were asked about the ease of obtaining short-term and long-term bank credit (see table 5-7). Two-thirds of the sample said they found it very easy or fairly easy to obtain short-term credit. The situ- ation with long-term bank credit was quite different: about three-quarters said they found it difficult or impossible to obtain long-term loans. Again, there was relatively little variation by ownership; de novo firms found it as easy, and as difficult, to obtain bank loans as the rest. By far the most common problem in obtaining bank loans, according to firms, was the cost of the loan (see table 5-8). Other common problems were banks' unwillingness to lend in a period of high inflation (long-term loans only), the financial situation of the firm, and inadequate collateral. Surprisingly, de novo firms were not faced with this last problem much more often than other firms-19 percent said inadequate collateral was a problem in obtaining short-term loans, compared with 14 percent for the total sample, and there was virtually no difference at all in long-term loans. The response of firms to these perceived obstacles is twofold. First, they say they are relying primarily on retained earnings to fund any fixed investment (just as Western firms usually do, by the way); and second, they continue to hope for (subsidized) state credits for their investment needs. Thus, of the firms intending to engage in fixed investment, 71 per- cent said they would use retained earnings to finance it in whole or in part, compared with 41 percent who said they would use DSCs, and only 30 percent who would use commercial bank credit. Similarly, firms' most Table 5-7. Ease of Obtaining Bank Credit on Commercial Terms (percet7tage offirtns) Short-term bank credit L7ong-term72 bank credit ResponIse o07 conmmercial terms on7 comm1lercial term7s Very easy 34 12 Fairly easy 33 12 Fairly difficult 15 19 Very difficult 8 28 Impossible 6 26 Other 2 3 Firms, Banks, and Credit in Ruissia 155 Table 5-8. Main Problems in Obtaining Bank Loans (percentage offirms citingfactor) Short-term Long-term Problem loans loans Banks don't have enough money 10 15 Size of loans too small 7 6 Poor financial situation of firm 16 15 High interest rates, loans too expensive 73 60 Lenders prefer clients with large deposits 5 7 Banks prefer better-known, established clients 3 3 Banks unwilling to lend because of high inflation 6 23 Inadequate collateral 14 16 Bank rejects investment projects 2 3 Lack of personal contacts with bank managers 2 2 Regulations and paperwork 6 7 Other 3 4 Hard to say 10 11 No problems in obtaining bank loans 4 1 commonly cited obstacle to fixed investment (after high interest rates; that is, the cost of credit) was their inability to obtain DSCs. The nrost commonly requested form of government assistance (after tax breaks) was credit on preferential terms. We turn now to an analysis of the factors affecting the supply of credit. It is possible to identify the supply curve for credit through the survey data because we have firms' estimates of the ease of obtain:ng credit. We follow Pinto and Van Wijnbergen (1994) by using an ordered logit procedure, with firms' estimates of the ease of obtaining credit as the dependent variable. We estimate the equation with a variety of inde- pendent variables, each tried separately. In each case, we estimate the equation first with the single independent variable on its own, and then again, adding industry dummies and a size variable (log employment) to control for size effects (bargaining power, for example). The results are summarized in table 5-9. We find some evidence that creditworthiness of the borrower is a fac- tor determining the supply of credit; the banking system has some hard- ness in it. Firms in financial distress (chronic loss-makers) do find it significantly more difficult to obtain short-term bank credit, even when 156 Financial Aspects of Enterprise Restructitring controlling for size and industry, but they find it no tougher to obtain long-term bank credit than other firms. (We use this "financial distress" indicator rather than the profitability figures reported by firms because of severe data problems with the latter.) The volume of overdue bank credit (overdue credit as a percentage of sales) shows no connection to the sup- ply of credit. It appears that this is because what does matter is the pres- ence of overdue bank credit (the actual volume seems not to matter). Firms with significant (greater than 2 percent of sales) overdue bank credit do find it more difficult to obtain short-term bank credit. The most important financial variable affecting the ease of obtaining both short- term and long-terrrm bank credit is past history of payments difficulties- firms that failed to service or repay bank debt in the past find it significantly more difficult to obtain either short-term or long-term bank credit now. We noted earlier that firms were commonly part-owners of their lend- ing banks. Somewhat surprisingly, this apparently does not translate into greater ease in obtaining bank credit. Being a shareholder-even a major shareholder-does not significantly correlate with a greater ease in ob- taining either kind of bank credit (in one case the relationship is perverse: when we control for industry and firrm size, we find that shareholding firms find it more difficult to obtain long-term bank credit, not easier). This lack of effective influence is also reflected in the actual volumes of borrowings and in the interest rates paid. Shareholders do not have larger bank loans (in relation to sales, total liabilities, or employment) compared with nonshareholders, nor are they granted more favorable repayment terms. Similarly, a bank holding shares in the firm is not associated with easier access to credit (again, cross-holding is, if anything, associated with more difficult access). There is some weak evidence that firms in which nonbank domestic financial institutions hold shares have easier access to short-term credit. The causality is unclear, however-the evidence is con- sistent with investment funds placing holdings in creditworthy firms as well as with possible (indirectly) connected lending. Larger firms do not have an advantage in garnering loans; log em- ployment is not correlated with easier bank credit. Past access to bank credit is also not correlated with current access. With respect to owner- ship groups, there is some indication that state-owned firms find it easier to obtain long-term bank credit when we control for size and industry. One possibility is that this may reflect banks' assessment of the long-term Firms, Banks, and Credit in Ruissia 157 riskiness of investing in private or privatized firms, where the results of expected restructuring are difficult to predict; another is that state-owned firms have better access to (implicit or explicit) government guarantees on their lending. De novo firms, by contrast, find it no more (or less) .iffi- cult than privatized firms to obtain bank credit, an encouraging finding. Identification of the demand for credit is more problematic. We limit ourselves here to looking for signs that adverse selection is operating, and bad firms rather than good firms are taking up bank credit. We use a simple approach and ask which factors are associated with firms holding any bank credit at all. The analysis is similar to the earlier strategy: we run two sets of logit regressions of whether the firm has any bank credit (1 = yes, 0 = no) on characteristics of interest, first without, and then with, industry dummies and a size variable. The results are reported in table 5-10. The findings suggest that firms that do hold bank credit are nmore troubled than the average. They tend to be larger; more often loss-nmak- ing; with larger overdue payables to suppliers, government (tax arrears), and their own workers; and with larger-than-average recent declines in output. There is no correlation with ownership-in particular, de novo firms are no more unlikely (or likely) to have a bank loan, confirming our earlier observations. These results are robust to inclusion of industry and size variables in the regressions. The last finding about new private firms aside, the results are worrisome and suggest that adverse selection may be operating in the Russian bank credit market-too many "bad" films are borrowers. Banks' Influence on Enterprise Decisions In this section we examine briefly, from the perspective of the enterprise managers, whether banks influence enterprise decisions-and if they (0o, in which areas-either as creditors or as shareholders. When managers were asked to rate the influence of different actors, including banks, on various categories of enterprise-level decisions (operational, employ- ment-related, and financial), relatively few enterprises indicated that banks had any major influence. Management (according to managers) overwhelmingly dominates enterprises across all kinds of decisions. Nevertheless, a significant 23 percent of enterprises responding indi- cated that banks had a moderate or great influence in decisions concern- 158 Financial Aspects of Enterprise Restrtuctitring Table 5-9. Factors Affecting the Supply of Bank Credit, Ordered Logit Results Shlort-term Long-termi Factor banik credit bank credit Financial distress (yes/no) ('Usually a loss-maker") -- 0 _ 0 Overdue bank credit as a percentage of sales 0 0 o 0 Overdue bank credit greater than 2 percent of annual - 0 sales (yes/no) - 0 Failed to service or repay bank loan on time in last two - - -- years (yes/no) - Shareholder in bank (yes /no) 0 0 0 _ Major shareholder in bank (yes/no) 0 0 o o A bank holds shares in the firm (yes/no) 0 0 0 _ A nonbank domestic financial institution holds shares 0 0 in the firm (yes/no) + 0 Bank exerts influence on enterprise decisions' 0 0 o o Obtained a bank loan in the last two years (yes/no) 0 0 o 0 Size (log employment) 0 0 o o Ownership group (relative to "privatized," 64 percent of sample) State-owned (26 percent of sample) 0 0 O + De niovo (newly established private firm, 11 percent 0 0 of sample) 0 0 Note: Factor is associated with bank credit being easier (+) / harder (-) to obtain. Results come in pairs. The first result is the significance level of the independent variable in a simple ordered logit procedure with the ease of obtaining bank credit as the dependent variable. The second result is the significance level of the independent variable after adding industry dummies and a size variable (log employment). ++ = Significant at 1 percent level, positive coefficient. + = Significant at 5 percent level, positive coefficient. 0 = Statistically in- significant at 5 percent level. - = Significant at 5 percent level, negative coefficient. - - = Sig- nificant at 1 percent level, negative coefficient. A small number of outliers with very high overdue bank credit/sales ratios were excluded from the tests involving this variable. a. See section entitled "Banks' Influence on Enterprise Decisions." Firmis, Banks, anid Credit in Ruissia 159 Table 5-10. Which Firms Hold Bank Credit? Factor Correlation Size (log employment) ++ ++ Financial distress (yes/no) ("Usually a loss-maker") ++ ++ Overdue payables to suppliers as a percentage of sales ++ + Overdue tax payables as a percentage of sales ++ ++ Arrears to workers (wage arrears) as a percentage of sales ++ . ~~~~~~~~~~~~++ Log change in real output 1993-94.H1 - - Ownership group (relative to "privatized," 64 percent of sample) State-owned (26 percent of sample) 0 0 De novo (newly established private firm, 11 percent 0 of sample) 0 Note: Factor is associated with bank credit being held (+) / not being held (-). Results come in pairs. The first result is the significance level of the independent variable in a simple logit procedure with the presence of bank credit as the dependent variable. The second result is the significance level of the independent variable after adding industry dummies and a size variable (log employment). ++ Significant at I percent level, positive co- efficient. + Significant at 5 percent level, positive coefficient. 0 Statistically insignificant at 5 percent level. - Significant at 5 percent level, negative coefficient. - - Significant at 1 percent level, negative coefficient. A small number of outliers with very high overdue payables of different categories were excluded from the tests involving these variables. ing sales or production, and about 17 percent noted that banks had some influence in financial decisions generally. Fewer than 10 percent of the en- terprises indicated influence of banks in employment-related decisions, with little difference between kinds of employment (management or workers). The relatively high frequency of bank influence on enterprise decisions regarding sales and production is interesting. The importance banks attach to the sales and production decisions of enterprises may be a reflection of the short-term nature of the majority of bank loans. There are no significant differences in bank influence among owner- ship groups, with the exception of de novo enterprises (see table 5-11). A Table 5-11. Bank Influence on Decisions of Enterprises (percent) Sales, production, Worker employment Profits, investments, marketing, operations and benefits Management employment financial issues Factor No inifluence Some influence No influence Some influtence No influence Some influence No influence Sonme influence State-owned 84 16 94 6 94 6 83 17 Privatized 75 25 92 8 93 7 84 16 De novo private 73 27 86 14 86 14 78 22 L S 200 77 23 91 9 91 9 81 19 200f4& 304] 32 33 [317] Pearson correlationsb (The following variables are expressed as a percentage of salesc) Sales to the government n.a. 0.02 0.04 0.01 (0.69) (0.43) (0.88) [345] [3311 [299] Nonmilitary sales to government 0.11i 0.06 0 0.01 -0.01 0.01 or budget organizations (0) (0.39) (0.93) (0391) (0.98) p[2711 [239] 1216 n.a. [34213139 Military production 0 0.03 -0.013.0.05 0.04 0.1 (0.94) (0.65) (0.11 (0.32) (0 44) (03) (271] [239[2[ [343] {$33 i [299] Revenue from military goods 0.03 0.01 0.12 0.03 0.17 0 purchased by the state as a (0.95) (0.82) . 00)(0.56) QJ percentage of total revenues d [312] 12521 29 [348] 31] 11 Average wage, current year 0 -0.02 -0.05 0.0 WI0.07 0.07 0.04 -0.03 (0.95) (0.80) (0.40) (0.11) .0). (0.22) (0.18) (0-44) (0.63) [3121 [254] [241] [339] 32] [287] [340] 30 29 Output growth in volume over -0 0 -0.02 -0.03 -005 -0.02 0.08 01 ~ 1 the previous year (0.96) (0.96) (0.84) (0.71) (0.50) (0.74) (0.28) (006) [158] [158) [155] [194] [186] [1831 [196) [1921 n.a. Not available. Note: Degree of significance is listed in parentheses, number of observations in brackets. a. First line: sign of correlation. b. First line: coefficient of correlation. c. 1994.H1 sales for 1994 transfers; 1990-91 sales for 1992 transfers. d. 1994.111 Ieve.tues for 1994 transfers, 1990 revenues for 1992 transfers. 2 Significant correlation at a 10 percent confidence level. m Sigrnificant correlation at a 5 percent confidence level. 192 Finianicial Aspects of Enterprise Restrzcturinig Figure 6-5. Concentration of Financial Flows According to Enterprise Size Percentage of recipients Volume index: sample average 100 80 Probability of being Total transfers per 300 recipient employee in 1994 70 (ethnscl)(right-hand scale) 50 F7 1994 ~~~~~~~~~~~~~200 40 30 100 20 10 0 0 0-100 101-350 351-800 801-1,500 1,501 5,001- 20,001- Total 5,000 20,000 100,000 Enterprise size measured by number of employees First, employment appears to be one of the main variables determin- ing the amount of transfers, as well as the probability of being a recipient. For enterprises with over 5,000 employees, the chance of being a recipient is more than one in two (see figure 6-5). Smaller enterprises have a much lower chance of receiving transfers. The volume of subsidies received per employee, however, is positively correlated with employment size for small and medium-size enterprises only (here defined as those with fewer than 1,500 employees). The largest enterprises received about 50 percent of the average transfers per employee for the overall sample in both 1993 and 1994 (table 6-2). It is equally interesting to note that these relationships lose most of their significance if the employment level is re- placed by output or by the volume of sales as the size indicator. This sug- gests that workers, more than the firm itself, are the real concern of the authorities in their decision to provide transfers, giving extra lobbying leverage to the enterprises that are the largest in employment. Governmient Financial Transfers to bndustrial Enterprises and Restrtuctturing 193 Even enterprises that are not very large in size but are in the same sec- tor can collude to increase their leverage on government authorities to extract funds (see Perotti 1994, in particular). This explains why participa- tion in industrial associations substantially increases enterprises' chances of becoming recipients. While for the sample as a whole, the average par- ticipation rate in associations is about 48 percent, for the recipient group it is almost 60 percent (see table 6-8). We further explore if enterprises' market power increases their ability to extract large volumes of govern- ment funds. Indeed, while 24 percent of all enterprises in the sample rec- ognized that they do not have competitors, 29 percent of those in the recipient group and 22 percent of nonrecipients indicated that they do not face market competition. In addition, nonrecipients estimate their aver- age number of competitors at twenty-five, while for recipients, it is twelve (see table 6-6). The likelihood-ratio Chi2 test of independence, however, does not significantly reject the null hypothesis. This might be because collusion is less effective in industries with nonhomogenous products, including manufacturing and defense-related industries. The number of enterprises in an industry is not as good a predictor of its jmar- ket power as the product itself (this is well described in Aslund 1995). General rent-seeking behavior is also exhibited by enterprise manag- ers. Almost 80 percent of the managers in the sample expressed an inten- tion to invest in the medium term, including 90 percent of the recipient-enterprises. The explanation of the latter phenomena is prob- ably institutional. The easiest way to extract government transfers in ]Rus- sia today is to design a large-scale investment program and to request federal funds for its implementation, arguing that the program is of na- tional importance. This does not mean that such intentions will be real- ized later, however, because transfers received might be reallocated, for noninvestment purposes. Table 6-8 shows that the Chi2 test rejects the null hypothesis of independence between "being a recipient" and "have an investment plan." Table 6-10 tells us what remains after the impact captured by owner- ship and sectoral dummies. We see that the compensatory factors-gov- ernment procurement and social benefits-have odds ratios significa:ntly close to 1. This means that enterprises within the same sector and of the same ownership status do not have a greater chance to get subsidies Sim- ply because of these compensatory factors. Not surprisingly, membership in industrial associations and the size effect are completely captured by 194 Financial Aspects of Enterprise Restructtring Table 6-10. Explained Variable: "Being a Recipient in 1994" Explanatory variable Odds ratio Government procurement 1.01** Social benefits 1.02* Military production 4.60* Industrial association N.S. Labor force (size) N.S. Itivestment plan 9.56*** Each of six owvnership categories > 1,000,000*** Sectors: heavy machinern, tool enginieerinig, agroniachinery 70-95** Sectors:fiels, MIC, metalwvorking 36-49* All other sectors N.S. Note: This table presents the results of a logistic regression on "being a recipient in 1994." The selected explanatory variables, taken individually, were significant at a 5 percent level, as shown in table 6-7. We still control for ownership and sectors. Explanatory dummy vari- ables are in italic. Results from the maximum likelihood estimation: Chi2(25) = 77.64; num- ber of observations = 190; Prob > Chi2: 0.000; Pseudo R : 0.326. *** = significant at a 0.5 percent level; ** = significant at a 5 percent level; = significant at a 10 percent level; N.S. not significant at a 10 percent level. sectoral and ownership dummies. A higher share of military production in total sales (although the significance level associated with this variable is rather low) and the existence of at least one investment program do strongly increase the probability of being a recipient. As we have seen earlier, there is no correlation between "being a recipient in 1994" and either the existence of a military conversion program or the amount of ac- tual investment spending. These results together suggest that forced sub- sidies are the prevailing category. Government Financial Transfers and Enterprise Performance We now turn to the relationship between government financial transfers and enterprise performance. Given some of the data problems discussed earlier and the lack of time series for variables measuring enterprise per- formance, we will not .be able to examine the causality issue. Instead, we will simply look at cross-section correlations at a given point in time. Aware of the simultaneity involved in this exercise, we will not be able to answer the question of whether subsidies foster or hamper the restructur- Government Finiancial Transfers to Indutstrial Enterprises anid Restrctulring 195 ing process. Instead, our emphasis will be on documenting the cdiffer- ences in performance between recipients and nonrecipients and within the subset of recipients. Not surprisingly, our data suggest that financial transfers are concen- trated in enterprises with the worst financial perfornances. It is true that the proportion of loss-makers in the group of recipients (10 percent) is smaller than among nonrecipients (15 percent), and that the proportion of recipients among firms in financial distress7 (24 percent in 1993) is smaller than for all other firms (36 percent) (see table 6-6). Nevertheless, because financial transfers are part of the enterprise's cash flows, positive profits that appear on balance sheets might actually be negative when net of all subsidies.8 Thus, many recipiefits in the profit-maker group should instead be considered like loss-makers. Indeed, table 6-8 shows that the costs/sales ratio, which can be read as a proxy for profitability (of oppo- site sign), is highly negatively correlated with the probability of being a recipient. This result holds after controlling for ownership and sector. Our results suggest that firms that adjust less were among the largest recipients-ex post they received more transfers than recipients thai. ad- justed more quickly. The data for the first half of 1994 are consistent with the hypothesis that subsidies helped them maintain output and employ- ment levels without making the needed adjustments. Table 6-9 shows that output decline in 1994 is negatively correlated with financial trans- fers, suggesting that 1994 subsidies might have helped support industrial activity. We also observe relatively fewer layoffs in the recipient enter- prises than in the rest of the sample. In addition, there is some indication that government transfers may have been used for additional wage in- creases. On average, enterprises within the sample preserved the level of their 1993 real wages in 1994, despite additional declines in sales. Al- though the recipients had lower wages in both years and a significantly lower wage growth in 1994 (see table 6-6), table 6-8 shows that the shiare of wage bonus in total cost is positively correlated with the probability of being a recipient. Therefore, it is not surprising that figure 6-6 shows ithat changes in the nominal labor productivity of recipients is about three times worse than that of nonrecipients. Our interpretation is that al- though large transfers have helped decrease the scale of the output drop to some extent, they have also encouraged labor hoarding and have re- sulted, ex post, in diminishing labor productivity, at least in real terms. 196 Financial Aspects of Enterprise Restructutring Figure 6-6. Labor Productivity inoniinal termss) Millions of rubles per employee 20 18- M 1992 m 1993 16 - m 1994 14- 12 - 10 8- 4- 2 Group 1 Group 2 Group 3 Group 4 Total Group 1 = nonrecipients Group 2 = recipients providing qualitative data only GrolLp 3 = recipients providing quantitative data Group 4 = nonrespondents This last hypothesis is further confirmed when one looks at the small group of recipients that are the largest enterprises. These firms, although faced with a sharp decline in their output level, have not reduced their employment. This subgroup of recipients constitutes the core of enter- prises facing real restructuring problems, and they have received sub- stantial government assistance. Unfortunately, these enterprises did not provide quantitative information about the amount of subsidies obtained. All these results show that recipient-enterprises, especially the largest recipient-enterprises, adjusted much less than the nonrecipient group. This lack of adjustment, together with the failure of 1992-94 government transfers to lead to an increase in enterprise investment activity, which in- stead went primarily to finance recurrent costs, particularly to protect wages, suggests that subsidies may have contributed to slowing down, rather than facilitating, restructuring by the recipient enterprises. Governmenti Financial Transfers to Induistrial Enterprises and Restrictutring 197 Conclusions and Policy Recommendations By the middle of 1994, after two-and-half years of market reforms, the Russian enterprise sector as a whole still faced a rather soft budget con- straint. The number of subsidy-free enterprises was very small. About one-third of enterprises in the survey received some government assis- tance in 1993, and about one-quarter received such support in the first half of 1994. Only 20 percent percent of the enterprises in the sample re- ported that they neither received government transfers nor accumulated tax arrears. Even these numbers may underestimate the actual scale of government financial assistance to enterprises because of some of the downward biases in the survey data discussed earlier. Transfers are highly concentrated: 1.5 percent of the enterprises within the sample received about half of total transfers. Large and extra-large en- terprises, as a rule, are the greatest recipients of government funds. The average size of transfers is relatively small, however, and it has been sub- stantially reduced since 1992. For the largest recipients (7-8 percent of the sample), the size of transfers is about 6 percent of their output. The bulk of recipients receive even smaller transfers. Government transfers of such a size do not provide recipients with the necessary funds for a genuine restructuring. Instead, they are mainly used to finance current operations and losses. Our analysis shows that one of the most important factors determining whether an enterprise receives government transfers, and how much these transfers might be, is the size of its employment. Overall, gov- ernment financial transfers do not seem to serve any other identifiable ob- jectives (such as promoting restructuring, high-tech, and export industries). The system of transfers is composed of about a dozen independent channels, which makes government assistance even less efficient and more difficult to monitor. The most explicit transfers, such as budget sub- sidies and investment grants, amount to a very small portion of the total transfer flows; far more funds are disbursed in the forms of tax exemp- tions and directed credits. While explicit government transfers have been reduced significantly, tax arrears have become a main source of soft enterprise financing since mid-1993. Even enterprises that do not receive explicit government transfers seem to operate in a rather soft financial environment. They can extract funds in the form of tax arrears and arrears to their suppliers because of poor tax administration, weak procedures of debt recovery, and underdeveloped bankruptcy procedures. 