34428 Privatization in Latin America MYTHS AND REALITY Edited by Alberto Chong Florencio López-de-Silanes INTER-AMERICAN DEVELOPMENT BANK STANFORD UNIVERSITY PRESS Privatization in Latin America Privatization in Latin America MYTHS AND REALITY Edited by Alberto Chong Florencio López-de-Silanes A COPUBLICATION OF STANFORD ECONOMICS AND FINANCE, AN IMPRINT OF STANFORD UNIVERSITY PRESS, AND THE WORLD BANK © 2005 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org E-mail: feedback@worldbank.org All rights reserved. 1 2 3 4 08 07 06 05 A copublication of Stanford Economics and Finance, an imprint of Stanford University Press, and the World Bank. Stanford University Press The World Bank 1450 Page Mill Road 1818 H Street, NW Palo Alto, CA 94304 Washington, DC 20433 The findings, interpretations, and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the le- gal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank, 1818 H Street, NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. ISBN 0-8213-5882-0 (World Rights except North America) ISBN (soft cover) 0-8047-5242-7 (North America) ISBN (hard cover) 0-8047-5241-9 (North America) Library of Congress Cataloging-in-Publication Data Privatization in Latin America : myths and reality / Alberto Chong, Florencio López- de-Silanes [editors]. p. cm.--(Latin American development forum) Includes bibliographical references and index. ISBN 0-8213-5882-0 1. Privatization--Latin America. I. Chong, Alberto. II. López-de-Silanes, Florencio. III. Latin American development forum. HD4010.5.P754 2005 338.98'05--dc22 2004062831 Latin American Development Forum Series This series was created in 2003 to promote debate, disseminate informa- tion and analysis, and convey the excitement and complexity of the most topical issues in economic and social development in Latin America and the Caribbean. It is sponsored by the Inter-American Development Bank, the United Nations Economic Commission for Latin America and the Caribbean, and the World Bank. The manuscripts chosen for publication represent the highest quality in each institution's research and activity out- put and have been selected for their relevance to the academic community, policymakers, researchers, and interested readers. Advisory Committee Members Inés Bustillo, Director, Washington Office, Economic Commission for Latin America and the Caribbean, United Nations Guillermo Calvo, Chief Economist, Inter-American Development Bank José Luis Guasch, Regional Adviser, Latin America and the Caribbean Region, World Bank Stephen Haber, A. A. and Jeanne Welch Milligan Professor, Department of Political Science, Stanford University; Peter and Helen Bing Senior Fellow, the Hoover Institution Eduardo Lora, Principal Adviser, Research Department, Inter-American Development Bank José Luis Machinea, Executive Secretary, Economic Commission for Latin America and the Caribbean, United Nations Guillermo E. Perry, Chief Economist, Latin America and the Caribbean Region, World Bank Luis Servén, Lead Economist, Latin America and the Caribbean Region, World Bank About the Contributors Francisco Anuatti-Neto is a professor in the Department of Economics at the Universidade de São Paulo and FIPE (Fundação Instituto de Pesquisas Econômicas), Brazil. Milton Barossi-Filho is a professor in the Department of Economics at the Universidade de São Paulo and FIPE (Fundação Instituto de Pesquisas Econômicas), Brazil. Katherina Capra is a researcher at the Unidad de Análisis de Políticas Sociales y Económicas (UDAPE), La Paz, Bolivia. Alberto Chong is a senior research economist in the Research Department at the Inter-American Development Bank, Washington, D.C. Ronald Fischer is a professor in the Department of Economics at the Universidad de Chile, Santiago. Sebastián Galiani is a professor in the Department of Economics at the Universidad de San Andrés, Buenos Aires, Argentina. Mauricio Garrón is a coordinator at the Organización Latinoamericana de Energía, Quito, Ecuador. Paul Gertler is a professor at the Haas School of Management, University of California at Berkeley. Antonio Gledson de Carvalho is a professor in the Department of Eco- nomics at the Universidade de São Paulo and FIPE (Fundação Instituto de Pesquisas Econômicas), Brazil. Rodrigo Gutiérrez is a professor in the Department of Economics at the Universidad de Chile, Santiago. Florencio López-de-Silanes is a professor in the School of Management at Yale University and an associate with the National Bureau of Economic Research, Cambridge, Mass. vii viii ABOUT THE CONTRIBUTORS Roberto Macedo is a professor in the Department of Economics at the Universidade de São Paulo and FIPE (Fundação Instituto de Pesquisas Econômicas), Brazil. Carlos Machicado is a researcher and PhD student in the Latin American Program of the Universidad de Chile (Santiago)/Instituto Tecnológico Autónomo de México (Distrito Federal)/Universidad Torcuato di Tella (Buenos Aires, Argentina). Carlos Pombo is a professor in the Department of Economics, Universidad del Rosario, Bogotá, Colombia. Manuel Ramírez is a professor in the Department of Economics at the Universidad del Rosario, Bogotá, Colombia. Ernesto Schargrodsky is a professor in the Business School at the Univer- sidad Torcuato Di Tella, Buenos Aires, Argentina. Pablo Serra is a professor in the Department of Economics at the Univer- sidad de Chile, Santiago. Federico Sturzenegger is dean of the Business School at the Universidad Torcuato Di Tella, Buenos Aires, Argentina. Máximo Torero is a senior researcher at Grupo de Análisis para el Desar- rollo (GRADE), Lima, Peru, and International Food Policy Research In- stitute, Washington, D.C. Contents Foreword xvii Acknowledgments xix 1. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 1 Alberto Chong and Florencio López-de-Silanes 2. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA: A MICROECONOMIC ANALYSIS 67 Sebastián Galiani, Paul Gertler, Ernesto Schargrodsky, and Federico Sturzenegger 3. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 117 Katherina Capra, Alberto Chong, Mauricio Garrón, Florencio López-de-Silanes, and Carlos Machicado 4. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 145 Francisco Anuatti-Neto, Milton Barossi-Filho, Antonio Gledson de Carvalho, and Roberto Macedo 5. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 197 Ronald Fischer, Rodrigo Gutiérrez, and Pablo Serra 6. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 275 Carlos Pombo and Manuel Ramírez ix x CONTENTS 7. PRIVATIZATION IN MEXICO 349 Alberto Chong and Florencio López-de-Silanes 8. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 407 Máximo Torero Acronyms and Abbreviations 479 Index 483 FIGURES 1.1 Economic Activity of State-Owned Enterprises, 1978­97 4 1.2 Revenues from Privatization in Latin America, 1990­2000 7 1.3 Availability of Privatization Data on Latin America 16 1.4 Profitability Changes after Privatization in Latin America 20 1.5 Operating Efficiency Changes after Privatization in Latin America 21 1.6 Percentage Changes in Employment after Privatization in Latin America 22 1.7 Median Changes in Output after Privatization in Latin America 24 1.8 Net-Income-to-Sales Gap between Privatized and Private Firms before and after Privatization 25 1.9 Cost-per-Unit Gap between Privatized and Private Firms before and after Privatization 26 1.10 Median Real and Industry-Adjusted Changes in Wages after Privatization 30 1.11 Transfers from Workers as a Percentage of Increased Profitability after Privatization 31 1.12 Median Changes in Profitability of Privatized Firms in Competitive and Noncompetitive Industries in Latin America 33 1.13 Changes in Employment and Output of Privatized Firms in Competitive and Noncompetitive Industries in Latin America 35 1.14 Rehiring after Privatization, by Region 48 2.1 Percentage of Accumulated Income from Privatizations, 1990­98 73 2.2 Logarithm of Population Connected to the Water Network and Fitted Values, Aguas Argentinas, 1980­99 93 CONTENTS xi 2.3 Displaced Workers' Earnings Rents: 1991 and 2001 102 4.1 Formal Employment before and after Privatization, 1995­99 166 5A.1 Cost per Unit before and after Privatization, Adjusted and Unadjusted 251 5A.2 Investment as a Fraction of Physical Assets (PPE) before and after Privatization, Adjusted and Unadjusted 254 5A.3 Investment as a Fraction of