198 Financial Aspects of Enterprise Restrtctutring The analysis of the factors determining transfer allocation revealed that, on average, government assistance does not go to enterprises that face the most severe financial problems, or those particularly affected by recent external shocks, such as losses of Eastern European and CIS mar- kets. Transfers are also not biased toward enterprises with more intensive investment spending. There is some indication that selected forms of transfers play a slightly compensatory role for price controls and/or gov- ernment procurement. Overall, however, the distribution and the level of transfers are related consistently to the size of employment and enter- prise membership in industrial associations. In other words, the main al- location criterion of government financial transfers is the bargaining power of the enterprises. The current level and system of government financial transfers clearly have not promoted enterprise restructuring, especially for the largest re- cipients. In order to encourage restructuring of the enterprise sector, fur- ther reduction and rationalization of government financial transfers will be required. Suggestions are listed below. D Deepening price and trade liberalization and reducing the total volume of transfers. Substantial reductions in transfers can be achieved through further liberalization efforts. This will reduce the need for compensatory subsidies and, at the same time, reduce dis- tortions in the economy. Medium-term liberalization objectives might include eliminating the remaining price and margin con- trols, especially at the local level; completing the divestiture of so- cial assets; and implementing market prices and a competitive framework for state procurement. Although most of this task had already been achieved at the federal level in 1995, much remains to be done by the local governments. - Reducing the number of channels. There are currently numerous channels through which government financial assistance is pro- vided to enterprises. Rationalizing transfers requires a substantial consolidation of these channels, as well as a transition to transpar- ent and explicit forms of allocation. In principle, all government fi- nancial assistance to enterprises should be channeled through the budget (as subsidies, investment grants, or loans). This will en- hance the government's ability to assess the overall fiscal costs of maintaining such transfers and to monitor the use of transfers more easily. Government FinTancial Tranisfers to Industrial Enterprises anid Restructutring 199 . Reducing the number of agencies involved in the allocation of transfers. Corresponding to the multitude of transfer chamnels, there are currently a large number of government and quasi-gov- ernment agencies involved in the allocation of financial assistance to enterprises, without effective coordination. There is clearly a need to reduce the number of agencies involved in the allocation of transfers to enterprises and to clarify responsibilities for allocation and monitoring among the core economic agencies. . Improving transparency of the allocation process. The current system of government financial transfers to enterprises lacks any transparent procedures and criteria for allocation. Even when crite- ria do exist, our analysis shows that they are not implemented con- sistently, if at all. The objectives of the transfers should be transparent, nondiscretionary, and known to the public. * Conditioning remaining transfers on restructuring. The lack of transparent procedures and criteria has also meant that most trans- fers do not have clearly stated purposes or objectives, and no meaningful conditions are attached to them. In order to make a continued reduction in the government financial transfers to the largest recipients sustainable, transfers should be clearly targeted for specific purposes (such as the divestiture of social services by enterprises or severance pay for workers) and conditional on timle- bound and measurable actions by the recipient enterprises toward restructuring. * Improving monitoring and supervision over the use of transfers. For the transfers to reach their intended recipients and achieve their stated purposes, an effective system of monitoring and super- vision of their use must be put in place. These mechanisms shouLld be relatively simple and focused on regular accounting of the transfer flow that has been disbursed and spent, and also on whether the enterprises have complied with conditions attached to the trans- fers. Penalties should be imposed on the misuse of transfers. Appendix Measures of Association and Qualitative Variables Association measures between binary variables (between "being a recipi- ent for a given year" and an explanatory dummy variable, for example) 200 Finanicial Aspects of Enterprise Restnrcturing are assessed with the likelihood-ratio Chi2 independence test (at one de- gree of freedom). The number in parentheses is the significance level at which the null hypothesis that variables are independent is rejected. Therefore, a small figure will mean that qualitative variables are signifi- cantly correlated. The sign indicates the impact on the marginal prob- ability: for instance, if it is positive, then the probability of being a recipient, given that the explanatory variable equals 1, will be greater than the unconditional probability of being a recipient. In the case of the Ozvnership variable (table 6-5), an ordered multival- ued explanatory variable (taking integer values from 1 to 6, increasing with the degree of independence from the state), we have added another test: the Kendall's tau-b. This measure of association accounts for the ranking of the explanatory variable and takes values between -1 and 1. In the specific case of ozun?ership, the measured Chi2 has 5 degrees of freedom. To test the correlation between "being a recipient for a given year" and any continuous explanatory variable, we use the t-test on means and test the null hypothesis that the means of the two subsamples-recipi- ent/nonrecipient-are equal. Finally, a simple Pearson correlation gives the measure of association between any two quantitative (continuous) variables. In table 6-9, we pre- sent the correlation coefficient, the significance level in parenthesis (a 0.00 value would mean that the correlation is significant at a 0.5 percent confi- dence level), and the number of observations in brackets. All calculations have been computed with a Stata-version 3.1 software. References Alfandari, Gilles. 1995. "Financing Industrial Restructuring." Cuadernos del Este 14: 35-52. Aslund, Anders. 1995. How Rtussia Became a Market Economy. Washington, D.C.: The Brookings Institution. Aukutsionek, Sergei, and Elena Belyanova. 1994. "Do Budget Constraints of Russian Enterprises Become Harder?" MEIMO 7: 124-43. Balcerowicz, Leszek, and Alan Gelb. 1994. "Macropolices in Transition to a Market Economy: A Three-Year Perspective." In Proceedings of the World Bank Annuial Conference on Development Economics. Washing- ton, D.C. Government Finantcial Tracnsfers to Indutstrial Entterprises anid Restrtctuirinig 201 Boycko, Maxim, Andrei Shleifer, and Robert Vishny. 1995. Privat;izing Russia. Cambridge, Mass.: MIT Press. Claessens, Stijn, and Kyle Peters. 1996. "Enterprise Performance and Soft Budget Constraints: The Case of Bulgaria." East Asia and Pacific Region, The World Bank, Washington, D.C. Photocopy. Delyagin, Michael, and Lev Freinkman. 1993. "Extrabudgetary Func[s in Russian Public Finance." RFE/RL Research Report 2 (48): 49-54. Fan, Qimiao, and Mark E. Schaffer. 1994. "Government Financial Trans- fers and Enterprise Adjustments in Russia, with Comparisons to Central and Eastern Europe." Economics of Transition 2 (2): 151-88. Freinkman, Lev. 1994. "Government Financial Transfers to the Enterprise Sector in Russia: General Trends and Influence on Country Macroeconomic Performance." Paper presented at the IIASA C'on- ference on Russian Output Trends, June 11-12, Vienna. Kornai, Janos. 1986. "The Soft Budget Constraint." Kyklos 39 (Fasc.1). Maddala, G. S. 1983. Limited-dependent and Qutalitative Variables in Econo- metrics. Cambridge, U.K.: Cambridge University Press. Perotti, Enrico. 1994. "Collusive Arrears in Transition Economies." Finan- cial Markets Group Discussion Paper No. 198, London School of Economics. Schaffer, Mark E. 1995. "Government Subsidies to Enterprises in Central and Eastern Europe: Budgetary Subsidies and Tax Arrears.' In David M. G. Newbery, ed., Tax and Benefit Reform in Central and Eastern Eutrope. London: CEPR. . 1990. "State-Owned Enterprises in Poland: Taxation, Subsid:iza- tion, and Competition Policies." European Economy 43: 183-201. World Bank. 1995. "Russian Federation: Towards Medium-Term Viabil- ity." Report No. 14472-RU, Washington, D.C. .1993. "Subsidies and Directed Credits to Enterprises in Russia: A Strategy for Reform." Report No. 11782-RU, Washington, D.C. Notes 1. Taking into account that only about 70 percent of respondents in the survey provided the volume of their nominal output. 2. The group of the largest recipients was determined on the basis of two com- plimentary criteria, absolute size of received transfers and relative size of trans- fers as a percentage of output. An enterprise was included in the group of large 202 Financial Aspects of Enterprise Restrtctutring recipients for 1993 if it had received either more than 1 billion rubles in 1993 (US$1 million) or had a transfer/output ratio exceeding 15 percent. For two other years, 1992 and 1994, this I billion threshold level was deflated according to the annual producer price index (PPI). About one-quarter of all recipients were classi- fied as the largest recipients. 3. Total directed credits, however, are estimated at 28 percent of total financial transfers in 1993 according to the survey, compared with 24 percent at the aggre- gate level. It is possible that, when asked, enterprises in the survey felt the need to justify the final use of the directed credits they received, and designated other kinds of directed credits as those for investment and conversion. 4. As indicated earlier, we suspect that most of the nonrespondents are recipi- ents/bad payers, unwilling to reveal themselves. 5. Please note that not all defense-related enterprises have necessarily identi- fied themselves in the survey as being members of the military-industrial com- plex (MIC). 6. The level of concentration of transfers is similar to the concentration of di- rected credits, which was considered in another recent study (Freinkman 1994). One-hundred largest recipients of directed credits allocated in 1992-93 received, depending on the credit program, from 25 to 50 percent of the total disbursed funds. At the same time, the cumulative employment of those recipients amounted to only 3-4 percent of total industrial employment. 7. Our definition of "financial distress" is the same as that in chapter 4 of this volume-firms that reported they are chronic loss-makers. 8. See Kornai (1986) and Schaffer (1990) for discussions of profits and subsi- dies in Hungarian and Polish state-owned enterprises, also relevant here. Part III Corporate Governance and Competition 7 Ownership Structures, Patterns of Control, and Enterprise Behavior in Russia John S. Earle, Saul Estrin, and Larisa L. Leshchenko We hiave created a broad basis of shareholders who have an economic interest in the suc- cess of the reform. A. Chubais, The Finiancial Times, 30 June 1994 Most enterprises conitinute to be ruin uinchallenged by old management teams, which often lack the humnan capital and interest to initiate sigiftlcant restrecturing. M. Boycko, The Fi- ntancial Times, 30 June 1994 According to The Financial Times (27 June 1994, p. 3), Russia's mass priva- tization program, carried out between late 1992 and mid-1994, "sold more than 11,000 state-owned enterprises, accounting for around 70 per- The authors would like to thank Zuzana Sakova for dedicated research assis- tance, and Mark Schaffer for detailed comments, which led to many improve- ments in the chapter. Helpful comments were also received from David Bernstein, Joseph Blasi, Simon Commander, Randall Filer, Alan Gelb, Mike Mv[ar- rese, Mario Nuti, Judith Sedaitis, and participants at seminars at the World Bank, CERGE, Leontief Center in St. Petersburg, and the Center for International Secu- rity and Arms Control at Stanford University (CISAC). The World Bank, the La- bor Research Program of the Central European University Privatization Proj,ect, and CISAC provided financial and logistical support. 205 206 Corporate Governance and Competition cent of Russian industry, in exchange for cash and 148m freely distrib- uted vouchers." From a very low level in 1992, employment in the private sector is estimated to have grown to around 50 percent of the labor force (European Bank for Reconstruction and Development, cited in the Initer- inational Herald Tribunle). It is thus not surprising that Russia's pro-reform politicians, as well as some Western analysts (see, for example, Lieber- man and Nellis 1994), have hailed the program as a success. But for many observers the speed of privatization has been bought at the price of suboptimal ownership structures, which may carry deleterious implica- tions for the restructuring process. There has been surprisingly little empirical analysis of which owner- ship forms have emerged, nor of the implications for the control and be- havior of formerly state-owned firms. Government sources suggest that some 40 million people, about half the labor force, have become share- holders (Reuters, 30 June 1994), and according to Professor Yasin, head of President Yeltsin's advisory economic council, "insiders own on average some 70 percent of the privatized enterprises" (reported in The Financial Tinies, 30 June 1994). Data from a sample of 142 firms collected by Blasi and Shleifer (1995) indicate that insiders held an average of some 65 per- cent of the shares in 1993. The ownership question may be crucial, for economic theory predicts differing performance, not merely depending on whether firms are privately or state-owned, but also according to whether privately owned firms are insider or outsider controlled, and whether the controlling group of insiders is made up of managers or workers (see, for example, Aghion, Blanchard, and Burgess 1994). While all the evidence suggests that it is employees who hold a majority of shares (see Blasi 1994, for example), control is usually argued to be vested primarily in the hands of senior management (see Blasi and Shleifer 1995; Commander, Dhar, and Yemtsov in chapter 2 of this volume). Neverthe- less, there has not yet been an attempt to describe the patterns of owner- ship and control, nor to analyze their impact on different areas of enterprise decisionmaking. It is these three issues-ownership structures, patterns of control, and enterprise behavior-and their interrelationships that are the subject matter of this chapter. Alternative Ownership Forms and Enterprise Behavior: Some Hypotheses Although the literature on transition has stressed that privatization is a critical component of the transition process, there have been few attempts Ownership Structutres, Patterns of Control, and Enterprise Belhavior in Rtussia 207 to evaluate the comparative strengths and weaknesses of alternative ma- jority ownership structures for the newly privatized companies. Earle and Estrin (1994) argue that the balance of advantage shifts among differ- ent ownership forms according to the problem under consideration. For example, outsider ownership may offer superior access to external capital markets, but it may also cause greater social dislocation, while worker ownership may slow employment restructuring. In this section, we pro- vide a simplified comparison of alternative majority ownership forrms in achieving four widely accepted objectives of the transition, including'1 . Developing a politically independent and market-oriented enter- prise sector, which we term "depoliticization"2 * Long-term restructuring * Short-term restructuring * Minimizing transaction costs associated with further evolution of ownership. The hypothesized impact of each ownership form, in relation to one another and against the base case of state ownership, is reported in table 7-1. The table summarizes the analysis that follows, and indicates, fo:r ex- ample, the predicted extent of depoliticization in worker-owned firms, relative both to state ownership and the other ownership forms. But a few words of caution are needed. First, the table summarizes results derived from theoretical models of 100 percent ownership by a single group. In defining our five ownership forms empirically, however, we take a ma- jority stake (or indeed the largest single stake if other holdings are diver- sified) as implying effective control of the firm. This may be misleading. In practice, the largest group of owners may have highly diversified hold- ings, while minority interests may be greatly concentrated, giving the lat- ter effective control. For example, enterprises classified as worker-owned according to ownership stake may actually be managerially controlled. We return to this issue below. There are many assumptions behind the behavior hypothesized in the table, not all of which will always be satisfied. Three cases will suffice. First, the extent of restructuring will typically be greater when procluct and factor markets are more competitive, ceteris paribus. If sectoral and. re- gional diversity is sufficient, these elements might swamp any inde- pendent ownership effects. Second, the precise institutional form of different ownership types may significantly affect behavior. Thus, fiims owned collectively by workers with limited share tradeability might be expected to perform much worse than those owned by workers on the 208 Corporate Governance anid Comipetition Table 7-1. Comparison of the Impact of Alternative Ownership Forms in Attaining Objectives of Transition Itemz WO MO 00 DN Depoliticization + + ...++ Long-term restructuring Unbundling + + ..U Investment + + ... U Intemnal organization + U Short-term restructuring Nonlabor cost minimization + + + U Labor cost minimization 0 + + U Evolution + + ... .. Note: All entries are relative to the status quo, state ownership. The denotes better; ++ denotes much better; ... denotes comparable to Western firms; U denotes not a relevant comparison; 0 denotes the same as the state sector. WO: enterprises with dominant worker stake. MO: enterprises with dominant manager stake. 00: enterprises with dominant out- sider stake. DN: newvly established privately owned enterprises. basis of individually held and freely tradable shares. Finally, the situation of the firm itself is relevant. Profitability clearly assists restructuring, re- gardless of ownership form. At the same time, collective employee own- ership might be beneficial in situations of extreme loss-making by geographically isolated firms, because such a form allows workers to trade wages for employment security. These provisos aside, the table re- ports predictions about the relative impact of alternative ownership forms on restructuring. The arguments are sumimarized below. Developinig a Politically Independent and Market-Oriented Enterprise Sector A fundamental objective for new ownership structures in transitional economies is to promote the clarification of property rights and to estab- lish new objectives for the firmn. All privatizations assign titles of owner- ship to particular individuals. But founding a new relationship with the state invTolves ensuring freedom for firms from arbitrary interference and a radical reorientation of goals from seeking rents to satisfying the de- mands of the market (see Boycko, Shleifer, and Vishny 1993; Frydman and Rapaczynski 1994). Compared with state ownership, de novo private and outside private ownership seem likely to be best able to ensure depoliticization of the Ownership Structures, Patterns of Control, and Enterprise Behavior in Ruissia 209 firm and reorientation of objectives. To the extent that the new oAners are entrepreneurs, they will be less a part of the old order, and perhaps have more restricted access to the flow of subsidies.3 Insiders will also have incentives to increase economic profits, since they personally stand to gain through their shareholdings. But they may also have closer ties to the state bureaucracy and greater opportunities to pursue special conces- sions than outsiders or new entrepreneurs. Within the category of insiders, one might also predict a difference be- tween managerial and worker ownership. If budget constraints are soft, it is arguable that transfers of ownership either to managers or to workers will have little or no effect on enterprise behavior, because both sets of new owners will remain motivated to maximize rents rather than profits or earnings per worker. Insider privatization is therefore unlikely to bring many benefits until budget constraints are tightened. Under somewhat harder budget constraints, the net returns to profit as against rent-seeking will be determined by both the opportunity costs and the benefits, which are in turn affected by the prospects of the firm, its environment, the po- litical situation, and so forth. There may be some differences, however, between employee and managerial ownership in this respect. Workers represent a new and more diffuse group of owners than managers, who are generally survivors from an earlier period, maintaining their good connections and bad habits. The costs to seek rents may be higher for worker-owned firms than those under managerial ownership because the former may have more diffuse and heterogeneous objectives. More im- portant, the benefits to rent-seeking may be lower in employee-owned firms because managers, given their longstanding connections under the previous regime, may be more effective at extracting subsidies. Managers may also be able to achieve higher returns to rent-seeking because they may be better able to appropriate the rents for themselves, or becaluse there are fewer of them to share the spoils. In such circumstances, insider privatization to employee owners, by weakening the old relationslhips, might be superior to managerial ownership. Nevertheless, we predict that both will prove inferior to outsider privatization in this area. Long-Term Restructuring We focus on three issues here: unbundling, organizational structure, and investment. The boundaries of firms in a market economy are supposed to be determined by efficiency considerations: the costs and benefits of in- 210 Corpomte Governanice anid Comlpetitioni tegration. But in socialist economies, as emphasized by Komai (1992), the relationship between the managers of firms and their superior, whether the director of a trust or a branch minister, differed little from the rela- tionship between the manager and the foreman or production supervisor under his or her direction. An important element in the transition process is therefore to reorganize the groups of productive units that previously comprised the enterprise sector to form a new industrial structure in which the boundaries of the firms minimize internal transactions costs. A market orientation should also be reflected in changes in enterprise organizational form. The structure of the organization should be adapted to respond to the changing demands of customers, to ensure adequate mechanisms for managerial control, and to provide appropriate informa- tion for rational decisionmaking. This may involve such moves as the es- tablishment of new functional divisions within the firms suitable for finance or marketing and the development of new control and monitor- ing systems. Finally, long-term restructuring involves investment in capi- tal equipment to introduce new technologies, to raise quality standards, to broaden product differentiation, and to address input wastage and its environmental consequences. An important issue is the ability of differ- ent ownership forms to mobilize capital and introduce new technologies. Restructuring, both long and short term, is primarily a problem faced by current and former state-owned firms, so we exclude de novo private firms from these comparisons.4 Provided outsiders are able to exercise their nominal property rights, outsider ownership is probably best suited to long-term restructuring. Given their profit orientation, outside owners will take the most dispassionate view of existing production and organ- izational structures, and in principle they suffer least from agency prob- lems in their dealings with external capital markets. Insider-owned firms might be expected to suffer more serious difficulties in raising outside capital because of the agency problems faced by lenders and minority in- vestors (see, for example, Shleifer and Vasilyev 1994 and Hansmann 1990 for summaries). Ownership by managers is also likely to dominate that of nonman- agerial employees in redefining the appropriate boundaries of the firm. Worker ownership may still be superior to state ownership, because rear- ranging the boundaries of the firm will be possible provided the employ- ees who gain can compensate the losers. In principle, even highly Ownership Structutres, Patterins of Conttrol, and Enterprise Behavior in Russia 211 egalitarian employee-owned firm with high solidarity may therefore be able to undertake some restructuring and unbundling, provided it offers a potential Pareto improvement and some form of compensation package can be agreed upon. In some situations, however, this compensation will not be possible, and potential Pareto improvements will not be convertible into actual Pareto improvements (for instance, because lump-sum transfers are infea- sible or because of severe capital market imperfections). The biggest problems are likely to arise from the difficulties of collective decision- making under uncertainty, particularly when some groups of workers are earning supracompetitive rents. Many enterprises have a large number of potential restructuring paths-for instance, changing product lines, reor- ganizing company divisions, or adopting different kinds of new technolo- gies-but each has different implications for the value of the human capital of groups of workers in the company. Given that the profit associ- ated with each path is also greatly uncertain, each group of workers will try to block paths that seem likely to downgrade their own skills. Thus, it may not be difficult for blocking coalitions to form ex ante, preventing ex post desirable restructuring. In resolving these agency problems, managerially owned firms have a clear advantage. They will be motivated to undertake any restructuring or rearrangements in the boundaries of the firm that increase profits. Su- pracompetitive wages may be reduced and workers laid off with little or no compensation. Agency problems apart, managerial ownership 1hus has the potential to yield restructuring benefits analogous to those of in- vestor ownership, benefits greater than those under worker ownership. Short-Tenn Restructuring The transition process demands that firms become responsive to market signals in the short term, both in the products they choose to supply ind in their use of factor inputs. In firms in which the optimal level of output has fallen, the ownership system must be able to effect large decreases in employment and other inputs. Because of inherited technologies and pro- duction practices that are wasteful in the use of inputs, including energy and labor, new owners must have the incentives and the ability to ensure that costs are reduced, that the factor mix is rationalized, that productiv- 212 Corporate Governanice and Competitioni ity is raised, and that quality is improved. These are the standard prob- lems of restructuring (see, for example, Belka and others 1995; Estrin, Gelb, and Singh 1993). Once again, one can predict that outside owners will have fewer qualms than insiders about reducing employment and implementing other short-term restructuring measures. Nevertheless, they might be un- able to exercise their property rights in such sensitive areas, especially if