Sales before and after Privatization, Adjusted and Unadjusted 257 5A.4 Net Income as a Fraction of Physical Assets before and after Privatization, Adjusted and Unadjusted 260 5A.5 Operating Income as a Fraction of Physical Assets before and after Privatization, Adjusted and Unadjusted 262 5A.6 Operating Income as a Fraction of Sales before and after Privatization, Adjusted and Unadjusted 264 6.1 Markup Rates for IFI Sample and Total Manufacturing, 1970­98 298 6.2 Total Factor Productivity Indexes for IFI Sample and Total Manufacturing, 1970­98 300 6.3 Investment Rates for IFI Sample and Total Manufacturing, 1970­98 301 6.4 Labor Productivity Indexes, IFI Sample and Total Manufacturing, 1970­98 303 6.5 Thermal Capacity versus Thermal Generation, 1991­99 312 7.1 Gap between Privatized Firms and Private Firms 358 7.2 The Overall Fiscal Impact of Privatization 369 7.3 Total Net Debt of the Public Sector 371 7.4 Deregulation Actions Taken before Privatization 381 7.5 Mexican Banks' Profitability Indicators before and after Privatization 383 7.6 Mexican Banks' Performance Indicators before and after Privatization 384 7.7 Terms of Related and Unrelated Loans 387 7.8 Default and Recovery Rates: Related and Unrelated Loans 388 7.9 Market Capitalization as Percentage of GDP and Number of Companies Listed in the Mexican Stock Exchange 390 8.1 Evolution of the Privatization Process, 1991­2000 414 8.2 Privatization Revenues by Sector, 1991­99 415 xii CONTENTS 8.3 Privatization Process Progress, 1991­2000 416 8.4 Public Approval of Privatization, 1991­2000 417 8.5 Evolution of Performance Indicators 428 TABLES 1.1 Proceeds from Privatization in Developing Countries, 1990­99 5 1.2 Recent Studies on Firm Performance after Privatization in Latin America 9 1.3 Reasons for Firm Exclusion from the Privatization Sample 18 1.4 Labor Restructuring before Privatization, by Region 44 1.5 Labor Restructuring and Privatization Prices in Latin America 45 1.6 New Hires and Rehires in Privatized Firms in Latin America 50 2.1 Privatization Revenues in Argentina, by Sector 72 2.2 Nonfinancial Companies Included in the Database 75 2.3 Nonfinancial Privatizations in Argentina Not Included in the Database 77 2.4 Privatized Banks in Argentina Included in the Database 79 2.5 Changes in Profitability for the Sample of Nonfinancial Privatized Firms 82 2.6 Changes in Performance for the Sample of Nonfinancial Privatized Firms 83 2.7 Changes in Profitability for the Sample of Privatized Banks 88 2.8 Changes in Performance for Privatized Public Banks 89 2.9 Access to Water and Sewerage Services, 1991 92 2.10 Difference in Difference of the Proportion of Households with Access to Water Connection, 1991­97 94 2.11 The Effect of Privatization on Child Mortality Rates, 1990­99 96 2.12 The Effect of Privatization on Displaced Workers' Earnings Flows 101 2B.1 Description of the Variables Used to Evaluate the Impact of Privatization on the Performance of Nonfinancial Firms 109 2B.2 Description of the Variables Used to Evaluate the Impact of Privatization on the Performance of Financial Firms 111 CONTENTS xiii 3.1 Common Objectives for Privatizations 119 3.2 Reasons for Excluding Firms from the Sample 129 3.3 Changes in Performance of the Sample of Privatized Firms on Bolivia 131 3.4 Industry-Adjusted Changes in Performance for the Sample of Privatized Firms 135 3A.1 List of Privatized Firms, 1992­2001 137 3A.2 Description of the Variables Used in Tables 3.3 and 3.4 139 4.1 Description of the Privatization Program and Coverage of the Sample, 1991­2000 150 4.2 Privatized Brazilian Companies by Industry Classification 152 4.3 Summary of Results 153 4.4 Changes in Performance: GMM-IV Panel Data Analysis 159 4.5 Employment in Selected Industries, by Public or Private Ownership, 1995­99 164 4A.1 Federal State Enterprises Privatized, 1991­2000 175 4A.2 Companies Privatized by BNDES on Behalf of Brazilian States, Minority Shares Privatized by Federal Government, and São Paulo State Privatization Program 180 4A.3 Remaining State-Owned Enterprises 182 4B.1 Description of the Variables 184 4C.1 Change in Performance: Tests of Means and Medians (Two Years before Privatization versus Two