insiders refuse to cooperate. Moreover, if product or factor markets are relatively more competitive and budget constraints hard, insiders may be forced to restructure and improve their efficiency in order to survive. Comparing managerial and worker ownership, it is important to stress that both have equivalent incentives to increase economic profits and cut nonlabor costs. Worker-controlled firms, however, are more likely than managerially owned enterprises to perpetuate inefficiencies in the allocation of labor. The flip side of this argument is that worker-own- ers would probably be able to get rid of managers more easily. In cases where managerial turnover is a sine qua nion for the firm to be turned around, managerial ownership has the disadvantage of entrenching bad managers. Evolution of Governance Form The transition process involves dynamic adjustment by organizations to changed and changing economic circumstances. The outcome of the proc- ess may be path dependent, and the appropriate institutional arrange- ments may gradually change as the process unfolds. In such circumstances, it may not be possible to specify ex ante the optimal ownership structure, but it would be desirable that whatever structure is first selected should have the flexibility to evolve as the dynamic path of transformation pro- ceeds. The lower the transaction costs involved in exchanging ownership rights, the less binding the initial allocation of those rights, because mar- kets would emerge to ensure a reallocation to achieve better matching of owners with assets. Institutions dealing with property rights should therefore be designed to lower transaction costs and to facilitate the de- velopment of financial markets. The new ownership configuration should also minimize the probability of degeneration back to state ownership. Widespread ownership by outsiders, whether in de novo or privatized firms, is likely to encourage the development of secondary markets, and Ownership Structuires, Patterns of Control, and Enterprise Behavior in Russia 213 thus further the evolutionary process of matching and rematching assets with owners. In contrast, concentrated insider ownership will discourage the development of takeover markets, because the lack of liquidity in small numbers of shares implies that it may be very difficult to ear-n the control premium on minority stakes previously acquired, and rematching is thus inhibited. If worker shareholdings are widely dispersed, secon- dary markets may develop more easily than if shares are concentrat.ed in the hands of a few managers. Although still difficult, it may be somewhat easier for outsiders to take over companies by buying up small numbers of shares than by negotiating with a single manager or a small group of managers. While there may be a collective interest among the insiders to exclude outsiders, individual employees may "free ride" by selling their small holdings to outsiders. Concentrated insider holdings are more likely to lead to entrenchment because of the informational advantage of insiders over outsiders. In an environment that carries great uncertainty over the prospects for any company and suffers a lack of functioning fi- nancial markets to provide estimates of value, the concentration of lhold- ings, together with the asymmetry of information, may give rise to adverse selection in the market for corporate control. Summary of Hypotheses In summary, outside ownership is predicted to provide the greatest pro- gress toward our four objectives for enterprises in transition. Where rele- vant, this performance would be matched by de novo owners. Insider privatization is expected to be superior to state ownership, but inferior to majority outsider control. If we compare forms of insider ownership, worker ownership is hypothesized to have deficiencies in long-term re- structuring, especially rearranging the boundaries of the firm, and in short- term restructuring when employment levels are at issue, but it is perhaps superior in depoliticization and evolution of governance structure. Institutional Features of Russian Privatization The Russian mass privatization involved large-scale giveaways to insid- ers, based on the argument that there was no politically feasible alterrna- tive form of privatization. This is because managers and workers lhad already accumulated tremendous political influence, and enterprises had 214 Corporate Governtance and Competition gained significant autonomy and de facto property rights. Early methods of ownership decentralization under Perestroika had already emphasized leasing arrangements, eventually resulting in insider buyouts at highly preferential prices. The institutional features implied by the State Privatization Program seem straightforward. The legal form of enterprises is an open, individu- ally owned joint-stock company, and shares are in principle fully trad- able, and voting rights (of voting shares) are freely and equally exercised. But there are some important qualifications relevant to our hypotheses, which we list in increasing order of importance. First, in addition to the better-known ways in which workers were able to acquire shares, there was possibility of a kind of ESOP (Employee Stock Ownership Plan), the FARP (Fund of Workers' Shares). On average, the FARP seems to hold only a minor fraction of shares, but may sometimes be more significant, exercising a governance role and restraining share trading.5 Second, un- der the "Option 1" method of privatization, 25 percent of company shares were given to company employees free-of-charge, but under the condi- tion that they be nonvoting.6 Third, as noted above, many companies were privatized outside the State Privatization Program, generally through the buyout of a lease granted to the workers' collective during the years of Perestroika. According to Webster and others (1994, p. 11), "al- most all former leaseholds were either closed joint stock or limited liabil- ity companies." In closed joint-stock companies, share trading is permitted only among employees and with the approval of the workers' collective (which apparently survives in many firms). Furthermore, many observers question the degree to which the legal institutions function in practice, even in nominally open joint-stock com- panies. For instance, there seems to be some evidence of ESOP-like trusts being formed to stifle worker influence. According to Blasi (1994), many managers intended to form a trust for the employees' shares itl order to control how those shares were voted. More generally, voting rights may not always be freely exercised. Managers have reportedly often post- poned the first general meeting of shareholders after privatization, and voting is sometimes said to be conducted neither by secret ballot nor in proportion to shareholdings. Despite frequent press accounts, it is diffi- cult to obtain reliable information on such practices or to estimate their prevalence. Ownership Struictutres, Patterns of Control, and Enterprise Behavior in Ruissia 215 There also seem to be many constraints on the tradability of shares, re- sulting partly from attempts by insiders to prevent the entry of outside investors, and partly from the limited development of secondary mar- kets. Probably the best evidence for the poor possibilities for share trad- ing was the extremely low cash value of vouchers, and the implied low value of company shares.7 Because the cash value of vouchers was deter- mined primarily by transactions involving minority investors, it seems likely that the control premium in this case is enormous: outsiders have little willingness to pay for minority stakes in insider-controlled firmrs.8 Finally, we come to the issue of the residual softness of budget con- straints. Little change in enterprise behavior can be expected to result from ownership changes when firms systematically fail to bear the costs or win the benefits of their actions. It is often assumed that subsicly re- ductions are necessarily associated with privatization, but in Russia this may not be true. Indeed, shortly after the voucher privatization prDcess began, and no doubt with the intention of encouraging the process to move forward, President Yeltsin signed a state decree, "On Not Permit- ting Discrimination Against Privatized Enterprises in the Provision of State Financial Support" (November 27, 1992). Nonetheless, there seems to be agreement that subsidies and money creation generally declined in 1993 and 1994, so that the "nondiscrimination" may be starting to apply in the form of hard budget constraints for all. If this is true, privatization could begin to affect behavior in Russia. We examine the evidence pro- vided by the survey on these points below. Corporate Control in Russian Enterprises In our subsequent empirical work, we address whether firms ownecl by different groups of majority or dominant owners behave differently. To examine this question, five categories of ownership groups were con- structed. The firms in the sample were first classified according to whether they were old enterprises (privatized or state-owned [SO]) or new private ones (de novo private firms, DNs).9 Categories for the possible controlling interests in the old firms were then defined on the basis of the informa- tion on legal form, method of privatization, status of privatization, and the structure of ownership, the last determined by the percentage of vot- ing shares held at the time of interview by ten categories of owners.10 216 Corporate Governance and Comlpetition Old firms were then categorized into state enterprises and those claiming their company "has been privatized."" The latter companies were designated as outsider-owned (00) if banks, investment funds, other domestic firms, foreign institutions, and individuals other than em- ployees together held more than the combined total for insiders.12 In- sider-owned companies were considered to be managerially controlled (MO) when the percentage of shares held by managers was at least as great as that held by nonmanagerial employees. When a larger share was held by nonmanagerial employees, we classified a firm as worker-owned (WO).13 Table 7-2 reports information on the ownership structure of the 439 companies in the sample. Of these, 45 are DNs and 325 are old firms, of which 110 still have a dominant state share and 214 are majority privat- ized (the remaining firms are unclassified). The sample of state-owned and privatized firms was randomly drawn from a list of the population of industrial firms employing more than fifteen workers, to which were added a predetermined number of de novo firms. The data therefore pro- vide an opportunity, which is particularly valuable in the absence of com- prehensive official statistics, to measure the ownership outcome of the Russian privatization process. Workers have become dominant owners in a majority of cases: WOs account for 138 firms, 65 percent of the total; 19 percent, or 40 firms, are MOs; while the remaining 16 percent, 36 firms, are O0s. Among all privatized companies, workers hold an (unweighted) average of 47.5 percent of all shares, and managers hold 20.8, which yields a total insider stake of 68.3 percent, over two-thirds of all shares. The remainder is divided between the state (10.7 percent) and outsiders (19.7 percent), while 1.1 percent of the shares were owned by unclass- ifiable "others."14 The sample contains significant diversity in category of dominant owner, which makes it well-suited for our purpose of relating these cate- gories to elements of the firms' behavior. There also appears to be an as- sociation between the extent of share ownership by workers and that of outsiders: each is more likely to own shares in a company dominated by the other than they are to own shares in a company dominated by either managers or the state. Managers and outsiders seem particularly unwill- ing to own shares in one another's companies. In addition, the state ap- pears to exhibit a slight preference for share ownership in companies dominated by managers over those dominated by workers or outsiders. Owvnership Structures, Patterns of Control, and Enterprise Behavior in Ruissia 217 Table 7-2. Distribution of Ownership by Dominant Owner Type Dominanit ozonera Category SO WO MO 00 DN Total State Mean 89 10 13 12 1 34 Standard deviation 21 14 15 13 5 40 Workers Mean 7 63 14 26 6 31 Standard deviation 14 20 20 14 17 31 Managers Mean 2 12 63 7 58 17 Standard deviation 5 11 23 7 39 26 Outsiders Mean 2 14 9 53 26 15 Standard deviation 6 16 12 21 36 22 Number of enterprises 110 138 40 36 45 439 Note: SO: enterprises with dominant state stake; WO: enterprises with dominant worker stake; MO: enterprises with dominant manager stake; 00: enterprises with dominant out- sider stake; DN: newly established, privately owned enterprises. The total column includes firms that were not classifiable according to dominant owner, and thus the numbers clo not correspond strictly to the sum (or average) of the previous five columns. a. It was possible to classify some firms (two WOs and twenty DNs) without complete in- formation on oNvnership shares. Together, these results provide some evidence against the somewhat prevalent views (for instance, in Webster and others 1994) that manag,ers and workers are in close coalition with one another in privatized Russian firms, and that managers are more likely than workers to become inde- pendent of the state. Official data on the ownership structure of the newly privatized com- panies is unavailable. Nevertheless, our results on ownership shares are of the same order as those obtained from three earlier surveys that at- tempted to collect some of this information for samples of privatized companies. In Pistor's (1994) sample of thirty-six firms, all employees to- gether received an average of 61.8 percent of all shares, while outsiders held an average of 19 percent, and the State Property Fund retained 19.3 percent. Blasi's (1994) survey of 127 privatized firms found 90 percent with majority employee ownership. On average, insiders held 65 percent of shares in his sample, with a median of 60 percent.15 Finally, Webster 218 Corporate Governance and Comepetition Table 7-3. Legal Form by Dominant Owner Type Donintanit owner Legalfornm SO WO MO 00 DN Total Joint stock 27 120 30 31 12 267 Limited liability 0 1 3 0 3 7 General partnership 1 0 1 0 1 4 Limited partnership 0 9 5 0 11 33 Cooperatives 0 1 0 0 2 2 Physical persons 0 0 0 0 14 15 State-owned joint stock 8 0 0 0 0 11 Leasehold 0 2 0 0 1 3 Nonincorporated, state-owned 68 0 0 0 0 70 Other 5 0 0 0 0 3 Total 109 133 39 31 44 415 Note: The total column includes firms that were not classifiable according to dominant owner, and thus the numbers do not correspond strictly to the sum of the previous five columns. and others (1994) reported on a survey of ninety-two privatized firms in Moscovskaya and Vladimirskaya oblasts conducted in October 1993. On average, only 10 percent of the shares of these companies remained with the state, managers held 17 percent, and workers have 61 percent. These studies, of course, rely on small, nonrandom samples and did not have information on major aspects of ownership rights, such as whether shares were voting or nonvoting. Our findings also differ, par- ticularly insofar as the managerial stake in the companies in our sample is significantly larger,16 and because we did find a significant number of outsider-controlled companies in the privatized group. The survey was also conducted later, and there may, of course, have been some evolution of the ownership structure, although most commentators believe such changes have been minimal so far (see Blasi and Shleifer 1995). Tables 7-3 to 7-6 provide information on other characteristics of our sample by our categories of ownership. Table 7-3 reports the breakdown according to legal form for the 415 companies for which this information is available. Among privatized companies, the joint-stock form overwhelm- ingly predominates, with 90 percent of the total, but we are unable to dis- tinguish closed from open joint-stock companies. DNs exhibit a wider variety of forms; the largest number are individual entrepreneurships. Owniership Strinctinres, Patternis of Control, and Enterprise Behavior in Russia 219 Table 7-4. Branch by Dominant Owner Type Donminant owner Induistry sector SO WO MO 00 DN Total Energy 5 1 1 0 0 7 Fuel 8 3 1 1 0 13 Ferrous metallurgy 1 5 1 3 0 10 Nonferrous metallurgy 1 5 1 1 0 8 Chemicals 3 8 2 0 4 17 Heavy machine building 6 11 2 1 1 21 Electrotechnical 3 5 2 1 2 13 Machine tools and computers 7 5 1 1 3 17 Automobile industry 1 5 1 2 2 11 Agricultural machinery 4 5 0 5 2 16 Light machine building 2 1 0 0 3 6 Defense industry 6 4 2 1 1 14 Shipbuilding 2 2 1 3 0 8 Radio industry 9 3 0 0 0 12 Communications and electronics 7 6 0 3 1 L7 Metal constructions 3 5 2 4 1 15 Machine repairing 6 5 2 3 0 16 Wood harvesting 8 2 0 0 0 1 0 Woodworking industry 3 6 3 2 3 17 Construction materials 6 7 1 1 11 26 Textiles 4 11 6 1 4 26 Clothing industry 2 13 6 1 4 26 Food processing 6 8 3 1 0 18 Meat and milk 1 9 0 1 1 12 Other industrial production 6 3 2 0 1 12 Commercial activity 0 0 0 0 1 1 Military-industrial complex 31 12 5 4 1 53 Total 110 138 40 36 45 369 In table 7-4 the distribution by industrial branch is shown, and in table 7-6 the distribution by region. In order to control for differences in tech- nologies and in shocks across firms, we have disaggregated branches ac- cording to the major product, which results in twenty-six roughlly two-digit industrial branches. The survey instrument also asked which firms were part of the military-industrial complex (MIC); 53 of the 369 placed themselves in that category, as against 14 in the defense sector. About 60 percent of MIC firms remain state-owned, a higher proportion than that of all firms, and of the roughly 40 percent that have been privat- 220 Corporate Governtanice and Competition Table 7-5. Dominant Owner by Industry Sector Group Dominantt owner Inditstry sector grouip SO 4VVO MO 00 DN Total Fuel and energy 13 4 2 1 0 20 Heavy industry 36 56 13 18 18 141 Light industry 42 34 8 13 16 113 Consumer goods 19 44 17 4 11 95 Total 110 138 40 36 45 369 Note: SO: enterprises with dominant state stake. WO: enterprises with dominant worker stake. MO: enterprises wvith dominant manager stake. 00: enterprises with dominant out- sider stake. DN: newly established privately owned enterprises. ized, more than half are worker-owned. Table 7-5 offers a simpler picture of the distribution of ownership classes across sectors. Sectors can be combined into four main groups: Group 1 includes sectors 1 and 2; Group 2, sectors 3-13; Group 3, sectors 14-20; and Group 4, sectors 21-26. Of enterprises in Group 1 (fuel and energy), 65 percent are SO, 20 percent are WO, 10 percent are MO, and 5 percent are 00; there are no DNs. It is clear that the state still controls these sectors of the economy, perhaps to levy taxes on their profits. In Group 2, about 70 percent of enterprises are SO and WO. These sectors need considerable investment, but their prod- Table 7-6. Region by Dominant Owner Type Dominanit owner Region Total SO WO MO 00 DN North 53 17 16 6 10 4 Volga-Vyatka 21 6 9 1 2 3 Povolzhski 49 21 18 5 1 4 North Caucasus 36 1 23 5 3 4 Urals 49 18 14 5 6 6 West Siberia 43 13 21 3 4 2 East Siberia 29 8 6 6 3 6 Moscow 43 16 11 6 3 7 Centre 46 10 20 3 4 9 Total 369 110 138 40 36 45 Note: SO: enterprises with dominant state stake. WO: enterprises with dominant worker stake. MO: enterprises with dominant manager stake. 00: enterprises with dominant out- sider stake. DN: newly established privately owned enterprises Ownership Structures, Patterns of Control, and Enterprise Behavior in Russia 221 ucts are in demand. In Group 3, 67 percent of enterprises are also SO and WO, perhaps so the state can continue to control sectors such as electron- ics. In Group 4 workers control more than 45 percent of enterprises, and the state less than 20 percent, perhaps because these sectors require lower levels of investment. Regarding regions, we have combined similar groups of oblasts into nine regions, closely following the usual division of the Russian Federa- tion into twelve economic regions, which differ in level of economic de- velopment and infrastructure, the availability of natural and human resources, their fields of specialization, and their geographic locations. Because of the small number of observations in some regions, however, we have combined the regions of the North and Northwest, Central and Central-Chernozem, and Eastern Siberia and the Far East. In Kaliningrad, we had no observations, and we treat Moscow as a separate region.17 Ownership and Control in Russian Firms What do these data on the structure of ownership imply about who con- trols Russian firms and the nature of enterprise behavior? Despite the relatively small proportion of managerially dominated firms, and of managerial ownership generally, most observers believe that top manag- ers have remained firmly in control (see Blasi 1994; Boycko, Shleifer, and Vishny 1993). In this section we look at the reported degree of "influence" over the kinds of decisions exercised by different owners to test whether nominal ownership and effective controls are positively correlated. "Influence" is measured in our data as a qualitative variable that can take on one of three values: "rarely or never influential" (1), "moderate influence" (2), or "dominant, most important" (3). We assume that these categories are adequate proxies for participation in decisionmaking about the firm's operation, and analyze their relationship with ownership shares. Tables 7-7, 7-8, 7-9, and 7-10 contain the means, by ownership-control category, of the reported influence of several kinds of 'actors" 18 over frcur different kinds of decisions: (a) sales, production, marketing, and current operations; (b) employment, hiring, and firing of workers and social and nonwage benefits; (c) employment, hiring, and firing of management and managerial compensation; and (d) allocation of profits, major invest- ments, sale or lease of major assets, and financial issues generally. One might expect the influence of outside owners to be greater in category d 222 Corporate Governanice anid Competitioni Table 7-7. Clarification of Property Rights: Influence of Actors by Dominant Owner Type Domirlnant ownler Actor SO WO MO 00 DN Total Management, Board of Directors 2.77 2.68 2.86 2.63 2.76 2.73 Manager-shareholders 2.48 2.48 2.58 2.48 2.65 2.52 Worker-shareholders 1.36 1.39 1.41 1.24 1.32 1.35 Outside individual owners 1.15 1.15 1.00 1.30 1.30 1.17 Outside institutional owners 1.26 1.25 1.00 1.30 1.00 1.21 Local government 1.34 1.20 1.16 1.13 1.23 1.23 Federal government 1.47 1.24 1.30 1.38 1.22 1.35 Banks 1.19 1.33 1.27 1.41 1.31 1.30 Note: SO: enterprises with dominant state stake; WO: enterprises with dominant worker stake; MO: enterprises with dominant manager stake; 00: enterprises with dominant out- sider stake; DN: newly established, privately owned enterprises. The total column includes firms that were not classifiable according to dominant owner, and thus this does not corre- spond strictly to the sum (or average) of the previous five columns. than in the other decision areas; of workers to be relatively greater in b; and of managers in a. One would also expect dominant owners to have sig- nificantly more influence on decisionmaking in general than other actors. None of these propositions seems to hold for the data. Rather, in every firm, "management and executive boards" are reported to have the great- Table 7-8. Decisions Concerning Employment: Hiring and Firing of Workers, Social and Nonwage Benefits Domintant owvnier Actor SO WO MO 00 DN Total Management, Board of Directors 2.71 2.60 2.78 2.51 2.66 2.66 Manager-shareholders 2.44 2.40 2.55 2.46 2.64 2.49 Worker-shareholders 1.45 1.43 1.47 1.27 1.21 1.41 Outside individual owners 1.14 1.11 1.00 1.17 1.20 1.11 Outside institutional owners 1.25 1.19 1.00 1.26 1.00 1.15 Local government 1.26 1.21 1.19 1.36 1.18 1.22 Federal government 1.19 1.13 1.14 1.21 1.14 1.17 Banks 1.08 1.11 1.09 1.03 1.14 1.11 Note: The total column includes firms that were not classifiable according to dominant owner, and thus this does not correspond strictly to the sum (or average) of the previous five columns. Ownership Stritctlres, Patterns of Control, and Enterprise Behavior in Ruissia 223 Table 7-9. Decisions Concerning Employment: Hiring and Firing of Management, Managerial Compensation Dominanit ownzer Actor SO WO MO 00 DN Total Management, Board of Directors 2.69 2.61 2.86 2.'4 2.66 2.69 Manager-shareholders 2.40 2.32 2.52 2.57 2.74 2.47 Worker-shareholders 1.24 1.33 1.36 1.26 1.28 1.31 Outside individual owners 1.11 1.12 1.00 1.17 1.10 1.11 Outside institutional owners 1.21 1.23 1.06 1.41 1.00 1.19 Local government 1.30 1.19 1.13 1.25 1.10 1.22 Federal government 1.26 1.16 1.14 1.10 1.14 1.19 Banks 1.10 1.11 1.10 1.03 1.14 1.11 Note: See table 7-2 for acronyms and notes. est influence on all kinds of decisions. They are closely followecd by managerial shareholders, while at first glance all other actors dwindle into insignificance. There are, however, a few areas in which dominant ownership cate- gory affects control over enterprise decisions. First, we note that worker- shareholder influence is consistently greater than the influence of the other actors, with the exception of managers. In this regard, it is particu- larly worrisome that workers are seen as moderately influential over the allocation of profit, especially in worker-owned firms. This sits somewhat uneasily with studies that dismiss the influence of workers outright (see, Table 7-10. Decisions Concerning Allocation of Profits, Major Investments, Sale or Lease of Major Assets, Financial Issues Generally Dominlant owner Actor SO WO MO 00 DN Total Management, Board of Directors 2.77 2.87 2.92 2.67 2.63 2.81 Manager-shareholders 2.47 2.53 2.81 2.41 2.71 2.59 Worker-shareholders 1.42 1.68 1.63 1.26 1.26 1.5.3 Outside individual owners 1.19 1.23 1.10 1.43 1.22 1.25 Outside institutional owners 1.46 1.34 1.12 1.63 1.00 1.37 Local government 1.34 1.27 1.23 1.33 1.29 1.29 Federal government 1.46 1.25 1.28 1.32 1.27 1.32' Banks 1.24 1.27 1.13 1.23 1.22 1.23 Note: See table 7-2. 224 Corporate Governtatce and Conzpetition for example, Blasi and Shleifer 1995). The flip side is that we find limited evidence of outside owners-either individuals or institutions-having significant influence over enterprise decisions, although outsiders do have some influence over financial decisions in OOs, and banks on pro- duction and sales. This weak outside control holds, despite the survey suggestion that their shareholdings are considerable (15 percent on aver- age) and that they are dominant shareholders in about 15 percent of pri- vatized firns. This suggests that, rather than searching for changed shareholdings, one has to look to changes in control and behavior before applauding the gradual increase in outsider shareholdings in Russian firms. Finally, we note a continued, if secondary, influence of the state, es- pecially in state-owned firms and in decisions regarding production and the allocation of profit. We go on to investigate more systematically whether these measures of influence are associated with the magnitude of ownership stakes, using correlation analysis. Table 7-11 contains simnple correlation coefficients between influence and ownership share. The coefficients are typically low, and relatively few are statistically significant.19 Nevertheless, the two groups that gain significantly more influence through higher owner- ship are managers-over the issues of long-run resource allocation-and outside individual owners-over all issues except questions of short-run sales and production. Banks as owners also appear to be able to exercise some control over production decisions through their shareholding. Table 7-11. Correlation of Ownership and Influence Tiype of decision Type of owfner n A B C D Manager-shareholders 257 0.108 0.133 0.143 0.176* Worker-shareholders 233 0.109 0.083 0.135 0.127 Outside individual owners 160 0.178 0.215a 0.188a 0.197a Outside institutional owners 123 0.030 0.051 0.154 0.157 Local government 202 0.017 0.020 0.063 0.071 Federal government 188 0.150 -0.051 -0.007 0.052 Banks 193 0.209a -0.060 0.065 0.142 Note: A: sales, production, marketing, current operations; B: employment, hiring, and fir- ing of workers, social and nonwage benefits; C: employment, hiring, and firing of manage- ment, managerial compensation; D: allocation of profits, major investments, sale or lease of major assets, financial issues generally. a. One-tailed significance: 0.01. Ozvnership Structures, Patterns of Control, and Enterprise Behavior in Rtussia 225 Worker shareholdings are positively correlated with influence, especially over questions of managerial employment and long-run allocative issues, but the effect is not quite significant. These results might be taken as evidence for the common view that Russian managers are largely in control of their firms, regardless of share ownership (see Blasi 1994; Boycko, Shleifer, and Vishny 1993; and Shleifer and Vasilyev 1994). It must be remembered, however, that in all cases the evidence relies on the self-reported perceptions of the managers them- selves. The widespread self-confidence of managers does not in itself con- stitute sufficient evidence. Table 7-11 suggests that the h:igher shareholding yields greater influence, both to outsiders and to banks., and while the evidence on worker shareholdings is weaker, one could imag- ine a normally quiescent workforce intervening to prevent drastic restruc- turing. We therefore go on to examine how closely the objectives of the firm, as demonstrated through observable actions, follow the interests of dominant shareholder groups. Ownership and Enterprise Behavior In this section we analyze empirically whether different structures of shareholding influence enterprise behavior in Russia. In particular, we test some of the hypotheses outlined earlier about the relative effect.s of privatizing to different dominant ownership groups. We report our find- ings in three subsections: * Changing the nature of the economic relationship between the I irm and the state ("depoliticization") * Long- and short-term restructuring strategies ("reorientation") * Short-term enterprise performance, in employment, sales, exports, and the like. The latter two subsections conflate the second and third "objectives of transition," mentioned earlier, in a manner dictated by the data. Unlike the previous section, where we looked at both the numbe:r of shares held by each ownership group and the firms categorized accord- ing to dominant owner, in the work that follows we look only at the five ownership groups by controlling shareholder interest. Our general ap- proach is to use regression analysis to investigate whether there are statis- tically significant differences in enterprise performance by dominant 226 Corporate Governance and Competition ownership category, and if so, whether these differences persist once we control for sectoral, regional, and firm-specific sources of heterogeneity within each ownership class. Our approach is to estimate four OLS regressions on each indicator of performance, commencing with the ownership dummies, then adding a lagged endogenous variable (where available), then including sectoral and regional dummies, and finally controlling for size by employment in 1991. The simplest equation provides information on the distribution of performance by ownership type. The second is a dynamic specification that indicates the impact of ownership on change in performance. Neither of these equations include any other explanatory variables, and they are intended to describe, in a statistically meaningful way, the differences among the ownership groups. The third equation tests whether owner- ship effects on the change in performance can be isolated when a fuller set of explanatory variables has been included as independent variables to control for firm-specific heterogeneity in the data set. In the absence of a formal model to guide the choice of independent variables, and for par- simony and consistency between equations, we prefer to report only re- gressions that control for competitive market pressures and locational effects, picked up by sectoral and regional dummies, respectively.20 Be- cause the size of the firm may be an important variable for certain aspects of Russian transition, however, especially when comparing de novo with current and former state-owned firms, we sometimes also report a final equation that further includes a proxy for firm size-employment in 1991.21 This helps in the analysis of the relative performance of de 7Zovo firms, which could perform differently because they are new and private, or because they are new and small (see Richter and Schaffer, chapter 8 of this volume).22 Distancing from the State In this subsection, we investigate the hypothesis that, relative to state ownership, outsider-owned firms, and especially DNs, will be the most successful in distancing themselves from the state. Between managerially and worker-owned firms, we want to test whether worker-owned firms become relatively less dependent on the state than their managerially owned counterparts. Ownership Strutctutres, Patterns of Control, and Enterprise Behavior in Riussia 227 Our initial approach is descriptive. In table 7-12 we report several proxies for state influence in, and support for, enterprises. The first three variables concern sales of products to state customers, the argument be- ing that the relationship between the enterprise and the state will be closer in enterprises producing primarily for procurement, whether mili- tary or not. PRFORST2 is the percentage of revenue from all "govern- ment customers," while PRFORST4 is the percentage of revenue from the sale of what we infer to be publicly procured goods.23 According to both measures, government sales are most important to SOs, followed in order by MOs and OOs, but they are least important to WOs. Although the standard deviations are large enough to suggest caution in interpreting the results, on average it does appear that the WOs have the fewest sup- ply ties to the state among old companies. It is surprising, however, that the proportion of total revenue derived from government sales on the part of DNs is quite high-30.8 percent-perhaps providing evidence of some dependency of the new private sector on the state. PROFORST measures the change in the percentage of revenue derived from sales of publicly procured goods since 1990 (the change in PPXORST2 is unavailable). The decrease averaged only 2.6 percent, with the size of the decline directly related to the current level, so that these sales fell the most in SOs, followed by MOs, OOs, and WOs. Regression results are reported in table 7-13. In the first column, the only inde- pendent variables are dummies for dominant owner groups. We conEirm that WOs, OOs, and DNs receive a smaller percentage of their revenue from the state, differences that are significant at the 1 percent level (for WOs), the 5 percent level (for DNs), and the 10 percent level (for OOs). But these results are level rather than rate-of-change phenomena; they vanish in the second column, where PRFORST4 from 1990 is added to the right-hand side. The lagged dependent variable has a coefficient of .66, which, with a t-statistic of 27.5, accounts for much of the variation in cur- rent sales to the state. This is evidence that there is significant inertia in sales to the state. The third column shows the results from adding con- trols for sector and region, many of which are significant, but the most important explanatory variable remains the four-year lagged dependent variable. In the latter two equations, we do not pick up any significant differences across ownership forms. This suggests that the significant rankings by ownership type are selection effects by history, region, atnd 228 Corporate Governance and Conmpetition Table 7-12. Depoliticization Dominanit owner SO WO MO 00 DN Total PRFORST2 Mean 32.72 22.56 24.03 23.82 30.77 26.27 Standard deviation 42.00 35.44 37.43 40.95 42.09 38.61 PRFORST4 Mean 9.93 2.21 7.58 3.10 3.02 5.50 Standard deviation 24.66 12.34 21.98 13.93 11.03 19.18 PROFORST Mean -5.56 -1.08 -2.89 -1.93 0.20 -2.56 Standard deviation 17.41 5.18 11.31 9.90 3.26 11.56 PRICONT Mean 0.57 0.32 0.32 0.30 0.24 0.38 Standard deviation 0.52 0.47 0.47 0.47 0.43 0.48 ARRTOST Mean 20.00 13.03 13.16 6.25 6.15 13.88 Standard deviation 33.67 27.24 24.77 21.65 22.19 28.56 STATLOAN Mean 0.20 0,14 0.13 0.14 0.09 0.13 Standard deviation 0.40 0.34 0.33 0.36 0.29 0.34 PREFLOAN Mean 22.09 15.21 6.40 20.58 14.00 16.03 Standard deviation 35.49 27.54 15.79 25.26 31.94 28.38 GOVSUP92 Mean 0.33 0.19 0.18 0.23 0.13 0.22 Standard deviation 0.47 0.40 0.38 0.43 0.34 0.41 GOVSUP93 Mean 0.46 0.32 0.28 0.37 0.16 0.32 Standard deviation 0.50 0.47 0.45 0.49 0.37 0.47 GOVSUP94 Mean 0.39 0.20 0.20 0.31 0.16 0.26 Standard deviation 0.49 0.40 0.41 0.47 0.37 0.44 GOVASS92 Mean 93.67 13.47 10.95 13.97 0.18 30.02 Standard deviation 449.48 57.94 31.22 48.98 0.79 225.60 GOVASS93 Mean 611.69 67.29 139.72 213.09 5.52 220.32 Standard deviation 3,150.09 239.02 519.93 905.69 28.30 1,621.70 GOVASS4 Mean 368.50 107.06 82.92 163.62 3.09 160.22 Standard deviation 1,281.96 700.49 231.05 613.77 11.71 784.17 GASS94BE Mean 0.16 0.10 0.06 0.11 0.03 0.79 Standard deviation 0.50 0.61 0.17 0.30 0.09 13.36 Ownership Structures, Patterns of Control, and Enterprise Behavior in Russia 229 Table 7-12 (continued) Dominant owner SO WO MO 00 DN Total GASS4BE Mean -0.61 -0.38 -0.37 -0.58 -0.15 0.30 Standard deviation 2.42 1.17 1.48 1.68 0.68 13.35 Note: PRFORST4: percentage of production to the state out of the total revenues in 1994; PRFORST(T-4): percentage of production to the state out of the total revenues in 1990; PRO- FORST: change in percentage of total revenue provided by these goods in 1994 compared to 1990; ARRTOST: percentage of liabilities to the state that are overdue more than three months; PRICONT: dummy, which takes on value of 1 if there is price control and 0 other- wise; STATLOAN: dummy that takes on value 1 if enterprise received any loan from gov- ernment; PREFLOAN: percentage of total loans received at the central bank discount: rate. The total column includes firms that were not classifiable according to dominant owner, and thus this does not correspond strictly to the sum (or average) of the previous five coluLmrns. GOVSUP94-92: dummy defined as 0 if there was no government support in 1994-92, : oth- erwise; GOVASS94-92: million rubles of government assistance in years 1994-92; GASS94BE: million 1994 rubles of govenment support per employee received in 1994. GASS4BE = GOVASS94/EMPLOYMENT91 - GOVASS931PI/EMPLOYMENT93, where IPI is Industrial Price Index. sector, and the ownership category is not yet significantly affecting the pace of change of sales to the state. Although it is unlikely to be under the direct influence of enterprises, the continuing existence of price controls does reflect lingering state in- volvement in enterprise behavior, as well as an issue for which influence costs could be quite high. Such controls persist largely through the ability of local governments to constrain the size of markups. PRICONT in table 7-12 is a dummy variable equal to 1 if the firm reports that there are "price controls or fixed profit margins on [their] major products," and zero othervise. By this measure, prices are far from fully liberalized in Russia, with a full 57 percent of SOs reporting price controls. Distinctly fewer privatized companies, 30-32 percent, face controls on their output prices, and the fraction for DNs, 24 percent, is still less, although the lev- els are high in absolute terms. We now turn to the vexing issue of state support for the enterprise sector. ARRTOST measures the percentage of tax liabilities that were more than three months overdue as of 1 April 1994. This follows our pre- dicted pattern exactly. Arrears were highest among SOs, at 20 percent, 230 Corporate Governance and Comipetition Table 7-13. Depoliticization Regressions Dependent variable: PRFORST4 Independentt variables 1 2 3 WO -7.72** 1.22 1.24 (2.46) (1.27) (1.43) MO -2.34 1.89 2.48 (3.52) (1.76) (1.91) 00 -6.83* 1.02 0.96 (3.77) (1.92) (2.16) DN -6.90** 2.17 1.99 (3.36) (2.47) (2.80) PRFORST (t - 4) No 0.66*** 0.67*** (0.02) (0.03) Regions No No Yes Sectors No No Yes it 323 279 279 Adjusted R2 0.023 0.736 0.737 Note: * = significant at 10 percent level; ** = significant at 5 percent level; *'*= significant at 1 percent level. PRFORST4 = percentage of production to the state out of the total revenues in 1994; PRFORST(T-4) = percentage of production to the state out of the total revenues in 1990. followed by MOs and WOs at 13 percent, and OOs and DNs at 6 percent. The next two variables measure loans received with state support. STAT- LOAN is a dummy variable that takes the value of 1 if either of the com- pany's two largest outstanding loans was received from, mandated, or guaranteed by the CBR or any state agency, and the value of 0 otherwise. Twenty percent of SOs received such loans, while they are extended to only 13-14 percent of privatized companies and only 9 percent of new private firms. A measure of preferential credits is PREFLOAN, the per- centage of all loans that carry an interest rate below the discount rate of the CBR. Once again, SOs receive the best treatment: 22.1 percent of their loans are preferential, compared with 20.6 percent among OOs, 15.2 per- cent among WOs, 14.0 percent among DNs, and an average of 6.4 percent for MOs. The final set of variables to measure the extent of depoliticization con- sists of a group of indicators of direct government assistance to the com- panies. As shown in table 7-12, GOVSUP92, GOVSUP93, and GOVSUP94 are dummy variables equal to 1 if the enterprise admitted receiving any Ownership Struicttures, Patterns of Control, and Enterprise Behiavior in Ruissia 231 support from the state-subsidies, investments, tax benefits or exemp- tions, preferential credits, or others-in 1992, 1993, and 1994, respectively. The percentage of companies reporting support rose from 22 percent in 1992 to 32 percent in 1993, before falling back to 26 percent in 1994. The highest percentage of companies receiving such support is the group of SOs-39 percent in 1994. Surprisingly, OOs were next, with 31 percent, followed by MOs and WOs with 20, and DNs, as expected, were last, with 16 percent. This pattern is confirmed by estimating logistic regressions with GOVSUP94 as the dependent variable, shown in table 7-14. The first col- umn shows the simple specification, while only ownership dummies are included on the right-hand side. DNs, WOs, and MOs have a significantly lower probability of receiving state support than do SOs, while between SOs and OOs there is no statistically significant difference. The results in columns 2 and 3, however, make evident that there is quite significant persistence in the receipt of government support: the lagged dependent variable is highly significant in both equations, implying that the same firms that receive support in 1993 tended to receive it in 1994. It is im- pressive that the coefficient on WOs remains statistically significant in these regressions, implying that with systematic regularity, more worker- owned firms lost support in 1994. The reported total values (in current million rubles) of all of the same categories of government assistance are represented in table 7-12 by GOVASS92, GO VASS93, and GOVASS94 for 1992, 1993, and 1994, respec- tively.24 Assistance declined sharply in 1994, to about 20 percent of its real value in 1993, once privatization had been accomplished. Mean assis- tance is highest in SOs, next highest in OOs, followed by WOs, MOs, and DNs. Privatized firms received substantially fewer subsidies than did state-owned enterprises. Because ownership types also differ by size, we divided government assistance by employment; GASS94BE equals the ra- tio of GOVASS94 to employment in 1994. Scaling by size reduces the dif- ferences among dominant owner types, while preserving their order in the receipt of assistance. The change in this ratio from 1993 to 1994 is vari- able GASS43BE, which showed there was little nominal change, but a strong real decline in all the enterprises that we could classify by dorni- nant owner. For example, WOs received only 42 percent of the assistance per employee in 1994 that they received in 1993 (measured in 1994 iu- bles), while OOs received about 32 percent, and MOs about 28 percen-t. 232 Corporate Governance and Comupetitioni Table 7-14. Depoliticization Regressions: Existence of Government Support (logits) Dependent variable: GOVSUP94 Inidependent variables 1 2 3 WO -0.94*** -0.84** -1.21*** (0.29) (0.38) (0.47) MO -0.94** -0.58 -0.56 (0.44) (0.57) (0.66) 00 -0.34 -0.10 -0.47 (0.41) (0.56) (0.66) DN -1.25 -0.30 0.05 (0.46) (0.58) (0.70) GOVSUP92 NO 0.51 0.21 (0.37) (0.45) GOVSUP93 NO 3.16*** 3.64*** (0.39) (0.49) Regions No No Yes Sectors No No Yes Correct predictions (proportion) 73.78 86.22 87.57 n 370 370 370 Note: ' significant at 10 percent level; ** significant at 5 percent level; i significant at 1 percent level. GOVSLuP94-92 = dummy defined as 0 if there was no government support in 1994-92, 1 othenvise. Figures in parentheses represent standard errors. By these measures, Russian budget constraints seem to have hardened quite significantly in 1994. The regression results in table 7-15 provide further support for this conclusion. In column 1, GOVASS94 is regressed only on ownership dum- mies, demonstrating again that the level of assistance provided to WOs, MOs, and DNs is statistically significantly less than that for SOs, while the OOs show no clear difference. Column 2 adds the lagged values of the dependent variable, which, as with table 7-14, reduces most of the own- ership dummies to insignificance. The coefficient on WO remains nega- tive and significant. In the following column, however, where sector and regional dummies are added, even the WO dummy loses significance. In this subsection, we have looked at government-enterprise relations in procurement, price controls, and subsidies. The findings taken together conform to our prior hypotheses-the influence of the state through these three channels is most marked in the remaining state-owned firms, and Ownership Structures, Patterns of Control, and Enterprise Behavior in Ruissia 233 Table 7-15. Depoliticization Regressions: Magnitude of Government Assistance Dependent variable: GOVASS94 Independent variable 1 2 3 WO -261.44** -150.79* -134.66 (110.18) (83.24) (91.16) MO -285.58* -131.82 -124.02 (158.13) (117.11) (125.15) 00 -204.88 -64.37 -81.16 (166.33) (124.29) (134.08) DN -365.41** -175.82 -176.75 (151.49) (113.52) (128.14) GOVASS92 No 0.79*** 0.82*** (0.14) (0.15) GOVASS93 No 0.20*** 0.19**:t (0.02) (0.02) Regions No No Yes Sectors No No Yes Number of plants No No No Employment in 1991 No No No Employment in 1994 No No No Adjusted R2 0.013 0.36 37 n 353 343 343 Note: * = significant at less than 10 percent level; * significant at less than 5 percent level; ***= significant at 1 percent level. GOVASS94-92 = million rubles of government assis- tance in years 1994-92. Figures in parentheses represent standard errors. least in de 11ovo private firms. Insider privatization does act to break the links with the state, although more markedly in worker-owned than in managerially owned firms. Surprisingly, however, the relationship be- tween the state and outsider firms remains very strong, comparable to that in state-owned firms. This could be explained by selection effects: outsiders may have tended to take control in firms with historically close connections to the state. In any case, our results demonstrate the power- ful inertia in the relationship between the state and the enterprise sector. Reorientation of Firms' Objectives and Restructuring We hypothesized earlier that privatized firms, particularly those control- led by outsiders, may be superior to state-owned firms in most areas of restructuring. In comparing insider-controlled firms, worker ownership 234 Corporate Governanzce and Conmpetition might lead to relatively less unbundling, investment, and reduction of la- bor costs than managerial ownership. We test these hypotheses in this section using qualitative data from the questionnaire that recorded man- agers' own views about their restructuring strategies. The questions cover four areas of enterprise decisionmaking: production, marketing, employ- ment policy, and investment. Managers are invited to indicate their pri- orities across a variety of responses in each area; they are allowed to respond on a scale from 1 (not important) to 3 (very important) for each response. The results are tabulated in table 7-16, which reports the rank- order of responses by ownership type and the average response on the 1- to-3 scale. In sharp contrast to the findings concerning depoliticization, we see little evidence that majority ownership stakes are yet influencing restruc- turing strategies among privatized firms, although DNs are clearly some- what different. The most striking thing about table 7-16 is how little the responses vary by ownership type. For example, the mean response across the ten possible actions under the heading of production strategy varies between 1.94 and 2.06. The variation is hardly greater within any particular answer. It is perhaps encouraging, however, that marketing and investment/finance strategies are generally regarded as slightly more important than production or employment strategies, regardless of ownership category. Commencing with production strategy, the rank-orders of importance are remarkably similar in all five ownership types. The ranks in table 7-16 rise with the importance attached to a strategy, so we note that all firms attach least significance in their production strategy to disposing of as- sets, seeking foreign consultants, and closing plants or shops, and most importance to increasing the efficiency of input use and to investments. The only major exceptions are privately owned firms, which presumably are not encumbered with poor practices, at least to the same extent as the others. Hence, as we would expect, they place less importance on employment policy, changing product mix, and improving efficiency of resource use, and emphasize, even more than other ownership groups, technology, product quality, and investment. Privately owned firms are also rather different in their employment strategies; employment reductions are seen as much less important, pre- sumably because as new organizations, they have not inherited the bloated labor forces of current and former state-owvned firms. Apart from Ownership Structueres, Patterns of Control, and Enterprise Behavior in Russia 235 Table 7-16. Responses on Importance of Management Strategies (rank order) Ranik Management strategy SO WO MO 00 DN Produtction strategy 1 Change in area of activity 6 4 7 4 5 2 Changing production mnix 8 7 8 8 6 3 Change of inventory policy 5 6 4 7 3 4 Closing of plant/shop 3 3 3 1 2 5 Change in product quality 7 7 6 9 9 6 Disposing of assets 1 1 2 2 1 7 More efficient use of productive resources 10 10 10 10 7 8 Changing technology 4 4 5 5 8 9 Seeking foreign consulting advice 2 2 1 3 3 10 New investments 9 9 9 6 10 Mean 1.94 1.97 1.9 2.05 2.06 Employment strategy 1 Decrease in labor 4 5 5 4 3 2 Increase in labor 2 1 3 3 5 3 Cutting social benefits 3 3 2 2 1 4 Cutting wages 1 2 1 1 2 5 Increasing wages 7 7 7 7 7 6 Increasing wage differentials 6 6 6 6 6 7 Modify or establish an internal wage scale 5 4 4 5 4 Mean 1.97 1.95 2.00 1.97 1.88 Investment strategy 1 Reducing new bank borrowing 6 5 6 5 7 2 Reschedule loans 3/4 2 3/4 2 5 3 Obtain new loans from banks 2 4 3/4 1 2 4 Obtain new loans from non banks 1 1 1 3 3 5 Lengthening period for payables 5 6 7 7 1 6 Reducing outstanding receivables 8 8 8 8 6 7 Change bank connections 3/4 3 2 4 4 8 Seeking foreign investors 7 7 5 6 8 Mean 2.12 2.15 2.14 2.21 2,07 Marketing strategy 1 Improve marketing 7 6 7 7 6 2 Change distribution network 3 5 5 6 4 3 Change suppliers 2 2 3 2 2 4 Seeking new domestic markets 6 7 6 5 7 5 Increasing export efforts 5 4 1 3 :3 6 Increase product price relative to competitors 1 1 2 1 1 7 Drop product price relative to competitors 4 3 4 4 5 Mean 2.06 2.14 2.07 2.13 2.04 236 Corporate Governance and Competition this, however, the similarities across ownership categories are much more revealing than the differences, and not entirely consistent with the view of unconstrained managerial control. In all ownership forms, the most important strategy by far for employment is an increase in wages, fol- lowed by the desire to increase wage differentials. Outsider-owned firms, however, place slightly more stress on establishing an internal wage structure than insider-owned firms, and-surprisingly-slightly less weight on employment reduction. Turning to investment strategy, some modest differences begin to ap- pear within the private group. New private firms place particular empha- sis on seeking foreign investors and reducing bank borrowing. A similar stress on foreign investment is found in both state- and worker-owned firms. Managerially owned firms in particular, and outsider firms as well, shy away somewhat from foreign involvement. Perhaps this is the case in the latter category because foreign advice and capital are less needed, and in the former case because it would threaten managerial entrenchment. Outside owners also place less stress on obtaining new loans than any other ownership form. On the marketing side, all ownership categories rate an improvement in marketing and discovering new domestic markets very highly, but place less emphasis on price adjustments or changing suppliers. One in- triguing difference, however, is that managerially owned firms place less weight on increasing exports, while state- and worker-owned firms re- gard international markets as potentially greater in importance. Enterprise Performance We conclude our evaluation of the impact of differing majority owner- ship forms by looking not at the self-reported intentions of managers, but at the behavior of their firms. We report the result of regression analysis undertaken to analyze elements of company performance in Russia, in- cluding sales, employment, exports, and pay. Means of the variables un- der consideration by ownership type are outlined in table 7-17, where some differences by ownership category do emerge, although the stand- ard deviations are typically large. The first variable in table 7-17 is sales in 1994. State-owned firms are the largest enterprises, followed by worker-owned, managerially owned, outsider-owned, and privately owned firms. The five kinds of firms in- Ownership Struictures, Patterns of Control, and Enterprise Behavior in Ruessia 237 Table 7-17. Company Performance Dominant owner Item SO WO MO 00 DN Total Sales Mean 157,022 5,970 3,785 3,071 682 7,382 Standard deviation 53,763 41,956 8,913 5,354 3,086 36,949 Profit-maker dummy Mean 0.86 0.86 0.90 0.89 0.87 0.87 Standard deviation 0.34 0.35 0.30 0.32 0.34 0.34 Capacity utilization Mean 54 50 56 43 43 53 Standard deviation 26 26 24 29 29 27 Nongovernment sales in 1994 (percent) Mean 90 98 92 97 97 95 Standard deviation 25 12 22 14 11 19 Exports to non-FSU in 1994 (% sales) Mean 3 5 2 8 9 0 Standard deviation 6 15 7 20 20 0 Capital stock aged 15 yrs (percent) Mean 36 29 30 42 9 32 Standard deviation 29 28 31 316 23 31 Employment in 1994 Mean 3,016 1,886 1,293 2,072 98 1,904 Standard deviation 7,959 8,196 1,808 3,639 146 6,269 Wage of workers Mean 135,988 127,062 131,510 144,357 173,633 1,355,45L,083 Standard deviation 111,337 98,102 102,536 118,705 141,316 53 Wage of managers Mean 162,957 159,976 174,029 205,718 226,103 1,734,741,599 Standard deviaiton 175,990 132,727 162,253 174,979 196,806 15 Note: SO: enterprises with dominant state stake; WO: enterprises with dominant worker stake; MO: enterprises with dominant manager stake; 00: enterprises with dominant out- sider stake; DN: newly established privately owned enterprises. creased sales at a similar average rate between 1992 and 1994. Size ac- cording to employment shows a similar pattern. The Russian firms in our sample are not major exporters outside the former Soviet Union-on