Years after, without Adjustment) 186 4C.2 Change in Performance: Tests of Means and Medians (Two Years before Privatization versus Two Years after, with Adjustment) 188 4C.3 Change in Performance: Tests of Means and Medians (All Years before and after Privatization, without Adjustment) 189 4C.4 Change in Performance: Tests of Means and Medians (All Years before and after Privatization, with Adjustment) 191 4D.1 Definition of the Control Variables Included in the Vector of the Econometric Model 192 5.1 Nationalization and Privatization of Firms in Chile, 1970­2001 199 5.2 State-Owned and State-Seized Firms, 1970­2001 204 5.3 Number of State-Owned Firms, by Year, 1973­2001 205 xiv CONTENTS 5.4 Privatized State-Owned Enterprises, 1972­2001 207 5.5 Revenues from Privatization of Chilean Public Enterprises, 1985­89 210 5.6 Net Income to Equity for Privatized Firms before Privatization, 1970­86 216 5.7 Employment Changes in Privatized Firms, 1970­92 220 5.8 Privatization of Chilean Telecommunications Firms, 1984­89 221 5.9 Privatization of Electric Power Firms, 1984­89 222 5.10 ENDESA: Investment, Power Generation, and Labor Productivity 228 5.11 Chilectra: Sales, Employees, Labor Productivity, and Energy Loss 229 5.12 Change in Node Prices and Residential Rates 229 5.13 Profits of the Main Electric Sector Companies: 1987­2000 231 5.14 Telecommunications Statistics, 1987­2000 232 5.15 Telefónica-CTC: Basic Fixed Phone Statistics 233 5.16 Cost of Local Monthly Telephone Service for the Average Family 234 5.17 Profits of Telecommunications Enterprises, 1987­2000 235 5.18 Concessions in Operation 239 5.19 Valparaíso: Time Spent Loading and Unloading and Transfer Speed 246 5A.1 Changes in Profitability of Privatized Firms 249 5A.2 Changes in Operating Efficiency of Privatized Firms 252 5A.3 Changes in Investment and Assets in Privatized Firms 255 5A.4 Changes in Profitability of Privatized Firms, Adjusted 258 5A.5 Changes in Operating Efficiency of Privatized Firms, Adjusted 261 5A.6 Changes in Investment and Assets in Privatized Firms, Adjusted 263 5A.7 Employment in Privatized Firms 265 5A.8 Physical Productivity before and after Privatization 266 5A.9 Concession Projects under Construction 268 5A.10 Projects in Concession Process 269 6.1 Privatization Program in the Real Sector in Colombia, 1986­98 279 6.2 IFI Privatization Program, 1986­97 281 6.3 ECOPETROL Privatization Program, 1993­99 285 6.4 Privatization in the Power Sector, 1995­98 289 CONTENTS xv 6.5 Average Changes in Manufacturing Basic Variables after Privatization for IFI Sample and Total Manufacturing 292 6.6 Changes in Performance for the Sample of Privatized IFI Firms 294 6.7 Industry-Adjusted Changes in Performance for the Sample of Privatized IFI Firms 296 6.8 IFI Firms: Role of Transfers from Workers 305 6.9 Changes in Performance in the Sample of Privatized Power Utilities and Public Enterprises of Medellín 307 6.10 Industry-Adjusted Changes in the Performance of Privatized Power Utilities 309 6.11 Annual Averages for Wholesale Electricty Market Efficiency Variables, 1996­2000 314 6.12 Markup Determinants for IFI Firms 317 6.13 Total Factor Productivity Determinants for IFI Firms 321 6.14 Determinants of Thermal Plants' Efficiency Scores 325 6A.1 Infrastructure Concession Projects with Ongoing Private Investment by 1998 329 6B.1 List of IFI Enterprises in the Sample 330 6C.1 The Indicators for IFI Firms in the Sample 330 6D.1 Colombia: Power Sector Statistics and Description of the Data Sets 337 6D.2 Thermal Plants: Input and Output Variables 339 6E.1 DEA Efficiency Scores in Thermal Generation before and after the Regulatory Reform 340 7.1 State-Owned Enterprises in Mexico, 1917­2003 351 7.2 State-Owned Enterprises, 1982­2003 353 7.3 The Privatization Program in Perspective 354 7.4 Changes in Performance for the Sample of Privatized Firms 356 7.5 Industry-Adjusted Changes in Performance for the Sample of Privatized Firms 360 7.6 Median Performance Changes in Privatized Firms in Competitive versus Noncompetitive Industries 361 7.7 The Role of Transfers from Workers 365 7.8 Other Benefits of Privatization Programs 366 7.9 The Fiscal Impact of Privatization: 1983­2003 