av- erage only 4 percent of sales go to such customers and the maximum ob- served in the sample is only 20 percent of total sales. Non-former Soviet 238 Corporate Governanice and Conmpetition Union exports are slightly higher on average in outsider-owned and worker-owned firms than in state-owned or managerially owned enter- prises, and negligible in de novo firms. The information on profits provided in our survey is poor, but the questionnaire did ask firms to report whether they were typically profit- makers. The average response to this question is reported in the second row of table 7-17. It can be seen that according to Russian accounting pro- cedures, most firms normally make profits, and the differences across ownership types are negligible. Turning to capacity utilization, rates in 1994 are very low, averaging around 53 percent across all firms. Never- theless, they are higher among de novo enterprises and lower in outsider- owned firms. The Russian capital stock according to the survey is relatively modern; only about 32 percent is reported as being more than fifteeen years old. It is not surprising that de novo firms have significantly younger capital, on average, but state-owned, worker-owned, and man- agerially owned firms are all close to the mean. The proportion of old capital is higher in outsider-owned firms: 42 percent of the total. Finally, average wages for workers and managers are highest in de novo enter- prises and lowest in worker-owned firms. State-owned and managerially owned firms are close to the mean, while pay for both groups is rather above average in outsider-owned firms. In the remainder of this section we use regression analysis to investi- gate whether these differences persist once we control for sectoral, re- gional, and firm-specific sources of heterogeneity within each ownership class. Our approach is to estimate the four versions of the performance equations outlined at the start of this section. Performance in. short-term restructuring is analyzed in tables 7-18 and 7-19, which explain 1994 sales and non-former Soviet Union exports, re- spectively. Commencing with sales, we note from column (1) of table 7-18 that de ntovo private firms are significantly smaller than state-owned firms (always the omitted class), as are worker-owned firms. In the dynamic specification of columns (2) and (3), however, there are no significant ownership effects, although the sign on all privatized firms is positive relative to state-owned firms. We interpret this to imply that majority ownership structures are not yet significantly affecting the rate of change of sales, although there is great persistence in turnover, as well as signifi- cant effects on the market environment from sectors and regions.25 Ownership Structutres, Patterns of Control, and Enterprise Behavior in Russia 239 Table 7-18. Sales in 1994 Category 1 2 3 WO -9,775* 2,912 3,277 (5,581) (2,021) (2,206) MO -11,916 3,035 2,136 (8,063) (3,037) (3,23C) 00 -12,631 2,437 1,642 (8,357) (3,188) (3,481) DN -15,020*`* 2,605 2,691 (7,524) (3,002) (3,385) Lagged endogenous variable (1 year) No 2.76*** 2.76f** (0.08) (0.08! Sectors No No Yes*` Regions No No Yes** Adjusted R2 0.006 0.86 0.86 it 298 246 246 Note: SO: enterprises with dominant state stake; WO: enterprises with dominant worker stake; MO: enterprises with dominant manager stake; 00: enterprises with dominant out- sider stake; DN: newly established, privately owned enterprises. * denotes significance at 10 percent level; ` denotes significance at 5 percent level; *** denotes significance at 1 percent level. Figures in parentheses represent standard errors. From table 7-19 we find that worker-owned-and to a greater extent, outsider-owned-firms export significantly more than the firms in the other three ownership categories. Despite considerable inertia in export performance over time, this result persists for worker-owned firms in the dynamic specification, and remains nearly as significant when sectoral and regional fixed effects are taken into account. De novo enterprises ex- port notably less; all other ownership forms have a positive sign relative to state-owned firms. We note from the fourth column, however, that the size of firms is not a significant explanatory variable for non-former So- viet Union exports; its inclusion leaves other results unchanged. Turning to capacity utilization, we find contrasting ownership effects in column (1) of table 7-20. There is no significant difference between the rate of capacity utilization in worker-owned, managerially owned, and state-owned firms. Capacity utilization is significantly lower in outsider- owned firms, and higher in de novo enterprises. The latter is easy to ex- plain-de novo private firms did not inherit the same excess capacity and are growing (see chapter 8 by Richter and Schaffer). Perhaps outsiders 240 Corporate Governance and Competition Table 7-19. Percentage of Sales Exported to Non-Former Soviet Union Economies Category 1 2 3 4 WO 2.24* -2.71** 2.58 0.10 (1.64) (1.47) (1.67) (2.19) MO 0.07 1.27 0.58 -3.00 (2.28) (2.06) (2.27) (3.39) 00 6.16*** 3.48 4.09 2.77 (2.52) (2.29) (2.45) (3.23) DN -2.55 0.29 -0.003 -4.08 (2.17) (1.98) (2.34) (4.47) Sectors No No Yes Yes Regions No No Yes Yes Size/1,000 No No No 0.14 (0.12) Lagged endogenous variable No 8.89*** 8.29*** 0.77*** (1.01) (1.12) (0.14) Adjusted R2 0.02 0.21 0.17 0.12 7l 325 325 325 243 Nofe: * denotes significance at 10 percent level; ** denotes significance at 5 percent level; *** denotes significance at 1 percent level. Figures in parentheses represent standard errors. have taken control only of firms with more serious restructuring prob- lems-for instance, having faced a larger output drop or worse inherited capital. It is interesting that these effects typically persist in the dynamic specifications, so the change in capacity utilization is also correlated sig- nificantly with ownership, positively for de novo enterprises and nega- tively for outsider-owned firms. There is also weak evidence that the further decline in capacity utilization tends to be correlated with em- ployee ownership. Once again, the size of the firm does not affect the other results, and the size variable is not significant. It is interesting to ask whether the differences by ownership type are associated with the vintage of the capital stock. There is some evidence for this relationship in table 7-21, at least with respect to de novo private firms. These are found to have a significantly lower proportion of capital more than fifteen years old than other firms. Nevertheless, there is no ex- planation for the poor showing of outsider- and worker-owned firms in their capacity utilization here; the coefficient on outsider-owned firms is Ownership Strluctuires, Patterns of Control, and Enterprise Behavior in Riussia 241 Table 7-20. Capacity Utilization in 1994 Category 1 2 3 4 WO -3.50 -3.97* -3.17 -4.74* (3.78) (2.43) (2.73) (2.81) MO 1.77 -5.46 -5.68 -733* (5.37) (3.47) (3.76) (3.,74) 00 -10.63* -3.38 -7.42* -7.(68* (5.87) (3.80) (4.20) (4.02) DN 16.00*** 5.87** 8.70** 7.03 (5.02) (3.23) (3.76) (5.-39) Size/1,000 No No No -0.14 (0.17) Lagged endogenous variable No 0.87*** 0.85*** 0.88*** (0.04) (0.05) (0.05) Sectors No No Yes Yes Regions No No Yes** Yes Adjusted R2 0.06 0.62 0.62 0.68 n 294 285 246 235 Note: * denotes significance at 10 percent level; ** denotes significance at 5 percent level; -** denotes significance at 1 percent level. Figures in parentheses represent standard errors. insignificant, and for worker-owned enterprises it is positive and weakly significant. Size of firm is once again not significant. A major issue that we predicted would distinguish insider and out- sider privatization was employment. The regressions reported in tab-le 7- 22, however, provide little support for our hypotheses. We do find in column (1) that de novo private firms are significantly smaller. The equa- tions also reveal very strong persistence of employment with significant sectoral effects, but no ownership impact in the dynamic specification.s. It would not be sensible to include a size effect here, as in other equations, because we measure size of firm by lagged employment to 1991. Finally, we look at insider (manager and worker) remuneration in tables 7-23 and 7-24. One might expect this to be higher in insider- than in outsider-con- trolled or state-owned firms, but there is no evidence that Russian man- agers or workers are taking advantage of their position to pay themselves higher wages. No insider-ownership variables are significant. Interest- ingly, however, wages of both managers and workers are found to be higher in de novo private firms, although this is a feature caused by iner- 242 Corporate Governiance and Conmpetition Table 7-21. Proportion of Capital Stock More than Fifteen Years Old Category 1 2 3 WO -6.84* -5.96 -5.82 (3.96) (5.19) (5.92) MO 6.17 3.42 5.01 (5.78) (7.41) (8.25) 00 6.57 8.74 10.04 (5.84) (7.6) (8.32) DN -27.00*** -23.6*** 1.24 (5.3) (7.0) (12.85) Average of sector No -0.02 0.02 (0.14) (0.19) Size/1,000 No No 0.17 (0.36) Sector No Yes*** Yes Region No Yes*** Yes Adjusted R2 0.09 0.08 0.02 it 308 244 193 Note: * denotes significance at 10 percent level; ** denotes significance at 5 percent level; ** denotes significance at 1 percent level. Figures in parentheses represent standard errors. Table 7-22. Full-Time Employment Category 1 2 3 WO -1,130 -183 -161 (895) (277) (279) MO 1,723 930 -176 (1,269) (391) (382) 00 944 -28 -327 (1,352) (415) (414) DN -2,918** 153 200 (1,206) (405) (418) Lagged endogenous variable No 0.92*** 0.91*** (0.02) (0.02) Sector No No Yes*** Region No No Yes Adjusted R2 0.007 0.91 0.93 fl 337 317 317 Note: I denotes significance at 10 percent level; ** denotes significance at 5 percent level; "I" denotes significance at 1 percent level. Figures in parentheses represent standard errors. Ownership Structu res, Patterns of Control, and Enterprise Behavior in Russia 243 Table 7-23. Average Monthly Wage of Managers Category 1 2 3 4 WO -2,981 -20,219 -4,737 -10,530 (22,733) (19,217) (20,887) (23,96'7) MO 11,071 -19,910 -6,645 -19,562 (33,007) (28,176) (29,000) (32,013) 00 42,760 -10,793 19,446 7,1 12 (34,183) (29,363) (31,141) (34,160) DN 63,146**` 23,213 30,414 61,325 (31,995) (27,675) (31,174) (60,562) Size No No No -2.37 (1.48, Lagged endogenous No 0.94*** 1.63*** 1.60`** variable (1 year) (0.16) (0.18) (0.21) Sector No No Yes`** Yes Region No No Yes*** Yes Adjusted R2 0.007 0.31 0.35 0.36 n 306 306 306 245 Note: * denotes significance at 10 percent level; ** denotes significance at 5 percent level; **1 denotes significance at 1 percent level. Figures in parentheses represent standard errors. tia, sector, and region rather than adjustment behavior. Large firms pay their workers more, but not their managers. In summary, therefore, enterprise behavior indicates more ownership effects than we found in rnanagers' self-reported restructuring intentions. These tend to concern the level of perfonnance, however, rather than. the pace of adjustment. There is particularly clear evidence of differences in behavior between de novo private firms and all other ownership catego- ries. Privatization does not yet seem to be affecting employment or sales adjustment. Conclusions The most widely noted features of Russian privatization have been its scale and remarkable speed. In this chapter, we have tried to explore the implications of the privatization program for dominant ownership forms and to analyze the effects of different ownership structures for enterp:rise behavior. Our findings confirm the central ownership role granted by the privatization process to managers and, particularly, to workers, although 244 Corporate Governance and Conzpetition Table 7-24. Average Monthly Wage of Workers Category 2 2 3 4 WO -7,418 -14,204 -4,352 -8,884 (15,319) (14,548) (14,872) (14,626) MO -2,969 -9,136 -11,830 -12,636 (22,690) (21,408) (20,866) (19,778) 00 98,780 3,492 7,872 8,496 (23,242) (22,205) (22,343) (21,043) DN 39,153'* 29,694 26,662 29,514 (21,741) (68,208) (22,155) (37,275) Size No No No 1.59* (0.92) Lagged endogenous No 0.792*** 0.58 1.64*** variable (1 year) (0.12) (0.12) (0.18) Sector No No Yes* Yes Region No No Yes' Yes Adjusted R2 0.004 0.13 0.27 0.43 n 310 310 310 248 Note: * denotes significance at 10 percent level; '* denotes significance at 5 percent level; +*x denotes significance at 1 percent level. Figures in parentheses represent standard errors. it also reveals a higher proportion of outsider-dominated firms-both privatized and de novo-than expected. What are the consequences of this ownership structure for enterprise behavior and restructuring, and what are the policy implications of these findings? Theory led us to expect much better enterprise performance across the board from outsider- than state-owned firms, with insider-controlled companies being somewhere in between. The balance of advantage be- tween worker and managerial ownership depended on the issue raised, with majority managerial ownership potentially offering advantages in long-term and short-term restructuring, but worker ownership perhaps superior in achieving a greater degree of depoliticization and possibilities for evolution. Our findings go some way toward confirming these hypotheses. We find significant differences across various aspects of control, behavior, and restructuring between state-owned and outsider-owned firms, most notably de novo enterprises. There are also differences between state- and insider-owned firms, although they are less marked. The balance of ad- Ownership Structtures, Patterns of Control, and Enterprise Behavior in Russia 245 vantage between managerially owned and worker-owned firms is un- clear overall, but we confirm that depoliticization is more closely associ- ated with the latter than the former majority ownership form. The results on de novo private firms are particularly encouraging, because in other work (see Belka and others 1995), one of us has argued that, in Poland at least, it is the small and mid-size enterprises of the de novo private sector that are leading the transition process. Our findings provide an initial in- dication that the same forces may be at work in Russia (see also chapter 8 of this volume). Our understanding of the Russian privatization process is also much enriched by focusing on the areas in which the data do not support our hypotheses. Although still preliminary, the most striking result is that the differences between state-owned and privatized firms, regardless of ma- jority ownership form, are typically not very great, especially regarding the key issue of restructuring. This phenomenon is probably explained by the nature of the current restructuring, which is occurring primarily from the hardening of budget constraints, and this generally affects all firms (if not more markedly the state-owned firms). Evidence from Poland (see Belka and others 1995; Estrin, Gelb, and Singh 1993) suggests that state- owned firms will adjust their behavior in the early phase of transition solely in response to hard budget constraints and increased market com- petition, without any significant impact from changes in ownership and control. The force of this point is increased when we note that the survey was undertaken relatively soon after the mass privatization was c(om- pleted, probably before major behavioral changes could be expected as a consequence of the new ownership structures. More subtly, the results for enterprises privatized to outsiders are dis- appointing. There is no evidence of greater depoliticization, nor of differ- ences in restructuring strategy, and apart from exports, virtually no difference in performance compared with the other privatized firms, the state-owned sector. One explanation may be that outsiders have simply not yet been able to establish effective control over the firms in w:hich they have a majority stake, a view consistent with the evidence about managerial dominance over decisionmaking. There is also some evidence to suggest that outsiders have taken majority control over firms that are somewhat inferior in capacity utilization, overemployment, profitability, and so forth. Perhaps insiders, who by all accounts controlled the firm's privatization process, only accepted majority outsider ownership when 246 Corporate Governiance aind Competition the situation of the firm was so desperate that the wider resources of out- siders were needed to ensure survival of the organization. In this case, the poor performance of outsiders would be related to the larger scale of the task in hand, rather than deficiencies of outsiders as a majority govern- ance group. We consider the consequences of worker ownership. Our study re- veals that Russian privatization has created an economy primarily com- prising majority worker-owned firms, but the effects on behavior and restructuring are not yet as disastrous as might have been predicted. Many of the reasons we have already noted: for instance, worker owner- ship may assist the process of depoliticization, but restructuring, where it may prove a major impediment, has hardly begun. Some may take heart that even in worker-owned firms, managerial control seems assured. Nevertheless, majority worker ownership may present a threat to effec- tive restructuring in the future, both in the long term, when the key is access to external capital markets, and in the short term, when firms need to address the problem of overstaffing. Policy conclusions follow directly from these findings. First, the mass privatization program has of necessity concentrated the attention of poli- cymakers on the former state-owned sector, but in performance and be- havior, prospects look better with de novo firms. The government may wish to develop a more systematic strategy for small and medium-size enterprise development, especially in the classic areas of weakness for these firms: access to outside (loan) capital, management training, and dealing with bureaucracy. The government may also wish to look more closely at what is going on in outsider-controlled firms, to see whether the problems arise from deficiencies in the legal institutions and arrangements for corporate gov- ernance. If so, regulatory changes or more effective enforcement of cur- rent legal requirements may be required. Finally, we return to the overhanging threat of majority employee control. We do not feel that the potential governance and behavioral problems of such control will necessarily be resolved by continued effec- tive managerial control. In situations of conflict between workers and managers-for example, over mass redundancies-either managers will give way to the dominant owner or they will in some way overrule work- ers, which is counterproductive insofar as it acts to undermine emerging property rights and the rule of law. The way forward is instead for major- Ownership Structures, Patterns of Control, and Enterprise Behavior in Russia 247 ity worker ownership to evolve to new ownership forms, most signifi- cantly outsider ownership. The key policy is therefore to ensure that sec- ondary markets are functioning so that worker shareholdings can be traded and purchasers can obtain full voting rights with their shares References Aghion, P., and W. Carlin. 1994. "The Economics of Enterprise Restruc- turing in Central and Eastern Europe." European Bank for Recon- struction and Development, London. Photocopy. Aghion, P., 0. Blanchard, and R. Burgess. 1994. "Restructuring E'nter- prises in Eastern Europe." European Economic Review 38: 1327-49. Belka, M., S. Estrin, M. Schaffer, and I. J. Singh. 1995. "Enterprise Adjust- ment in Poland: Evidence from a Survey of 200 Privatized, Private and State Owned Firms." Centre for Economic Performance Dis- cussion Paper No. 233, London School of Economics. Blanchard, O., S. Commander, and F. Coricelli. 1994. 'Unemployment: and Restructuring in Eastern Europe." In Simon Commander and Fabrizio Coricelli, eds., Unemployrment, Restructuring and the Labor Market in Eastern Europe and Russia. Washington, D.C.: World Bank. Blanchard, O., R. Dombusch, P. Krugman, R. Layard, and L. Sumrners. 1991. Reforms in Eastern Europe. Cambridge, Mass., and London, U.K.: MIT Press. Blasi, J. 1994. "Ownership, Governance, and Restructuring." In Ira Lieber- man and John Nellis, eds., Russia: Creating Private Enterprises and Efficient Markets. Washington, D.C.: World Bank. Blasi, J., and A. Shleifer. 1995. "Corporate Governance in Russia: An Initial Look," Rutgers University and Harvard University. Photocopy. Boycko, M., A. Shleifer, and R. Vishny. 1993. "Privatizing Russia." B10ook- ings Papers on Economic Activity 2: 139-92. Earle, J. S., and S. Estrin. 1994. "Employee Ownership in Transition." Pa- per presented to the World Bank-Central European University Conference on Corporate Governance in Central Europe and IRus- sia, December 1994 (revised February 1995). Earle, J. S., R. Frydman, and A. Rapaczynski. 1993a. "Notes on Voucher Privatization in Eastern Europe." In D. Fair and R. Raymond, eds., The New Europe: Evolving Economic and Financial Systems in East and West. Kluwer Academic. 248 Corporate Governance and Conmpetition . 1993b. "Privatization Policies in Eastern Europe: Diverse Routes to a Market Economy." In Privatization in the Transition to a Market Ecoizonomy: Stludies of Policies and Preconditionts in Eastern Eturope. Lon- don: Pinter and St. Martin's. Estrin, S., editor. 1994. Privatization in Central and Eastern Euirope. London: Longrnan. Estrin, S., A. Gelb, and I. J. Singh. 1993. "Restructuring, Viability and Pri- vatization: a Comparative Study on Enterprise Adjustment in Transition," Policy Research Department, World Bank, Washing- ton, D.C. Frydman, R., and A. Rapaczynski. 1994. Privatizationl in Eastern Eulrope: Is the State Withering Away? Budapest: CEU Press. Frydman, R., A. Rapaczynski, and J. S. Earle, editors. 1993a. The Privatiza- tion Process in Central Eutrope. Budapest: CEU Press. Frydman, R., A. Rapaczynski, and J. S. Earle, editors. 1993b. The Privatiza- tion Process in Russia, Ukraine, and the Baltic Reputblics. Budapest: CEU Press. Hansmann, H. 1990. "When Does Worker Ownership Work?" Yale Law Reviewu 99 (8): 1751-816. Ickes, B., and R. Ryterman. 1993. Paper delivered to a Conference on Pri- vatization and Restructuring in Russia at IIASA, July 1993. Kornai, J. 1992. The Socialist Systemt: Tlhe Political Economy of Communi1iuismll. Princeton, N.J.: Princeton University Press. Lieberman, Ira, and John Nellis. 1994. Russia: Creating Private Enterprises and Efficient Markets. Washington, D.C.: World Bank. McFaul, M. 1994. "Agency Problems in the Privatization of Large Enter- prises in Russia." In M. McFaul and T. Perlmutter, eds., Privatiza- tion, Conversion and Enterprise Reform in Ruissia. Palo Alto, Calif.: Center for International Security and Arns Control, Stanford Uni- versity. OECD. 1994. Trends and Policies in Privatization 1 (3). (Special Feature on Management and Employee Buyouts in the Context of Privatiza- tion.) Pinto, B., M. Belka, and S. Krajewski. 1993. "Transforming State Enter- prises in Poland: Evidence on Adjustment by Manufacturing Firms." Brookings Papers on Economic Activity 1: 213-70. Ownership Structures, Patterns of Control, and Enterprise Behavior in Russia 249 Pistor, K. 1994. "Privatization and Corporate Governance in Russia: An Empirical Study. In M. McFaul and T. Perlmutter, eds., Privatiza- tion, Conversion and Enterprise Reform in Rulssia. Palo Alto, Calif.: Center for International Security and Arms Control, Stanfor(d Uni- versity. Radygin, A. D. 1994. "Privatization and Investment in Russia: Wlhy the Model Needs Revision." Stuidies on Russian Economic Development 5 (5): 437-43. Shleifer, A., and V. Vasilyev. 1994. "Managerial Ownership and the Rus- sian Privatization." Paper presented to the World Bank-Central European University Conference on Corporate Governance in Central Europe and Russia, December 1994 (revised 1995). Webster, Leila, and Joshua Charap. 1993. The Emergence of Private Maanu- factutring in St. Petersburg. World Bank Technical Paper No. 228. Washington, D.C. Webster, L., J. Franz, I. Artimov, and H. Wackman. 1994. Nezwly Privatized Russian Enterprises. World Bank Technical Paper 241. Washington, D.C. Notes 1. Justification for these objectives and further discussion is contained in Earle and Estrin 1994. 2. The term "depoliticization" does not precisely capture the concept that we are investigating. According to Merriam Websters Nezw Collegiate Dictionary, it means "to take out of the realm of politics." We are concerned with inculcating a profit orientation and establishing market discipline over firms. This relates to the nature of control over enterprises (for example, state versus private), the objec- tives of the controlling group (such as rent versus profit maximizing), and to the environment in which they operate (for example, soft versus hard budget con- straints). For ease of exposition, we henceforth use the term depoliticization to re- fer to this complex process of firms distancing themselves from the state. 3. Webster and Charap (1993), in an early survey of ninety-nine private manu- facturing firms in St. Petersburg, found that the vast majority of Russian entrepre- neurs formerly held high-level posts in state-owned enterprises. While their skills in the sector are clearly relevant, however, it is unclear that these new entrepre- 250 Corporate Governance and Comnpetition neurs would also be able to take with them favored access to government grants, let alone rent-seeking attitudes. 4. The de novao private sector as a whole, of course, can influence industrial and economic restructuring. They may, for example, invest and minimize labor costs. But they are not adjusting from a former state-owned structure to a market-deter- mined optimum, but instead adjusting as profit-maximizers to changing market conditions. As such, we exclude them from the table. 5. Unfortunately, our data allow neither a computation of the shares held in a FARP nor an assessment of its effects on behavior. 6. Although the data do not distinguish voting from nonvoting shares, we know the method of privatization and were able to make adjustments for this fac- tor in our appraisal of corporate control. 7. According to Boycko, Shleifer, and Vishny (1993), imputing the value of the entire Russian capital stock on the basis of the cash value of vouchers would re- sult in a figure around the net worth of one large U.S. company. 8. Some evidence may be found in Pistor 1994. She states, for her sample of recently privatized companies, that "trading volumes were low, and usually oc- curred among employees and former employees" in the summer of 1993. More- over, the prices on the secondary markets were reportedly still much lower than in the original voucher auctions, again implying extreme shyness on the part of outsiders. Webster and others (1994) also found little evidence of share trading. 9. A major gap in the sample concerns the date of privatization. We can assume that most of the privatizations in the State Program were implemented from late 1992 until mid-1994, but lease buyouts may have taken place earlier. 10. The structure of ownership was not available in some observations, but often it could be inferred and imputed-for instance, in unincorporated state en- terprises. In other cases, firms claimed to be privatized, but reported that a major- ity of their shares were still held by the state; we classified them as state-owned (SO). Problems also arose because of missing values, answers of an unspecified or ambiguous "other," and the presence of nonvoting shares. Option 1 in the State Pri- vatization Program gave employees 25 percent of the shares free-of-charge, but the shares carried no voting rights; we subtracted those shares from the numbers given for insiders, and on this basis reclassified a number of companies. 11. In the group of the potentially privatized, we designated as SO all compa- nies in which the federal and regional property fund still owned 50 percent or more of the shares. 12. Unfortunately, we had no information on which of these entities might themselves still be state-owned, and in our analysis we are implicitly assuming they are all private. Ownership Structures, Patterns of Control, and Enterprise Behavior in Ruissia 251 13. When the data did not permit us to classify companies by dorninant owner, including cases of inconsistent answers across questions, the firmns are designated "unclassified." 14. The open-ended answers to the ownership (and other) questions allowed several "other" owners to be classified reliably into one of our categories. 15. Thus the distribution is positively skewed, implying that there we're few firms that had a small proportion of insider ownership. Outsiders held an aver- age of 21.5 percent of the shares, and the state retained an average of 13 percent. Blasi also provided information on the division of shareholdings between top managers and all other employees: top managers had an average of 8.6 percent of all shares (the median was 5 percent). 16. Blasi has pointed out that he defines managers as top managers, while we refer to all managers. Using our definition, he finds median managerial owner- ship to be 15 percent. 17. Our ten areas were constructed as follows: Moscow (Moscow city), Center (Vladimirskaya, Voronezhskaya and Moskovskaya oblasts), Urals (Permskaya and Sverdlovskaya oblasts and Bashkorstan), West Siberia (Novosibirskaya, Tyumenskaya, and Kemerovskaya oblasts and Altayskiy kraj), East Siberia (Kras- noyarskiy and Primorskiy kraj), Povolzhski (Tatarskaya, Saratovskaya, and Samarskaya oblasts), North Caucasus (Rostovskaya oblast and Stavropolskiy kraj), North (St. Petersburg and Leningradskaya oblast), North (Arkhangelskaya and Volgogradskaya oblasts), and Volga-Vyatka (Nizhnegordskaya oblast). 18. This includes owners and other actors not specifically identified as own- ers, but whom we use as proxies for the corresponding ownership group, local and federal governments, and banks. 19. This result remains essentially unchanged when the ownership share vari- ables are allowed nonlinear effects, as follows: less than 10 percent was reclassi- fied as "1," 10 to 25 percent as "2," and over 25 percent as "3." 20. A persistent problem with this data set is that, because of missing values scattered across variables, a change in specification of the equation can lead to major changes in the size of the data set used to estimate the model. These differ- ences are minimized by including only lagged endogenous variables (since iirms usually report the previous value for a variable if they report it currently) and sectoral and regional dummies (which we have for all firms). The number of ob- servations will generally be smaller when firm size is included in the fourth speci- fication. 21. In the cases where we estimated equations with such a specification, the data set is a slightly updated version. 22. We are indebted to Mark E. Schaffer for suggesting this line of enquiry. 252 Corporate Governance and Competition 23. The variable is defined as the sum of "military goods" and "nonmilitary goods" purchased by the state (hospital products, schoolbooks, and so forth); PRFORST4 is therefore smaller than PRFORST2. 24. Because data are provided in the table only for the first half of 1994, the comparisons in this paragraph multiply the amount of assistance for 1994 by two. When calculating real changes, we employ the relevant price index for the first half of 1994. 25. The fourth equation, controlling for size, is not included in table 7-18 because of the close relationship between sales and employment. As expected, employment was found to have a positive and significant coefficient in the regression. 8 The Performance of De Novo Private Firms in Russian Manufacturing Andrea Richter and Mark E. Schaffer The preceding chapter noted, among other things, that the de novo (newly established) private finns in the World Bank survey were quite different from their state-owned and privatized counterparts-in particular, their economic performance was clearly better. The differences among the various categories of state-owned and privatized firms were for the most part not nearly so great. We will consider in more detail the performance and prospects of de novo private firms in Russian manufacturing. The motivation for focusing on the de novo private sector is twofold. First, early observers such as Komai (1990) and Murrell (1990) suggested that the de novo private sector could play a critical role in the transition from socialism. Recent evidence from other transition economies indeed suggests that the newly established private sector can play a leading role in recovery, including manufacturing output growth. In Poland, for ex- ample, very strong growth in industrial output is being driven largely by the entry and expansion of de inovo private manufacturing firms. Second, it has been suggested by Barberis and others (1995) that the perform;mce of Russian firms following privatization may depend critically on the in- troduction of new human capital. Strong performance by new private firms, which are by definition mostly new bundles of human capital, 253 254 Corporate Governance and Competitionz compared with similar state-owned or privatized firms would support this argument. We can think of several reasons why the de novo private sector might grow rapidly early in the transition. One explanation just noted has to do with human capital and entrepreneurship in new private firms. Another derives from the nature of the Soviet economy: planning favored large units, and the small and medium-size enterprise (SME) sector was very small by Western standards. Therefore, during transition one would ex- pect to find rapid expansion of the previously suppressed SME sector; new small firms are established and existing ones expand as the SME niche is filled. A similar argument can be made about selected industrial sectors that were suppressed under the socialist system-retail trade and residential construction are examples-and that can be expected to grow rapidly as new firms enter the sector and existing firms in the sector grow. Transition also means restructuring across regions as well as across sectors, and de iiovo firms are thus likely to be overrepresented in cities and towns that are expanding commercial centers. Our goal in this chapter is twofold. First, we aim to describe in more detail the characteristics of newly established private manufacturing firms in relation to state-owned and privatized firms. Second, we try to establish whether these differences can be attributed to SME effects, in- dustrial sector effects, or location effects, or whether the dynamism we observe in the de nzovO private sector requires other explanations-for example, human capital and entrepreneurship. We do not explore differ- ences among the categories of privatized firms (worker-owned, manager- owned, and the like), but instead refer the reader to chapter 7 in this volume. Here we note only that our grouping together of all privatized firms is justified in part by the findings detailed in that chapter that the differences between the categories of privatized and state-owned firms are in general not nearly as substantial as the differences between de 1lovO and all other firms. We find that in most performance indicators, de nzovo private manufac- turing firms look significantly better than their state-owned and privat- ized counterparts. They are actually growing rather than contracting, operating at higher levels of capacity utilization, expanding employment rapidly, and investing more. Their outlook for future performance is similarly more positive, with higher expectations for growth of output and employment, and more planned investments. In most cases these dif- The Perfornmance of De Novo Private Firms inh Russian Manulfactutring 255 ferences appear to be inherent to the de novo character of the firms and cannot be attributed to their size, location, or industrial sector. In our con- cluding remarks we consider the role these dynamic enterprises are likely to play in the recovery of the Russian economy. Sample Characteristics and Methodology As explained in more detail in the appendix to this volume, the initial sample size of fifty for de novo firms was chosen arbitrarily. Furthernore, the survey was constructed to exclude microfirms (those employing fewer than fifteen persons); we would expect at this stage in the transition that most ie novo firms would be found in the microfirm sector. As we shall see, the de novo firms included in the sample are still generally small. Consequently, these firms account for only a very small fraction-about one-half of one percent-of total employment in our sample. This cannot be taken as an estimate of the weight of the Russian de novo sector in industry. A good indication of the actual size of the de novo sector in Russian in- dustry is given by results from a population survey conducted by the Centre for the Study of Public Policy,1 which found in mid-1995 that 6 percent of manufacturing sector employment was in the de novo sector (table 8-1). This is roughly ten times the share we found in our samnple. We have further evidence on small industrial enterprises in Russia from Goskomstat. At the end of 1993 there were 78,000 "nonstate" small and medium-size firms in industry (defined as firms employing up to 200 persons),2 employing about 1.5 million persons in total, or 8 percent of to- tal industrial employment; the sector, moreover, is apparently growing rapidly.3 Much or most of this activity is likely to be in de novo firrns. In sum, it appears that the contribution of the de novo private sector to Rus- sian industrial output is already significant in the aggregate. As noted above, we might expect that a priori smaller firms (whether state-owned, privatized, or de 1lovO) could be expanding rapidly to fill the SME niche. Another reason that smaller firms in our survey might appear to be more dynamic derives from the selection bias resulting in the pres- ence of only "survivors" in the survey. Ceteris paribus, we would expect a small firm to be less likely to survive a negative shock than a large firm.4 The smaller firms in the survey will not represent "losers" as well as they do "winners." Since de novo firms are mostly small, the effect of this survi- 256 Corporate Governance and Competitioni vorship bias would be to exaggerate the performance of the de novo groups compared with privatized and state-owned firms. We attempt to isolate de novo effects from size effects (both SME-niche effects and survivorship effects) and from industry and location effects in a very simple way-with regressions that include dummy variables for these characteristics. The usual test derives from a single regression of a firm characteristic or response on three dummy (1/0) variables: state ownership; whether the firm is de novo; and whether the firm is an SME, defined as a firm employing up to 200 persons. The benchmark category is therefore non-SME privatized firms. Because we have nearly 100 pri- vatized or state-owned SMEs in the sample, we have some hope of being able to separate size effects from de novo effects. In most cases regressions are run a second time, when fifteen industry dummies and four location dummies (major city, oblast capital, other town, rural) are included along with the de novo and SME dummy variables; in this case we are trying to isolate de novo effects from size, industry, and location effects. The regres- sions take the form (8-1) y = ax + ,DN + y STATE + 8SME (8-2) y = oc + ,DN + y STATE + 8SME + industry dummies + location dummies where DN = 1/0 if the firm is/isn't a de novo firm, and the other dummies are similarly defined. In regression 1, a test of the statistical significance of the coefficient on the de novo dummy (3) is a test of whether, holding size constant, de novo firms are different from privatized firms. Similarly, the statistical significance of the state-owned dummy (y) tells us whether, holding size constant, SOEs are different from privatized firms. Finally, a test on the coefficient on the SME dummy (5) is a test of whether, holding ownership constant, small and medium-size firms are different from larger firms (including possible survivorship-related reasons). The inter- pretations of the coefficients in regression 8-2 are similar, except we are holding industry and location constant as well. In all cases the regression methods depend on the form of the variable being tested: continuous variables use ordinary least squares, 1/0 variables use logit regressions, and ranked categorical variables use an ordered logit procedure. The Performance of De Novo Private Firms in Ruissian Manufacturing 257 Table 8-1. Size Distribution of Firms Overall Category sample State Privatized De novo Unknown Number of firms 439 110 272 45 12 Percentage of sample 100 25.1 62.0 10.3 2.7 Firm employment size (firms reporting employment) 50 36 3 9 24 0 51-200 91 20 52 16 3 201-1,000 126 39 83 3 1 1,001-5,000 103 19 84 0 0 > 5,000 28 14 14 0 0 Total number of firms reporting employment 384 95 242 43 4 Average employment 1,945 3,053 1,867 97 163 Total employment 746,724 290,057 451,828 4,189 650 (percentage of sample) (100) (38.8) (60.5) (0.6) (0.1]) Memo item: CSPP population survey, April 1995 Percentage of persons working in manufactur- ing, by firm ownership 100 41a 53 6 n.a. n.a. Not available. a. The 41 percent state = 25 percent working in state-owned firms, 15 percent working in mixed state/private firms, 1 percent in budgetary units. Source: CSPP data from New Russia Barometer-IV press release, Centre for the Study of Public Policy, University of Strathclyde, July 1995, and authors' calculations. CSPP survey covered approximately 2,000 adults. Before proceeding, several caveats regarding the sample are in order. First, the number of de novo firms in the survey is not large. After exami- nation of survey responses and recategorization of several firms based on these responses (see the appendix to this volume), we are left with a sam- ple of forty-five newly established private firms. Missing values and par- tial responses often reduce the number still further. More fundamentally, 258 Corporate Governance anid Conmpetitionz whereas the selection procedure for the basic sample of state-owned and privatized firms was genuinely random, selection of the de novo sample was essentially ad hoc (again, see the appendix); perhaps in effect random, but perhaps not.5 Finally, we have noted already that in principle we ex- pect a sample of small firms to be prone to a survivorship bias. It is also possible that turnover of de novo firms is more rapid than average, even holding size constant-for example, because de novo firms are typically young, and birth and death rates are generally higher for young firms. This could mean that simply holding size constant may not be good enough to eliminate the survivorship bias in the de 11ovO sample relative to the state-owned and privatized sample. Basic Characteristics of De Novo Firms The most salient feature of de inovo manufacturing firms in our sample is that they are small (table 8-1). Nearly all the sampled de novo firms had 200 or fewer employees during 1993-94 (average employment was 100), with over one-half of these firms in the smallest size category (fifty em- ployees or fewer). They are overwhelmingly an urban phenomenon, with two-thirds located in oblast capitals. Table 8-2 shows this overrepresenta- tion in larger cities to be statistically significant even when controlling for size, or size plus industry; it is not an SME effect (holding ownership or ownership + industry constant, we find SMEs are, if anything, overrepre- sented in rural areas). That such a large share of our subsample finds it- self in oblast capitals, rather than in just the two largest cities (Moscow and St. Petersburg), can be seen as a positive sign that entrepreneurial ac- tivity has taken root across the cities of Russia, not just in the major urban centers. Nearly one-quarter of the de novo firms in our sample produce build- ing materials, reflecting the boom in residential construction currently under way in Russia (table 8-3). Another fifth produce light industrial goods. Most of this sectoral distribution appears to be driven by size, apart from representation of de novo firms in the construction materials and food processing industries, which was far higher than average in the former case, and somewhat lower in the latter. When examining a new firm's capital stock, one would expect second- hand equipment from state or privatized firms to be the norm, because Russia's developing capital markets may prohibit start-ups from obtain- The Performanice of De Novo Private Firms in Ruissian Manufacturing 259 Table 8-2. Geographical Distribution (percent) Significance tests vs. privatized non-SME Area State Privatized De novo State De novo SMdE Major citya 27 16 24 0 + 0 O + 0 Oblast capital 24 38 67 - ++ _ ++ 0) Other town 42 38 7 0 - ( 0 - 0 Rural 7 8 2 0 0 + O 0 1 Total 100 100 100 Note: Significance tests: These are tests deriving from a single regression of a firm charac- teristic or response on three dummy (1/0) variables-state ownership, whether the firm is de novo, and whether the firm is a small or medium-size enterprise (SME), defined as a firm employing up to 200 persons. The benchmark category is therefore non-SME privatized firnms. In most cases regressions are run twice; the second significance result comes from a regression also including industry dummies (fifteen industry categories) and city durmmies (four city categories). Regression methods depend on the form of the variable being tested: continuous variables use ordinary least squares, 1/0 variables use logit, and ranked categorical variables use or- dered logit. Significance levels: ++ = positive and significant at the 1% level; + = positive and signifi- cant at the 5% level; 0 = insignificant at the 5% level; - = negative and significant at the 5% level; - - = negative and significant at the 1% level. (For more details, see text). a. Major city: Moscow, St. Petersburg. ing financing for the acquisition of new equipment and machinery. This conjecture is supported by Webster and Charap (1993), who found in an early study of new private manufacturing firms in St. Petersburg that in the ninety-nine firms surveyed in late 1992, equipment in de novo firms was usually second-hand and old. An auspicious result from our survey is that de novo firms report, quite in contrast with those in the Webster and Charap study, that most of their equipment is relatively new (table 8- 4). While start-up firms stated that roughly 59 percent of their capital stock is less than five years old, only 15 percent of the capital stock o0 es- tablished firms is of this vintage. This difference remains significant when we control for size, location, and industrial sector. Over 80 percent of the 260 Corporate Governance and Conmpetition Table 8-3. Sectoral Distribution (percent) Significance tests vs. prioatized non-SME Sector State Privatized De novo State De novo SME Energy 5 2 2 0 0 0 Fuels 7 2 0 + (none) (none) Metallurgy 2 8 0 0 (none) 0 Chemicals 3 7 9 0 0 0 Heavy machinery 8 1 7 0 0 - Engineering 6 3 7 0 0 0 Car manufacture 1 3 4 0 0 0 AgricuLtural machinery 3 4 0 0 (none) 0 Machine-building 22 13 4 0 0 - Other machinery 11 11 13 0 0 -- Wood/paper 10 6 7 0 0 0 Building materials 5 6 24 0 ++ 0 Light industry 5 15 18 - 0 0 Food processing 6 9 2 0 - -- Other 5 2 2 0 0 + Total 100 100 100 Memo item: Military-industrial complex firm 28 11 2 ++ 0 -- Note: Significance tests: logit of industry dummy on ownership and SME dummies. See note to table 8-2. capital of new firms is less than ten years old. Having such a young capi- tal stock places de niovo firms in a good position to take advantage of new opportunities opening up in the manufacturing sector. The picture for SMEs is quite different; after controlling for ownership effects, we find that the vintage of the capital stock of SMEs differs only moderately from larger firms. Why do our findings differ from those of Webster and Charap? Aside from sampling issues, the difference may be evidence that de novo firms are progressing and modernizing rapidly: our survey was done eighteen months later, and the firms included were an average of about 50 percent larger in full-time employment. The Performance of De Novo Private Firms in Ruissian Mantufacturing 261 Table 8-4. Vintage of the Capital Stock (percent) Significance tests vs. privatized non-SME Age State Privatized De novo State De novo SME 0-5 years old 14 15 59 0 ++ 0 0 ++ 0 5-10 years old 21 25 21 0 0 - 0 0 Cl 10-15 years old 28 25 11 0 -- 0 0 -- 0 > 15 years old 36 34 8 0 0 -- 0 Total 100 100 100 Note: Significance tests: OLS without (line 1) and with (line 2) city + industry dummies. See note to table 8-2. Another interesting finding is the degree of inter-de novo firm trade. The total sample showed a strong correlation between ownership type of the enterprise, and ownership category of primary customers: 34 percent of the sales of state-owned enterprises were to other state-owned enter- prises, and 39 percent of privatized firms' sales were to other privatized firms. In both of these cases, the strong correlation can probably be inter- preted as a continuation of trading ties established before transition. The same cannot be said in the case of de novo firms, for which other de novo enterprises represent a significantly greater share of sales than they do for privatized/state firms (24 percent of their customer base, nearly three times as great as for other firms). These new businesses did not inherit trading partners as corporate entities, although their managers or owners may have been able to draw on their personal network of busi- ness contacts. Wage and labor data are presented in table 8-5. De novo firms reported higher average wages for May 1994 than did state or privatized firms--17 and 23 percent higher, respectively. The differences show up primarily in higher blue-collar wages in de novo firms, with firm location and ind-us- trial sector, but not size, explaining most of the differences.6 As discussed in more detail in chapter 3, de novo firms provide fewer social benelits 262 Corporate Governance alnd Competition Table 8-5. Wage and Labor Data Significance tests vs. privatized nioni-SME Item State Privatized De novo State De novo SME WVage, May 1994 (thouisaind rubles) Firm average wage 163 155 191 0 ++ 0 o o 0 Blue collar 164 149 195 0 ++ 0 o o 0 White collar 176 152 160 0 0 0 0 0 0 Managerial 236 218 254 0 0 0 0 0 0 Social beuiefits Benefits provided? 96 96 68 0 0 - - (percentage of firms) 0 0 -- Number of benefits 5.3 5.1 2.0 0 0 -- _ Cost of benefits 14.7 16.2 9.7 0 0 (percentage of wage bill) 0 0 - Unionization Unionization (percent) 83 73 32 + -- 0 0 -- 0 Any unionization? 88 80 33 0 - - 0 (> 1 percent) 0 -- 0 Note: Significance tests: Wage, cost of benefits, percentage unionization - OLS without (line 1) and with (line 2) city + industry dummies. Any benefits provided, any unionization - logit. Number of benefits - ordered logit. than the typical firm in the survey. This is largely, but not entirely, a size effect. The significance tests show across all measures of benefit provision that levels of provision in SMEs are lower than in larger firns, but that if we consider only the number of benefits offered, we find de novo firms of- fer fewer benefits than privatized and state firms, even when controlling for size. Unionization levels are much lower in de novo firms (32 percent) than in state-owned (83 percent) or privatized firms (73 percent). This is The Performnance of De Novo Private Fimls in Ruissian Manutfacturing 263 Table 8-6. Output, Employment Growth, and Capacity Utilization (percent) Significance tests vs. privatized non-SME Priva- De De Item State tized novo State novo SME Cuirrent performance Output growth,' 1993-1994.Hl -7.9 -18.7 3.7 0 ++ 0 (annualized) 0 + 0 Employment growth,a 1993- -9.1 -12.3 17.6 0 ++ - - 1994 (annualized) 0 ++ -- Capacity utilization, mid-1994 53.8 49.9 71.4 0 ++ 0 o ++ 0 Employment vacancy rate 2.1 1.5 8.4 0 ++ 0 o ++ 0 Expectationsfor next 6 months Expecting output increase 37.8 42.8 69.8 0 ++ 0 o ++ 0 Expecting employment 7.0 11.6 32.6 0 + 0 increase 0 0 0 Note: Significance tests: Output, employment growth, capacity utilization, vacancy rate - OLS without (line 1) ancl with (line 2) city + industry dummies. Log capacity utilization is tested. Output and employment expectations - raw data is ranked categorization (1 = high, 5 low, and so forth). Tests use ordered logit with/without city + industry dummies. a. Growth rates calculated and tested are log growth rates. These are converted to the usual growth rate format for presentation purposes. clearly not a size effect: the difference between de novo firms and other firms is statistically significant at the 1 percent level, whether controlling for just size or for industry and location as well, whereas SMEs (control- ling for ownership) have levels of unionization that are no different from those of larger firms. Economic Performance of De Novo Firms Here we consider the evidence on the recent performance of surveyed firms in the realms of output, employment, and investment, and also look Table 8-7. Job Creation and Job Destruction, Mid-1993 to Mid-1994 State-owned and privatized De novo State + privatized, De iovo, with average Total State- with average employment employment Activity sample owvned Prizatized 50 51-200 201-1,000 1,001-5,000 > 5,000 All 50 51-200 > 200 Job creation rate 1.5 1.4 1.5 0.0 0.9 0.5 1.7 1.3 25.0 26.5 17.3 33.5 Job destruction rate 7.8 6.2 8.8 14.4 12.2 14.6 11.1 4.8 6.1 7.7 9.5 1.4 Net growth rate -6.3 -4.8 -7.4 -14.4 -11.3 -14.0 -9.4 -3.5 19.0 18.8 7.9 32.1 Job creation share 100 36.2 55.3 0.0 0.6 2.6 38.4 49.4 8.5 1.4 2.7 4.4 job destruction share 100 31.5 68.1 0.1 1.7 14.2 48.1 35.5 0.4 0.1 0.3 0.03 Employment share 100 39.5 60.0 0.03 1.1 7.6 33.6 57.2 0.5 0.1 0.2 0.2 Percentage job creators 20.8 16.7 15.0 0.0 23.4 13.8 11.0 25.0 60.5 64.0 53.3 66.7 Number of firms 370 90 233 7 64 115 109 28 43 25 15 3 1993-94 employment Total 748,792 295,474 448,828 236 7,864 56,804 251,448 428,271 3,826 607 1,742 1,478 Mean 2,024 3,283 1,926 34 123 490 2,307 15,295 89 24 116 493 Note: Job creation rate = For a given group of firms, the total increase in employment in firms expanding employment between period t and period t + 1, as a percentage of average total employment in the group in periods t and t + l. Job destruction rate = For a given group of firms, the total decrease in employment in firms decreasing employment between period t and period t + 1, as a percentage of average total employment in the group inl periods t and t + 1. Net growth rate = For a given group of firms, the net increase in employment in all firms between period t and period t + 1, as a percent of average total employment in the group in periods t and t + 1. J job creation rate - job destruction rate. The net growth rate differs from the usual growth rate format in that the denominator is average rather than start-period employment; it is also not a simple mean of the sample firms but an aggregate growth rate (a weighted mean). Average employment = (Lt + Lt + 1)/2. Job creation share = Jobs created in a given group of firms as a share of total jobs created in the sample. Job destruction share = Jobs lost in a given group of firms as a share of total jobs lost in the sample. Figures for total sample include data for four firms with unknown ownership. Calculations for state, privatized, and de novo firms that use the total sample exclude these four firms. The Performance of De Novo Private Firms in Ruissian Manufactutrinig 265 at their expectations for the short-term (six months to one year). We begin with changes in real output7 in the first half of 1994 compared with 1993 (table 8-6). The mean response for privatized firms was a fall in output of 19 percent, which is fairly close to the corresponding official Goskomstat industrial production figure. In contrast, de novo firms grew by 4 percent. The growth rate is not as high as might be expected (perhaps because of underreporting of rapid growth), but the difference is still statistically sig- nificant, at levels of 1 percent to 5 percent, in a variety of formulations. By contrast, there is no size effect: once we control for ownership, we find