368 7.10 Foreign Direct Investment in Privatization 370 7.11 Restructuring Actions before Privatization 373 7.12 Prior Restructuring: Dos and Don'ts 375 xvi CONTENTS 7A.1 Definition of Variables 394 8.1 Privatization Revenues and Investment, 1991­2001 412 8.2 Firm Performance Measures 419 8.3 Nonfinancial Companies Included in the Study 425 8.4 Nonfinancial Companies Not Included in the Study 427 8.5 Privatized Banks Included in the Study 430 8.6 Changes in Performance for the Privatized Firms 433 8.7 Performance Indicators of Privatized Utilities, Difference in Difference 435 8.8 Changes in Performance in the Financial Sector after Privatization 439 8.9 Performance Indicators of Privatized Banks, Difference in Difference 442 8.10 Changes in Employment after Privatization 444 8.11 Impact of Layoffs on Performance Indicators for Major Privatized Firms 446 8B.1 Description of Variables 451 8C.1 Evolution of Privatization in the Electricity Sector 458 8C.2 Evolution of Privatization in the Financial Sector 460 8D.1 Basic Statistics of Privatized Firms 463 8E.1 Changes in Performance after Privatization for Telefónica del Perú 468 8E.2 Changes in Performance after Privatization for Electrolima 470 8E.3 Changes in Performance after Privatization for Electroperú 472 Foreword In the 1980s, a number of Latin American countries launched significant privatization programs. Following decades of statist economic policies, trade protection, heavy-handed regulation, and even nationalization, priva- tizations were introduced as the linchpin of Washington consensus policies. Indeed, not only countries in Latin America, but many transition economies and other developing economies expected privatization to ignite economic growth. A decade later, many privatizations in Latin America were completed, but the process reached a standstill. The initial hope and optimism gave way to doubt, disappointment, and a widely shared belief that privatiza- tion had failed. Indeed, the alleged failures of privatization became central to the denunciations of Washington consensus policies. So what happened? Has privatization delivered benefits or not? Were the critics scholars or demagogues? This volume discusses a number of criticisms of privatization and then painstakingly assembles empirical data designed to evaluate them. A broad range of evidence, collected from a variety of countries, points to increased productivity and profitability, accelerating restructuring and output growth, mounting tax revenues, and improving product quality following privatization. In the cases where privatizations failed, the prob- lems appear to be linked to continued state involvement and regulation, as well as a weak corporate governance framework. Indeed, the volume pro- vides substantial evidence that improved corporate governance and regula- tory environment are complementary to privatization. In general, private ownership delivers the same significant benefits in Latin America as it does in other parts of the world. But do the increased shareholder profits of privatized firms come at the expense of other stakeholders? The evidence provides no support for the view that increased profitability comes from monopoly pricing, exploitation of workers, or reductions in tax payments. To the con- trary, increased profitability comes from productivity growth rather than redistribution. The studies in this volume also show that the manner in which privati- zation is carried out matters. Transparency and homogeneity of procedures, xvii xviii FOREWORD speed, and moderation in preprivatization restructuring lead to better outcomes and allow less room for corruption. The evidence presented in this volume amounts to a compelling case that privatization in Latin America has been a success. To the extent that they pay attention to the evidence, critics of privatization in Latin Amer- ica must recognize the basic fact that benefits have been substantial. It is hoped that these studies advance the course of privatization and capitalism in other parts of the world as well. Andrei Shleifer Whipple V.N. Jones Professor of Economics, Harvard University January 2005 Acknowledgments THIS BOOK WAS WRITTEN WITH THE SUPPORT of the Latin American Research Network at the Inter-American Development Bank (IDB). Created in 1991, this network aims to leverage the IDB's research department's capabilities, improve the quality of research performed in the region, and contribute to the development policy agenda in Latin America and the Caribbean. Through a competitive bidding process, it provides grant funding to lead- ing Latin American research centers to conduct studies on the economic and social issues of greatest concern to the region today. The network cur- rently comprises nearly 300 research institutes all over the region and has proven to be an effective vehicle for financing quality research to enrich the public policy debate in Latin America and the Caribbean. Many individuals provided comments and suggestions: César Calderón, Guillermo Calvo, Virgilio Galdo, Arturo Galindo, Magdalena López-Morton, Eduardo Lora, William Megginson, Alejandro Micco, Ugo Panizza, Andrei Shleifer, and Luisa Zanforlin. The authors also thank Bank and Yale University colleagues who participated in formal and in- formal discussions and workshops on background papers, and who pro- vided comments during revisions. Valuable input was also provided in the production of this book by Norelis Betancourt, Madison Boeker, Adriana Cabrera, Rita Funaro, Raquel Gómez, Martha Grotton, Maria Helena Melasecca, Mariela Semidey, and John Dunn Smith. Book design, editing, and print production were coordinated by Santiago Pombo, Janet Sasser, and Monika Lynde in the World Bank's Office of the Publisher. The views and opinions expressed in this book are those of the authors and do not necessarily reflect the official position of the IDB, its Board of Directors, or the Advisory Committee. xix 1 The Truth about Privatization in Latin America Alberto Chong and Florencio López-de-Silanes AFTER DECADES OF POOR PERFORMANCE and inefficient operations by state- owned enterprises, governments all over the world have earnestly embraced privatization. Beginning in the 1980s, thousands of state-owned enterprises have been turned over to the private sector in Africa, Asia, Latin America, and eastern and western Europe. This trend was spurred by the well- documented poor performance and failures of state-owned enterprises and the efficiency improvements after privatization around the world.1 Privatiza- tion efforts have greatly stalled in recent years, however, despite worldwide evidence that points to improved performance, firm restructuring, fiscal ben- efits, increased output, and quality improvements following privatization. Academia, politicians, and the media have recently attacked privatiza- tion, voicing concerns about its record, the sources of the gains, and its impact on social welfare and the poor.2 The negative reaction to privati- zation is reflected in opinion polls and some governments' reluctance to further their privatization programs.3 Popular support for privatization, as for other structural policies, generally follows a J curve, declining at first and recovering when the policy matures (Przeworski 1991). If politi- cians retreat from the now-unpopular effort to restructure the role of the state in the economy, the window of opportunity for deepening privatiza- tion efforts may close.4 Many countries have implemented large privatiza- tion programs, but in many others the state retains a large presence, often across many sectors of the economy (La Porta, López-de-Silanes, and Shleifer 2002). In these circumstances, it is imperative to analyze the real record of privatization and draw lessons from it. This chapter evaluates the privatization experience and assesses the em- pirical validity of the main concerns voiced against it. We focus on Latin America because after the transition economies of eastern Europe, Latin 1 2 CHONG AND LÓPEZ-DE-SILANES America is the region with the largest decline in the state's share of production in the past 20 years. The extent of privatization in Latin America and the quality of the data allow researchers to produce compre- hensive analyses that provide appropriate academic responses to some of the main criticisms