SMEs to have growth rates that are no different from those of larger firms. The newly established private firms are also operating at significantly higher levels of capacity utilization. Average capacity utilization in mid- 1994 was 71 percent in de novo firms versus 51 percent in privatized and state-owned firms. Again, the difference was statistically highly signifi- cant under most formulations and controls; and again we observe no SME effects. We note in passing that the declines in capacity utilization reported by the firms in the survey are roughly consistent with their re- ported declines in real output. The differences between de novo firms and privatized and state firm,s are most dramatic with respect to employment growth. In mid-1994, employment in state and privatized firms in the sample had fallen by 9 percent and 12 percent, respectively, compared with a year earlier-- somewhat greater than the employment declines reported by Goskorn- stat. Employment in the newly established firms in our sample, by contrast, had increased by 18 percent between mid-1993 and mid-1994. What is more, when we control for size, the difference between de novo firms and the rest is even greater; that is, we find a negative size effect. Put another way, controlling for ownership, we find that SMEs are shed- ding employmentfaster than larger firms. In part, the differences in labor growth between privatized/state-owned and de novo firms reflect the amount of restructuring needed in the former category, where labor shedding has lagged the fall in output. Evidence from the survey sug- gests this process still had some way to go as of mid-1994. A useful way to analyze the dynamics of employment is the frame- work of "gross job flows," or job creation and job destruction. We report some basic statistics on gross job flows by ownership and firm size in ta- ble 8-7; the notes to table 8-7 provide definitions of the different indica- tors. The net employment declines8 in 1993-94 of 5-7 percent in the state 266 Corporate Governaance and Competition and privatized sectors were driven mostly by job destruction (shedding workers), with little offsetting job creation (hiring workers). Only 15-17 percent of state and privatized firms created jobs betwveen mid-1993 and mid-1994, and the job creation rate-that is, additional employment in these privatized/ state job creators-amounted to 1-2 percent of total em- ployment in these firms. Job destruction in shrinking privatized/state firms amounted to 6-9 percent of total employment in these firms. De novo firms look very different. The job destruction rate in these firms is actually little different from that in privatized/state firms, at 6 percent, and the de novo contribution to job destruction is roughly in pro- portion to their weight in the sample. Job destruction is actually fairly common in this group: 40 percent of the de novo firms shed workers dur- ing the survey period. Job creation, by contrast, occurred at the rapid rate of 25 percent in newly established firms. While the de flovO firms as a group account for less than one-half of one percent of total employment in the sample, they were responsible for 9 percent of all jobs created. The figures for the different size categories are also revealing.9 Job creation is much greater in both small (up to 50 employees) and medium-size (51- 200 employees) de novo firms, and job destruction somewhat less, than in comparably sized privatized and state-owned firms. Again, the dyna- mism of the newly established private sector does not appear to be strictly a result of small size, and the state-owned and privatized SME sectors look anything but dynamic. Expansion of de novo firms is also evident in the number of job vacan- cies that they report as open and vacant for more than two months (table 8-6). About one-third of both ownership groups report that they have such vacancies, but the number of vacant workplaces amounts to 8 per- cent of employment in the newly established firms, versus less than 2 percent in privatized/state firms. The regression analysis suggests the high vacancy rate is neither an SME phenomenon-privatized/state SMEs also have a much lower vacancy rate (again less than 2 percent)-nor related to industrial branch or location. That it is closely connected to expansion and growth is reflected in the reasons given by managers for not being able to fill the vacancies. Just over half of privatized/state and de niovO firms with vacancies blamed the absence of qualified applicants, but two-thirds of the former cited the main reason as the inability to pay competitive wages to attract workers, compared to one-third of de novo firms. The Performance of De Novo Private Firms in Russian Manuifactulring 267 Across various financial questions, de novo firms do not differ sigrLifi- cantly one way or the other from their state-owned and privatized coun- terparts, once we control for size and the other firm characteristics. They are no more or less likely to be financially distressed, for example; 11 per- cent of de novo firms said they were "usually loss-makers," compared with 14 percent of state and privatized finns. As reported in chapter 4 in this volume, while de novo firms have fewer arrears to creditors than state-owned and privatized firms, these differences do not survive the in- clusion of size, industry, and location controls. Chapter 5 reports that ease of obtaining bank credit is not significantly different for newly established private firms. The only important financial indicator where the difference between de novo firms and others was statistically strongly significant, even after controlling for firm characteristics, was in response to a ques- tion on what factors restricted fixed investment activity by the firms. De novo firms rated "poor financial situation of the firm" as a less important constraint on investment activity, even when controlling for size, indus- try, and location. Why these newly established private firms should look similar to their state-owned and privatized counterparts with respect to their financial situation, when their economic performance is clearly bet- ter, is unclear. One possibility is that these firms have a greater incentive to conceal strong financial performance-for example, because they are more likely to be targets of organized crime. Another possibility is that a financially "soft" environment blurs differences in financial performance and financial constraints across firms. More work is needed here. We turn finally to investment, past and present, in more detail. Firnis were asked to compare their current investment levels with levels in the pre-reform (1990/91) period. Not surprisingly, privatized and state- owned firms have seen large decreases in investment, with two-thirds re- porting falls of 10 percent or more, and 40 percent reporting investment levels of one-half or less of the pre-reform level. Only 15 of the de novo firms responded to this question, reflecting in part their recent founding; most of these reported investment levels that were the same or higher than in 1990/91. Firms were also asked whether they were currently in- vesting. Of the privatized and state-owned firms in the survey (the re- sponse rate to this question was 100 percent), 57 percent are investing, compared with 89 percent of newly established private firms. This differ- ence is statistically significant at the 1 percent level, controlling for size, 268 Corporate Governance and Competition location, and industry. We note that the size effect here is again negative: whether controlling for just ownership, or ownership, location, and in- dustry, SMEs are significantly less likely, not more likely, to be investing in fixed capital. Performance Expectations Not surprisingly, we find that de novo firms are far more optimistic re- garding future performance than their privatized or state counterparts (table 8-6). Managers were asked about their expectations of firm output and employment six months into the future-that is, at the end of 1994- according to broad categories (> 20 percent increase, 10-20 percent in- crease, and so forth). Managers of de novo firms clearly expected to expand output faster than privatized/state firms: close to half expected their firm's output to be more than 10 percent higher, and about 30 per- cent expected increases in excess of 20 percent, compared with about 25 percent and 13 percent, respectively, for privatized and state-owned firms. The differences are statistically highly significant even if we control for size, location, and industry. SME managers, by contrast, once we con- trol for ownership effects, were not expecting to expand output any more or less rapidly than their counterparts in larger firms. Two-thirds of managers of privatized/state firms forecast slow labor shedding or no change in employment by end-1994; fewer than 10 per- cent foresaw any increase in employment. On the whole, managers in de irovo private firms expected to expand employment, but by surprisingly little: fewer than one-fifth expected to increase employment by more than 10 percent; only one-third expected to create any jobs at all (that is, in- crease employment). The difference in employment expectations between the two groups is statistically significant when controlling for size, but disappears when controlling for location and industry as well. Why de )tovo firms are not expecting to increase employment at anything near the pace of the previous year (and yet are expecting strong output growth) is not clear. Finally, although the newly established private firms had better re- cords of recent and current investment, there is little difference in their fixed investment plans compared with privatized and state-owned firms. When asked if the firm was planning any fixed investment the next year, 78 percent of privatized/state firms said yes, compared with 83 percent The Peiformance of De Novo Private Firms in Ruissian Manrufactutring 269 of de novo firms. The difference is statistically insignificant, and remains so after controlling for size, location, and industry variation. Given the differences between the two groups in current investment activity, it is tempting to conclude that managers in privatized/state firms continued to be excessively optimistic compared with their counterparts in de nIovo firms, but verification of this awaits a resurvey of the firms. Russian De Novo Performance Compared with Their Polish Counterparts How do these findings compare with those from the World Bank survey of 200 Polish state-owned, privatized, and de novo private manufacturing firms, as reported in Belka and others (1995)? Output was growing in all the ownership categories in the Polish survey in 1993, reflecting the gen- eral economic recovery then under way. Output in the Polish de novo firms was growing extremely fast (60 percent yearly, on average, in 1993), much faster than in the new private firms in the Russian survey, and the differential between the de novo firms and the privatized and state-owned firms was greater than that observed among our Russian firms. Employ- ment growth was, however, very similar in the two surveys; in the Polish survey, de novo firms increased their employment by 23 percent in 1993 and privatized/state firms shed 7 percent of their employment, com- pared with 18 percent de novo employment growth and roughly 10 per- cent privatized/state labor shedding in 1994 in the Russian survey. The estimates of Russian managers of the amount of excess labor in both the newly established and the privatized/state firms were very simi- lar to those of managers in the corresponding Polish ownership catego- ries. Three-quarters of the de novo firms in both surveys reported they had no excess employment, compared with about half of the state-owned and privatized firms in both surveys. Nevertheless, there are some notewor- thy differences in the reasons given for not reducing excess employment. While the expectation of a recovery in demand and social/ethical reasons were the most common explanations given in both surveys, about one- quarter of Polish state-owned and privatized firms (but no Polish de nowa firms) cited workers' resistance to layoffs, versus only 1 percent of Rus- sian privatized/state firms (and again, no Russian de novo firms). We re- call here that Russian de novo firms are significantly less unionized than their privatized and state-owned counterparts, even when controlling for 270 Corporate Governance anid Competition size, industry, and location effects; this may be evidence that Russian newly established private firms are hostile to unions despite the generally lower levels of labor activism in Russia. Their Polish counterparts are still more hostile to unions; despite, or perhaps because of, a strong tradition of labor activism in Poland, none of the Polish de novo firms in the survey by Belka and others (1995) had any union representation at all. Reported profitability and financial health in general in Polish de novo firms was above average, a more clear-cut result than we found in the Russian newly established manufacturing firms in our survey. Finally, in- vestment levels are significantly higher in de 7lovo firms in both surveys. We are also able to compare our job creation/job destruction results with a similar analysis done for the entire Polish industrial sector in 1991 (Konings, Lehmann, and Schaffer 1996). The results of that study hint again at the dynamism of the new private sector. A word of caution is in order on the comparison of the results, however, because the Polish study was based on comprehensive (essentially census-style) data of almost the entire industrial sector with the exception of unincorporated (that is, small) firms. The comprehensiveness of the Polish data compared with our survey data means that selection bias issues-observing or surveying only successful or surviving firms-will likely be a greater danger in our sample. This is probably the main reason that the Polish study found higher rates of job destruction in both the state and private sectors (15-20 percent) than we find here. There are, nevertheless, a number of points of similarity in the two studies. The main difference between the state and "domestic 100 percent private" (the authors argue that this is composed mostly of de novo firms) sectors in the Polish study is in the much higher rate of job creation in the latter (at 18 percent, close to what we find in our Russian survey); job de- struction rates in the two categories are similar, as we find in the Russian sample. Job destruction rates are lower in larger state-owned firms, again, as we find. Job creation rates in the Polish state sector decrease with size, in contrast to our finding of no clear size/job creation rate relationship. Conclusions In most performance indicators, de ilOVO private manufacturing firms look better-or much better-than both privatized/ state firms in general, and privatized/state SMEs in particular. They are growing rather than con- The Performance of De Novo Private Firms in Russian Manufactutring 271 tracting, operating at higher rates of capacity utilization, expanding em- ployment rapidly and creating jobs, and investing more. Their expecta- tions about future performance are similarly more positive, with higher expectations for growth of output and employment, and more plamned investments. These differences do not appear to result from simple size effects-small firms expanding to fill the SME niche and/or survivorship effects-nor from simple industry or location effects-observing more en- trants in expanding sectors or locations. Indeed, controlling for owner- ship effects, we find that Russian SMEs are, if anything, less dynamic than their larger counterparts (and this in spite of the survivorship bias mentioned above, which ceteris paribuis would make the SME sector look stronger than it really is). Deeper explanations of the strong performance of newly established private firms are required. This is an area for future research. We offer here several observations regarding the implications of our findings for future investment and growth in the Russian manufacturing sector. First, while the de novo private sector in Russia is not negligible, it is clearly much smaller than its counterparts in the transition countries that began their liberalizations early. In Poland, in particular, the current rapid rates of growth in manufacturing are driven primarily by the de novo private sector. If the pattern of de novo-led growth is followed in Russia, then it may take a few years for manufacturing output growth to accelerate, as the de novo sector catches up in weight of total manufactur- ing output. Of course, Russia may not follow this pattern. Russia has the benefit of having rapidly privatized the bulk of its industrial firms. One possibil- ity is that privatization itself will lead to improved performance by firrns. Early research, however, does not yet indicate very significant perform- ance effects of privatization per se (see chapter 7 in this volume). The work by Barberis and others (1995) suggests that the performance of Rus- sian firms following privatization may depend greatly on the introduc- tion of new human capital. De novo firms are by definition mostly new bundles of human capital led by new entrepreneurs, and our findings of very dynamic de novo firms would tend to support this argument. This evidence has the additional advantage of bypassing endogeneity issues in identifying human capital effects in privatized firms.10 Following this line of reasoning, the performance of the privatized sector will depend in part on incentives to bring in new management, for 272 Corporate Governianice anid Competition example. But it will also depend on the ability of privatized firms to at- tract talented new people; it is an open question how successful privat- ized firms will be in hiring new talent, given the attractiveness to entrepreneurially minded people of starting and expanding their own firms. Privatization may, however, improve the growth prospects of the de novo sector by making it easier for rapidly expanding new businesses and new entrepreneurs to acquire privatized firms in part or in whole (and introduce new management, restructuring programs, and the like). In any case, further empirical work is needed. References Alimova, Tatyana, and others. 1995. "Trends of Development of Small- Scale Enterprises." Entrepreneuirial Activity in Ruissia 1: 1-16. Barberis, Nicholas, Maxim Boycko, Andrei Shleifer, and Natalia Tsukanova. 1995. "How Does Privatization Work? Evidence from Russian Shops." NBER Working Paper No. 5136, Cambridge, Mass. Belk'a, Marek, Saul Estrin, Mark E. Schaffer, and I. J. Singh. 1995. "Enter- prise Adjustment in Poland: Evidence from a Survey of 200 Pri- vate, Privatized, and State-owned Firms." Centre for Economic Performance Discussion Paper No. 233, London School of Eco- nomics. Johnson, Simon, and Gary W. Loveman. 1995. Starting Over in Eastern Euirope: Entrepreineurslhip and Economic Rene.val. Cambridge, Mass.: Harvard B-usiness School Press. Konings, Jozef, Hartmut Lehmann, and Mark E. Schaffer. 1996. "Job Crea- tion and Job Destruction in a Transition Economy: Ownership, Firm Size, and Gross Job Flows in Polish Manufacturing 1988-91." Centre for Economic Performance Discussion Paper No. 282, Lon- don School of Economics. (Forthcoming in Labour Economics.) Komai, Janos. 1990. The Road to a Free Economy. New York: Norton. Murrell, Peter. 1990. "Big Bang versus Evolution: Eastern European Eco- nomic Reforns in the Light of Recent Economic History." PlanEcon Report VI(26). Webster, Leila, and Joshua Charap. 1993. The Emergence of Private Manul- facturing in St. Petersburg. World Bank Technical Paper 241. Wash- ington, D.C. The Perfornmance of De Novo Private Firms in Rtussiani Manufacturing 273 Notes 1. The survey covered approximately 2,000 adults and -was undertaken in July 1995 by the Centre for the Study of Public Policy, University of Strathclyde. 2. The Russian convention is to refer to firms employing up to 200 persons as "small" firms, but for the purposes of this chapter we will call such firms "SMEs." 3. Statistical Yearbook of Russia 1994, pp. 75, 295; Statistical Yearbook of the Ruts- sian Federation 1992, p. 75. Further evidence on the small-scale business sector is provided by Alimova and others (1995), who report that Goskomstat had regis- tered more than 900,000 small-scale enterprises, employing 9 million people, at the beginning of 1995. 4. Or, if it does survive, its poorer financial state may make it less willing to take part in the survey. 5. We note here that the sampling procedure for the newly established firms imposed a constraint of a maximum of 50 percent to come from Moscow a,nd St. Petersburg (see the appendix to this volume). In the event, this constraint was not binding, because only one-quarter of the sample came from these two cities. Nev- ertheless, we cannot, for example, unambiguously rule out the possibility that firms from these two cities are overrepresented in the sample. 6. The significance of the difference between the wages of de ilovo firms and the wages in privatized firms survives the inclusion of the SME dummy (first line in each set of results in table 8-5) but not the addition of industry and location dummies (second line in each set of results). 7. Because of low response rates for reported real output growth, these growth rates are derived from nominal output and hence may differ from growth rates used in other chapters. 8. The calculation of growth rates in our job creation/job destruction analysis follows the conventions used in gross job flows literature, and hence differs slightly from that used elsewhere in the chapter and the volume. See the notes to table 8-7 for details. 9. Nota bene: the figures for de novo firms employing more than 200 persons are reported for the record only, since only three firms were in this category. 10. As Barberis and others (1995) point out, it can be difficult to separate cause and effect when good firm performance is correlated with introduction of new human capital; for example, a good privatized firm may attract good new manag- ers, or a good new manager can turn around a firm, or both. Appendix The World Bank Survey of 439 Industrial Enterprises Une J. Lee All the chapters in this volume are organized around a World Bank sur- vey of 439 Russian industrial enterprises conducted in mid-1994. The principal concern of the survey was to look at how enterprises were ad- justing to the shocks of economic transition. The survey not only docu- ments the evolution of financial and real variables but also providles information on factors that govern decisionmaking at the firm level. More important, this is the first comprehensive, randomly selected survey of Russian industrial enterprises, and therefore provides a unique opportu- nity for rigorous analysis of enterprise adjustment issues. The timing of the survey, furthermore, documents enterprise performance in Russia fol- lowing the longest spell of tight credit policy since the start of the eco- nomic reforms, and it captures the sweeping changes in ownership that have taken place since 1993. The focus here is to discuss the methodology of the sample selection, outline the general characteristics of the fieldwork, and provide more de- tailed information on the survey instrument. Some data issues are then explored. Finally, an overview of the survey results is provided. 275 276 Appendix Sample Selection The sample was essentially split, with a main sample of 400 industrial en- terprises and a separate sample of 50 de novo (newly established) private industrial/manufacturing firms. The main sample was drawn from two different populations of enterprises, the list of enterprises in the 1991 Goskomstat database of industrial enterprises and the 1991 Goskomstat database of industrial enterprises within the military-industrial complex (MIC).' The two databases together hold information on approximately 23,000 enterprises, including the name and address of the enterprise, as well as its branch, ministry, ownership code, employment, and other variables. The fifty de novo private enterprises were selected from lists at local statistical offices. The selection of the main sample was carried out by the World Bank, while the selection of the fifty de novo private enter- prises was done by the Russian Centre for the Study of Public Opinion (VTSIOM), the finn that carried out the fieldwork. The main sample was selected randomly, based on the following crite- ria. The sample was stratified by region as well as by industrial branch. All enterprises located in the territories that were represented by the con- sultants were extracted from the main database. Selected branches, in- cluding those that fell in the "other" branch category, were excluded. In addition, enterprises with fewer than fifteen employees were not included. An initial sample of 400 enterprises and an alternative sample of an additional 400 enterprises were selected randomly for the main sample of enterprises. The alternate sample was used to select substitute enterprises for those in the original sample. The main criterion for substitution was territory, with branch/sector replacement and size as the secondary crite- ria. When a replacement could not be found from the alternate list, VTSIOM was allowed to use their own database of industrial enterprises to select alternate enterprises, using the same criteria for replacement. VTSIOM was also instructed to select fifty de 1lovo private industrial enterprises that employed at least fifteen individuals and were owned by persons or groups of persons. No more than 50 percent of these enter- prises were to come from Moscow and St. Petersburg. These enterprises were selected from lists at industrial departments of local statistical offices. The World Bank Survey of 439 Indutstrial Enterprises 277 In the end, 439 enterprises, including all 50 de novo private enterprises, completed the survey. While the target was 400 enterprises from the main sample, 389 enterprises completed the survey. Of these, 60 percent of the enterprises came from the original sample, 25 percent from the alternate sample, and 15 percent were selected independently by VTSIOM. A num- ber of enterprises did not complete the quantitative section of the ques- tionnaire and therefore could not be included. Most of these enterprises, it was learned later, were part of the military-industrial complex. In gen- eral, the interviewers were confronted with particularly uncooperative at- titudes on the part of enterprise managers in six of the covered territories: Altai, Kemerov, Tyumen, Bashkorstan, Rostov, and Saratov. The Survey Instrument The interviews with managers were conducted by VTSIOM during June and July of 1994. A total of 160 interviewers took part in the survey. T he questionnaire consisted of two parts, a quantitative portion with 39 ques- tions and a qualitative segment with 82 questions.2 The qualitative por- tion was completed during interviews with either managers of enterprises or their deputies. The quantitative section was usually given to the accounting departments within each enterprise for completion. Fre- quently, however, the quantitative section was completed in an inter- viewer's presence. Interviews lasted an average of 2 hours, but ranged anywhere from 40 minutes to over 7 or 8 hours. The quantitative section is primarily made up of questions on production, output, and sales; mar- ket structure; corporate structure; employment and wages; and enterpri-se finances. Enterprises were asked about current information, as well as historical data going back to the pre-reform period. The qualitative sec- tion of the questionnaire included questions on adjustment, adjustment strategy, management strategy, market structure, finlance, future expecta- tions, and privatization. This section was designed to complement and supplement the information from the quantitative section, as well as to provide additional information. In general, the qualitative section of questionnaire had a better re- sponse rate than the quantitative section. Potentially sensitive questions were the most likely to be left unanswered. These areas included ques- 278 Appendix tions on the cost and profit structure of the firm and requests for financial details, particularly government financial transfers and bank loans. Over- all, the average response rate was 74 percent for the entire questionnaire, with an average of 66 percent for the quantitative section and 80 percent for the qualitative section. Classification An effort was made in all the chapters in this volume to use a consistent set of variable definitions, including definitions for size, branch, and ownership. The chapters use the following size definitions: small enter- prises are those with 200 or fewer employees; medium-size enterprises are those with between 200 and 1,000 employees; large enterprises are those with between 1,000 and 10,000 employees; and very large enter- prises are those with more than 10,000 employees. While the sample was limited to include only enterprises with fifteen or more employees, based on the information provided within the 1991 database for the main sam- ple and information provided at the local statistical offices for the de novo enterprises, the final sample of enterprises contained several with fewer than fifteen employees for both 1991 and 1994 employment. Branches are classified up to the three-digit standard branch codes for industry used in Russia. Given that the majority of the enterprises from the main sample were selected based on 1991 information, a few of the enterprises had changed their main products, and some are now classified within the category of "other" branches. Ownership is a particularly sensitive category, and a great deal of time was spent in determining this classification. Since all the enterprises from the main sample were selected using a 1991 database, with the ex- ception of those selected directly by VTSIOM, current ownership infor- mation was not known. All the chapters in this volume use three basic classifications of ownership-state-owned (SOE); privatized; and new private or de novo private-with the exception of chapter 7 by Earle, Estrin, and Leschenko. This chapter employs a modified version of the classification, including a more detailed definition for privatized enter- prises. Because more precise definitions are required for privatized firms-specifically privatized and majority worker-owned, manager- owned, or outsider-owned enterprises-privatized firms that could not The World Bank Sutrvey of 439 Indutstrial Enterprises 279 be classified as such were left unclassified. For the SOEs and privatized enterprises, information on their current and past corporate structure, privatization information, and data concerning enterprises' shareholder structure were used for ownership classification. After careful analysis of the evidence gained from firms' responses to questions on corporate structure, privatization, and shareholder struc- ture, some of the firms from the separate de novo sample were reclassified as privatized. In addition, a few enterprises from the main sample were classified as de novo private firms. Based on these changes, the samplle of de novo private firms was reduced to forty-five, with most (93 percent) coming from the separate de novo sample mentioned earlier. Overview of Sampled Enterprises The tables below outline the main characteristics of the final sample, in- cluding ownership, size, region, and branch distribution. The population averages are given alongside the sample averages for the main sample where applicable. The enterprises selected separately by VTSIOM for the main sample are included with the main sample for comparison with t:he population averages. The population averages are not known for de novo enterprises. A table with the distribution of enterprises associated with, or members of, the military-industrial complex is also provided. The sample picks up the large transfornation in the legal status of en- terprises (see table A-1). The majority of firms have moved to the private sector. Only slightly more than one-quarter of the main sample of enter- prises (excluding the de novo firms) remained in the state sector by mid- 1994. Table A-2 shows the size distribution of the sample. Not surprisingly, the vast majority of de novo private firms are small, with no Table A-1. Distribution of Sample by Ownership Owznership category Percentage offirms Number offirms State-owned enterprises 26 110 Privatized enterprises 63 272 New private/de novo enterprises 11 45 Unclassified enterprises 3 12 Number of enterprises 100 439 280 Appendix Table A-2. Distribution by Size of Enterprise (percent) Population Maini Main averages, Size of enterprise sample, sample, De novo, De novo, 1991, main (employnient) 1994a 1991b 1994a 1991b sample 200 24 15 93 91 44 200-1,000 36 30 7 9 37 1,000-10,000 36 48 0 0 17 > 10,000 3 7 0 0 2 Number of enterprises 327 332 43 11 10,582 Note: The population excludes enterprises with fewer than fifteen employees. a. Based on 1994 employment. b. Based on 1991 employment. Table A-3. Distribution over Industrial Branches (percent) Main sample (exciliding De novo Popuilationfor Induistrial branichl de novo) enterprises main sample 1. Energy 3 2 1 2. Fuels 4 0 1 3. Ferrous metals 3 0 1 4. Nonferrous metals 3 0 1 5. Chemicals/petrochemicals 6 9 3 6. Heavy machinery 9 7 4 7. Machine tool engineering 4 7 3 8. Automobiles 2 4 1 9. Agricultural machinery 4 0 1 10. Shipbuilding, aircraft, defense 17 4 4 11. Other machine-building and metalworking 11 13 12 12. Wood and paper 7 7 12 13. Construction materials and glass 5 24 9 14. Light industry 12 18 19 15. Food processing 8 2 21 16. Other branches 3 2 9 Number of enterprises 394 45 10,582 Thze World Bank Survey of 439 Industrial Enterprises 281 Table A-4. Sample Distribution over Regions (percent) Maiin sample (excliudinzg De novo Popillation for Region de novo) enterprises main sample Northern Arkhangelsk Oblast 2 0 3 Vologda Oblast 2 0 3 Northwestern St. Petersburg 7 9 5 Leningrad Oblast 3 0 2 Central Vladimir Oblast 4 9 3 Smolensk Oblast 2 0 2 Moscow 12 16 8 Moscow Oblast 3 0 9 Volgo-Vvatskiy N. Novgorod Oblast 6 7 6 Central Chernozemniy Voronezh Oblast 2 11 4 Povolzhskiy Samara Oblast 4 4 4 Saratov Oblast 4 4 4 Republic of Tatarstan 5 0 3 North Causasus Stavropol Territory 5 9 4 Rostov Oblast 4 0 6 Uralskiy Perm Oblast 5 9 5 Sverdlovsk Oblast 5 0 5 Rep. of Bashkorstan 5 4 4 Western Siberia Altai Territory 2 0 4 Kemerovo Oblast 4 0 3 Novosibirsk Oblast 5 4 4 Tyumen Oblast 2 0 2 Eastern Siberia Krasnoyarsk Territory 4 9 5 Far East Primorye Territory 3 4 3 Number of enterprises 394 45 10,582 282 Appendix Table A-5. Enterprises within the Military-Industrial Complex (MIC) MIC Percentage of all szurvey enterprises MIC 15 SOE 48 Privatized 47 De novo 2 Unclassified 3 Non-MIC 85 Number of enterprises 428 de novo private firms in the large and largest categories of enterprises. Even after excluding enterprises with fewer than fifteen employees from the population, it is clear in table A-2 that the survey enterprises from the main sample are skewed toward the large and largest enterprises. Com- pared with the population averages, small enterprises are underrepre- sented. This is consistent with the branch distribution of the surveyed enterprises (table A-3). While all of the main industrial branches are cov- ered in the survey, there are many more enterprises from the heavy in- dustrial sectors in the main sample relative to the population averages. The heavy industrial branches were historically far larger in employment in Russia than the light industrial sectors. Compared with the population averages for the main sample, the re- gions are fairly well represented proportionally, even in the regions inter- viewers considered uncooperative (see table A-4). Moscow and St. Petersburg, however, are somewhat overrepresented. With the exception of the northern economic region, de iovo enterprises were represented in all major economic regions. Finally, about 15 percent of all sampled enterprises indicated that they belonged to the military-industrial complex, as did 16 percent of the main sample of enterprises, excluding de novo private firms (see table A-5). This is a much higher rate than the population average of 7 percent for the main sample, and is again consistent with the size and branch distribu- tion of the sample. MIC enterprises are generally larger by employment and attached to the heavy industrial sectors of the economy. T7he World Bank Survey of 439 Indiustrial Enterprises 283 Notes 1. MIC enterprises are enterprises that have been classified as such by the Ministry of Internal Affairs. They are predominately associated with, but not lim- ited to, certain industrial branches such as shipbuilding, aviation, and defense in- dustry. 2. The survey instrument benefited enormously from a similar survey com- pleted in Poland in late 1993 as part of a World Bank research project, Enterprise Behavior and Economic Reform, headed by I. J. Singh. Index (Page numbers in italics indicate material infigures or tables.) Arrears, 87-88, 92-93; causes of, 126- Balance sheets, 91; Goskomstat data on, 28; correlates of (at the firm level), 134-36; of industrial sectors, 100, 101 116-17, 118, 119, 120, 121-28; esti- Bancruptcy, 89, 131-32 mate of real net flow of, 96; finan- Banking system, softness in, 7 cial indicators and, 121; in Bank loans and bank credit, 142, 154- financially distressed firms, 110-13, 57; arrears and difficulty in obtain- 114-15, 116; firm characteristics ing, 124-25; as bad debts, 7-8, 145; and, 119; frequency of, 124; indus- definition of "bad" or "overdue," trial sector and, 121; interest on, 94, 148; ease of obtaining, 156, 162; fac- 95-96; largest category of, 98; as tors affecting supply of, 155- 57, late payments, 128; liquidity and, 158; held by "bad" or financially 109-10; location of firm and, 122; distressed firms, 150-54; most cem- measuring, 94-97; in Russian in- mon problem in obtaining, 154; dustry, 102; term structure of, 122, overdue, 7, 142, 145, 146, 147-53 123; time trends of, 99, 102-3. See Banks: arrears to, 123-24, 133; bad also Bank arrears; Enterprise ar- debts owed to, 7-8, 145, 150-54; rears; Overdue payables; Tax ar- commercial, 141-42; as creditcrs, rears; Trade credit arrears; Wage 163; cross-holdings between firms arrears and, 144-45; enterprise decisions Arrears crisis, 102, 108 and, 157, 159, 160, 161, 162-63; as investors and shareholders, 11, 14L5, Bad debts, 88, 91, 152, 162; vs. late pay- 161, 162; overdue liabilities to, 162; ments, 89-91; owed to banks, 7-8, risk exposure of, 142, 162; with 145, 150-54; rolling over banks', shares held by firms, 144-45, 156- 148, 149 57 Bad debt stock problem, 90 Benefits. See Social benefits to workers; Balance sheet items, data on, 93-94, 95 Workers' compensation 285 286 Index Bulgaria, financial transfers by govern- Depoliticization, 249 n.2 ment to enterprises in, 167 Directed state credits (DSCs), 141, 144- 45,171 Capacity utilization, 2, 21-22, 238, 239- 40, 241 Eastern European market for Russian Capital stock, vintage of, 240, 242 enterprises, 189 Childcare facilities as benefit, 55, 57, 58, Employee Stock Ownership Plan 71, 75-76 (ESOP), 214 Collateralization of loans, 143-44, 148, Employment, 23-24, 34; excess, 24, 26, 162 27; growth of, 264, 265-66; indus- Compensation. See Workers' compen- trial, 188; output and, 16, 17-18, 19, sation 20-21; in private sector, 206; priva- Council for Mutual Economic Assis- tization and, 241, 242, 243. See also tance (CMEA), 188-89 Jobs Courts to collect overdue debts, 130-31 Employment stability, 4, 10, 45; wages Credit on preferential terms, 155 and, 15 Credit. See Directed State credits (DSC) Enterprise decisions: banks' influence Credit control, 128-32 on, 157, 159, 160, 161, 162-63; by in- Creditworthiness, 155-56 siders, 4, 19 Czech Republic: govemment financial Enterprise ownership, 1, 214; behavior transfers to enterprises in, 180; tax of the enterprise and, 206-12, 225- arrears in, 107; trade credit in, 103, 33; control and, 215-21; 221-25; em- 104, 106 ployment policies and, 234,236; enterprise performance and, 236-41, Debt: courts to collect, 130-31; pay- 243; by workers, 246-47. See also ment periods for, 96, 106. See also Privatization Arrears; Bad debt; Bank loans; Enterprise restructuring, 209-11; short- Credit; Late payments term, 211-12 Debtors, insolvency of, 131 Enterprises: balance sheet structure of, De novo private firms, 253-58, 270-72; 95, 97-99; bank shareholding in, with bank credit, 142, 143; bank 145, 161, 162; corporate control in, credit for, 157; benefits and com- 215-21; Eastern European market pensation offered by, 5, 67, 77, 241, for products of, 189; government 261-63; capacity utilization by, 238, subsidies and restructuring of, 196, 239; characteristics of, 258-63; eco- 230-31; insider ownership of, 206; nomic performance of, 263, 264, liabilities of, 97; ownership effects 265-68; equipment of, 258-60; loans in, 31-35; ownership forms of, 1; for productive fixed income invest- worker compensation and restruc- ment and, 143; performance expec- turing of, 53; worker ownership of, tations for, 268-69; performance of 8, 246-47. See also De t7ovo firms; Russian compared to Polish, 269- Firms; World Bank survey of enter- 70; pre- and post-reform invest- prises ment by, 267-68; as shareholders in Enterprise sector, developing a politi- banks, 144; size of, 9, 238; trading cally independent market-oriented, partners of, 261 208-9 Inidex 267 Financial distress, 90-91, 106-7; arrears Goskomstat data, 134-36 and, 110-13, 114-15, 116; payment Government-directed credit. See Di- priorities and, 127-28; tax arrears rected State Credits (DSCs) and, 128 Government financial assistance, most Financial stress, 91 commonly requested forms.;-f, 155 Firms: with bank credit, 142, 143, 159, Government financial transfers to en- 162; bargaining on wages and em- terprises, 5-6, 71-73, 166-71, 197- ployment and, 39-45; benevolence 200, 230-32; bargaining position of, 10; budget constraints of, 5-8; with government and, 190, 198; cross-holdings between banks and, concentration of, 176, 177, 178-82, 144-45; decisionmaking in, 4, 19, 197; difficulties in estimating, 168- 29-31, 48, 157, 159, 160, 161, 162-63; 70, 183; easiest way to obtain, 193; depoliticization of, 208-9; employ- enterprise performance and, 194- ment strategies of, 234, 235, 236; 96; implicit government objectives evolution of governance forms of, and, 182-85, 186, 187, 188-90, 191, 212-13; financially distressed, 150; 192-94; nature of, 171, 172-73, 1.74, financial management strategies of, 175-76; policy recommendations 129; that hold bank credit, 157, 159; regarding, 198-99. See also Directed investment strategies of, 235, 236; state credits; Subsidies to firms with lending banks as sharehold- Government procurement, 185, 187, ers, 145; location of, 39, 122; manag- 189 ers of, 8, 9, 10-11; marketing Government policies, market-restrict- strategies of, 235, 236; need to sup- ing, 184-85, 186, 187 port customers and, 131; non-bank loans to, 143; objectives of, 35-37, Health facilities as benefit, 55, 57, 58, 38, 39, 208, 233-34, 236-41, 243; 75-76 with overdue bank credit, 152-53; Housing as benefit, 56, 57, 58, 59, 60, ownership of, 9, 21-22, 208-9, 212- 61, 63, 64; cost recovery and, 70; la- 13, 221-25; ownership impact on bor mobility and, 55 performance of, 236-41, 243; pay- Hungary: bancruptcy in, 89; bank ment priorities of, 127; performance credit in "bad" firms in, 152; con- of new private, 9; problems with trol of overdue receivables in, 129- accounting systems of, 139 n.12; 30; overdue bank credit in, 147--48; production strategies of, 234, 235; tax arrears in, 107, 116; trade credit response to difficulty in obtaining in, 103, 104, 106 bank loans, 154; as shareholders in lending banks, 144-55, 156-57; tax Incentive payments for workers, 78 arrears of financially distressed, Industrial output, 21-22; restructuring 6; workers' role in, 9, 30; use of of Russian industry and, 2 loans by, 143. See also De nIovo firms; Industrial sectors, balance sheets of, Enterprises 100, 101 Firms' performance, effects of owner- Inflation, 94-95, 96, 103, 139 n.18, 183; ship and control on, 206 bank loans and, 143, 147, 163; meas- Fund of Workers' Shares (FARP), 214 ures to control, 109; trade credit ar- rears and, 108 288 Inidex Insiders: enterprise ownership and, Outsider shareholders, 9 206; firm-level decisions by, 4, 29; Overdue payables, term structure of, influence of, 37, 39; rent-taking by, 122, 123 4 Overdue receivables, firms' control of, Interest arrears, 147; capitalization of, 128-29 162 Overdue receivables/sales ratio, 96 Interenterprise debt. See Trade credit Investors, banks as, 11, 145, 161, 162. Payment periods, 96 See also Shareholders Payments gridlock, 108 Poland: bank debt in "bad" firms in, Jobholding, multiple, 4, 79 152; control of overdue receivables Jobs, 264, 265-66. See also Employment in, 129-30; de niovo-led growth in, 271; government financial transfers Labor: cost of, 77; excess, 3; firms' un- to enterprises in, 180; overdue willingness to shed, 15; flows of, 24; liabilities to banks in, 148, 150; pay- part-time, 5; productivity of, 19, 20- ment priorities of financially dis- 21, 44, 49; time allocation and, 79- tressed firms in, 128; tax arrears in, 81. See also Unions; Workers 107, 116; trade credit in, 103, 105, Labor hoarding, 24-25, 27, 195 106; wage arrears in, 133; workers' Labor mobility, 77, 78; housing as benefits in, 67-68 benefit and, 55 Price controls, 184-85, 186, 187, 198, Labor productivity, 195 228, 229 Late payments, 88, 91; arrears as, 128; Principal, rescheduling of, 162 vs. bad debts, 89-91. See also Arrears Private sector, informalization of, 5 Legal reforms needed, 132 Privatization, 205-6, 243-47; employ- Legal system, reforms needed in, 132. ment and, 48, 241, 242, 243; firm be- See also Courts havior and, 8-9; institutional Liabilities in balance sheets, 91-92; features of Russian, 213-15. See also overdue, 92-93 Enterprise restructuring Liquidity, arrears and, 109-10, 128 Privatization program, ownership forms Liquidity problems, 7 of enterprises following the, 1 Loans, collateralization for, 143-44. See Productivity, wages and, 44, 49 also Bad debt; Bank loans; Credit Product mix, changes in, 3 Local governments' transfers to enter- Profit, 238; as firms' objective, 36; firms' prises, 171, 175 response to deterioration of, 71 Managers, 8, 9, 10-11; inside, 35 Receivables/sales ratio, 96 Military-industrial complex, govern- Reform: in enterprise and financial sec- ment financial transfers to, 189 tors, 140-41; needed legal system, Military procurement, payment delays 132 by government and, 121 Rent-seeking behavior by enterprise managers, 193 New firms. See De nov?o firms Restructuring, 10; measures of, 2 New Zealand, uncollectible taxes in, Romania, financial transfers by govern- 107 ment to enterprises in, 167 Index 289 Sales, wages and, 47 Tax evasion, 5 Shareholders: employment reductions 30:70 rule, 125 and insider, 34; outsider, 9; worker Trade credit, 87; international compari- layoffs and outsider, 32-33. See also sons in, 103, 104-5, 106-8; overdue, Investors 7; in transition countries, 103, 104- Shocks faced by Russian firms, 2; com- 5, 106; volume of, 129 pensation for, 188-89 Trade credit arrears, 87, 98, 103, 132; Slovakia, tax arrears in, 107 macroeconomic policy and, 108-9; Social benefits and services to workers, narrow range of, 109; in transition 4, 55-59, 60, 61, 62, 63-68; asset economies, 104-5, 106 structure and, 57-58; costs of, 68- Transition countries, trade credit in, 70, 73; de novo firms and, 61, 63; en- 103, 104-5, 106 terprise ownership and, 61, 65; financing of, 5; by firm setting, 64; Unemployment, 5; low rate of, 3, 79; firm size and, 58-59, 63, 64-65; threat of, 77 firms' spending on, 53; increase in Unemployment benefits, 3, 76, 77 market value of, 74; by industrial Unions, 31 sector, 60; loss of, 71; offsetting costs of, 71; reasons for continuing, Wage arrears, 7, 44, 132-33; wage level 57, 76; recovery of costs of, 70-73; and, 125 by region, 62, 63; results of provid- Wage decisions, 31 ing, 53-54; subsidies and, 188; un- Wage differentials, 78 derestimation of Wage to gross profit per worker ra,tio, Social benefits and costs of, 73; wage 45 level and, 66. See also Workers' Wage to gross surplus per worker ra- compensation tio, 44, 46 Stock/flow ratio, 96 Wages, 3, 241-42; bargaining on, 39--45; Stock problems, 90 consumer prices and, 73; firms' Subsidies to firms, 5-6, 7, 171, 172, 174; evolution and, 27-29; firms' finan- compensatory, 183, 184-85, 188; cial characteristics and, 43, 45; low, forced, 184, 189-90, 192-94; privati- 27-29; output and, 19; outside zation and, 215; protectionist, 184, shareholders and workers', 34; pro- 188-89; strategic, 184, 188-89. See ductivity and, 44, 49; sales and, 47; also Government financial transfers setting, 79; in state and privatized to enterprises firms, 77; traded for employment Survey. See World Bank survey of in- stability, 15. See also Social benefits dustrial enterprises and services to workers; Workers' compensation Tax arrears, 6, 90, 107, 108, 116, 123-24, Wage-tax regulations, 54-55 133, 175-76, 197, 229-30; financial Worker effort, 77 distress and, 128 Worker-manager harmony, 48 Tax benefits, 171 Workers: attachment to original firms, Tax breaks and exemptions, 155, 175 4-5, 78; bargaining power of, 39; Tax credits, 71 benevolent retention of, 3; decision- Taxes: excess wage, 54, 73; payroll, 55 making by (in firms), 8; incerntive 290 Index Workers (continuted) Worker welfare as firms' objective, 35- payments for, 78; layoffs of, 23-24, 37, 38, 39 32; motivating, 78; ownership of World Bank survey of industrial enter- firms by, 8, 246-47. See also Labor prises, 275; classification of the en- entries terprises and, 278-79; overview of Workers' compensation, 45, 75-76; en- the enterprises, 279-80, 280, 281, terprise restructuring and, 53; 282; sample instrument used in, monetary, 27, 73; nonmonetary, 4, 277-78; sample selection, 276-77 29, 49. See also Social benefits and services to workers; Wage entries Distributors of World Bank COTE D'IVOIRE INDONESIA Publications Centre d'Edition et de Diffusion Pt. Indira Limited Prices and credit terms vary Arom Aficaines (CEDA) Jalan Borobudur20 vayrm 04 BP. 541 P.O. Box 181 couintry to country. Consult your Abidtjan 04 Plateau Jakarta 10320 local distributor before placing an order. CYPRUS IRAN ALBANIA Center For Applied Research Ketab Sara Co. Publishers Adron LtdN Cyprus College Khaled Eslamboli Ave., PeIa Rexhepi Str. 6,Diogenes Street, Engomi 6th Street Pall. 9, Shk. 1, Ap. 4 Nicosia RO. Box 15745-733 Kusheh Delafrooz No. 8 Tirana CZECH REPUBLIC Tehran ARGENTINA National Information Center Oficina del Libro Internacional prodejna, Konviktska 5 Kowkab Publishers Av. Cordoba 1877 CS -113 57 Prague 1 P.O. Box 19575-511 1120BuenosAires DENMARK Teran AUSTRALIA, FIJI, PAPUA NEW SamfundsLitteratur IRELAND GUINEA, SOLOMON ISLANDS, RosenoemsAlle11 Government Supplies Agency GUINEAT, SOLMND WESTRNDSA A DK-1970 Frederiksberg C Oifig an tSolathair VANUATU, AND WESTERN SAMOA 45HrolRa 648 WhitehorseaRoad EGYPT, ARAB REPUBLIC OF Dublin- 2 848 Whitehorse Road ~Al Ahram Distribution Agency Mitcham3132 Al Galaa Street ISRAEL Victoria Cairo Yozmot Literature Ltd. AUSTRIA P.O. Box 56055 AUSTRIA The Middle East Observer 3 Yohanan Hasandlar Street Gerold and Co. 41, Sherit Street Tel Aviv 61560 Weihburggasse 26 Cl A-1011 Wien R.O.Y International BANGLADESH FINLAND PO Box 13056 Micro Industries Development Akateeminen Kiiakauppa Tel Aviv 61130 MsicoIdstrneoies y DeeoMentS P.O. Box 1 28 Assistance Society (MIDAS) FIN- Helsinki Paiestinian Authority/Middle East Dhou di RoAd R1A Index Information Services Dhanmondi R/Area FRANCE P.O.B. 19502 Jerusalem Dhaka1209 Word Bank Publications BELGIUM 66, avenue d'1na ITALY Jean De Lann 75116 Paris Licosa Commissionaria Sansoni SPA JeAn de Lanoy 2Via Duca Di Calabna, 111 Av. du Rol 202 GERMANY Casella Postale 552 1060 Brussels UNO-Vedag 50125 Firenze PoppelsdorferAllee 55 BRAZIL ~~~~53115 Bonn JAMAICA Publicacoes Tecnicas Intemacionais Ltda. Ian Randle Publishers Ltd. Rua Peixoto Gomide, 209 GREECE 206 Old Hope Road 01409 Sao Paulo, SP. Papasotidou S.A. Kingston 6 CANADA 35, Stoumara Str. Renouf Publishing Co. Ltd. 106 82 Athens JAPAN Eastern Book Service 5369CanotekRoad HONG KONG, MACAO 3-13Hongo3-chome, Bunkyo-ku Ottawa, Ontario KlJ 9,1 Asia 2000 Ltd. Tokyo 113 CHINA Sales & Circulation Department China Financial & Economic Seabird House, unit 1101-02 KENYA Publishing House ~~~22-28 VVyndham Street, Central Atfica Book Service (E.A.) Ltd. 8, Da Fo Si Dong Jie Hong Kong Quaran House, Mfangano Street Beijing P.~~~~~~~~~~~~~~O. Box 45245 Beifig 9HUNGARY Nairobi Foundation for Market Economy COLOMBIAtd 112 PF249 KOREA, REPUBLIC OF Carrera 6 No. 51-21 1519 Budapest Daejon Trading Co. Ltd. ApartadoAereo34270 ~~~~~~~~~~706 Seoun Bldg Apartafedeo Aer ta 3 0 INDIA 44-6Youido-Dong, Yeongchengpo-Ku Santatd de Bogot6, D.C. Allied Publishers Ltd. Seoul 751 Mount Road Madras - 600 002 L Wennergren-WilliamsAB ugal P 0. Box 1305 -- s1 S-17125 Solna Seoul Rua Oo Carmo 70-74 1200 Lisbon SWITZERLAND MALAYSIA Librairie Payot University oi Malaya Cooperatwe ROMANIA Serviceinstitutionnel Bookshop, Limited Compani De Ubrarii Bucuresti S.A. Cotes-de-Montbenon 30 P.O. Box 1127 Str. Lipscani no. 26, sector 3 1002 Lausanne Jalan Pantai Baru Bucharest 59700 KualaLumpur AOECO Van Dierren Editions RUSSIAN FEDERATION Techniques MEXICO tsdatelstvo cVes Mir> Ch. de Lacuez 41 INFOTEC 9a, LolpachniyPereulok CH1807Blonay Apartado Postal 22-8B0, T)aipan Moscow 101831 14030 Mexico, D.F TANZANIA SAUDI ARABIA, QATAR Oxford University Press NETHERLANDS Jarir Book Store Maktaba Street De Lindeboorntn0r-Publikahies PO. Box 3196 PO Box 5299 P.O. Box 202 Riyadh 11471 Dar es Salaam 7480 AE Haaksbergen SINGAPORE, TAIWAN, THAILAND NEW ZEALAND MYANMAR, BRUNEI Central Books Distribution EBSCO NZ Ltd. Asahgate Publishing Asia 306 Silom Road Private Mail Bag 99914 Pacific Pte. Ltd. Bangkok New Market 41 Kallang Pudding Road #04-03 Auckland Golden Wheel Building TRINIDAD & TOBAGO, Singapore349316 AND THE CARRIBBEAN NIGERIA Systematics Studies Unit University Press Limited SLOVAK REPUBLIC 9 Watts Street Three Crowns Building Jericho Slovart G.T.G. Ltd. Curepe Private Mail Bag 5095 Krupinsla 4 Trinidad, West Indies lbadan PO Box 152 852 99 Bratislava 5 UGANDA NORWAY Gustro Ltd. NIC Info A'S SOUTH AFRICA, BOTSWANA PO Box 9997 Book Department Oxford University Press Madhvani Building P.O. Box 6125 Etterstad Southern Alrca Plot 1614 Jinja Rd. N-06020slo6 P.O. Box 1141 Kampala CapeTowrn 8000 PAKISTAN UNITED KINGDOM Mirza Book Agency Intemational Subscription Service Microinfo Ltd. 65, Shahrah-e-Quaid-e-Azam PO. Box 41095 RO. Box 3 Lahore 54000 Craighall Alton, Hampshire GU34 2PG Johannesburg 2024 England Oxford University Press 5 BangaloreTown SPAIN VENEZUELA Sharae Faisai Mundi-Prernsa lbrcs, S.A. Tecni-Ciencia Ubros, S.A. Karachi-75350 Castello 37 Centro Cuidad ComercialTamanco 28001 Madrid Nivel C2 PERU Caracas Editorial Desarrollo SA Mundi-Prensa Barcelona Apartado3824 Cori%stl deCent, 391 ZAMBIA Lima 1 08009 Barcelona University Bookshop University of Zambia PHILIPPINES SRI LANKA, THE MALDIVES Great East Road Campus International Bookscurce Center Inc. Lake House Bookshop PO. Box 32379 Suite 720, Cityland 10 100, SirChittampalam Gardiner Lusaka CondominiumTower2 Mawatha H.V dela Costa, comer Valero St. Colombo2 ZIMBABWE Makati, Meto Manila Lonrgrnan Zimbabwe (PIe.)Ltd. SWEDEN TourleRoad,Ardbennie POLAND CE Fritzes AB RO. Box ST125 International Pubtiahing Service Regeringsgaton 12 Southerton, Ilarare UI. Piekna 31/37 S-106 47 Stockholm 00-677Wat2awa Headquarters 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Telephone: (202) 477-1234 Facsimile: (202) 477-6391 Telex: Mci 64145 WORLDBANK Mci 248423 WORLDBANK Cable Address: INTBAFRAD WASHINGTONDC World Wide Web: http://www.worldbank.org E-mail: books@worldbank.org European Office 66, avenue d'Iena 75116 Paris, France Telephone: (1) 40. 69. 30. 00 Facsimile: (1) 40. 69. 30.66 Telex: 640651 Tokyo Office Kokusai Building 1-1, Marunouchi 3-chome Chiyoda-ku, Tokyo 100, Japan Telephone: (3) 3214-5001 Facsimile: (3) 3214-3657 Telex: 26838 I51I _~~~~~~~~~~~IB -2332-