raised. Overall, the empirical record shows that privatization leads not only to higher profitability but also to large output and productivity growth, fiscal benefits, and even quality improvements and better access for the poor. Instances of failure exist, but in light of the overwhelming evidence, these failures should not be turned into an argument to stop privatization. The analysis in this chapter suggests that privatization failures can be un- derstood in a political economy framework. Their roots can be traced to substantial state participation in opaque processes, poor contract design, inadequate reregulation, and insufficient deregulation and corporate gov- ernance reform that increase the cost of capital and limit firm restructur- ing in a competitive environment. The chapter is organized as follows. The next section gives a brief overview of the rationale and extent of privatization around the world. The rest of the sections are structured around what we consider the four main ar- eas of concern about privatization. The first hurdle is to confirm that the profitability increases recorded by the literature are robust, unbiased, and not solely explained by sample selection of the best firms. The first generation of privatization papers suffered from this problem. A recent series of Latin American studies analyzed here, however, uses comprehensive firm-level data that provide robust evidence on performance changes after privatiza- tion. The second hurdle is to address criticisms of privatization concerning the welfare of workers, consumers, and the state, which we do by exploring who pays for the profitability gains. The evidence suggests that although la- bor cost reductions and price increases account for part of the gains, the bulk of the profitability improvement lies in deep firm restructuring and produc- tivity growth. The third hurdle is to examine concerns about the proper role of the state in firm restructuring before privatization and about the opacity of procedures, which may lead to collusion and corruption. Our final hurdle is to assess the role of complementary policies such as deregulation, reregu- lation, and corporate governance reform. We place particular attention on sectors with market power and inefficient regulation following privatization. The final section concludes, providing some policy implications from the pri- vatization record thus far. A Brief Look at the Privatization Experience around the World In the mid-1900s many famous economists and politicians favored state ownership of firms in several industries, where monopoly power and externalities often produced market failures. In the 1990s, however, the THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 3 evidence on the failures of state-owned enterprises around the world and developments in contract and ownership theory led to a reassessment of the benefits of state ownership in production (Shleifer 1998). The litera- ture emphasizes two reasons for the poor record of state ownership. One strand of the literature focuses on managerial shortcomings; it reflects the idea that imperfect monitoring and poor incentives for managers of state- owned enterprises translate into inferior performance. There are many reasons to believe this would be so. The average state-owned firm is not traded on the stock market; the threat of a takeover does not exist since control rests in the hands of the state. Discipline from creditors does not play much of a role either, because most loans to state-owned enterprises are public debt, and losses are typically covered by subsidies from the treasury. Additionally, the boards of directors rarely implement good cor- porate governance practices, and management turnover obeys political rather than market forces (Vickers and Yarrow 1988). The other strand of the literature emphasizes the political economy aspects of state production. The political view highlights the inherent conflict of interest in running state-owned enterprises, as managers seek to maximize their political capital and thus pursue inefficient decisions. Political interference in the firm's production results in excessive employment, poor choices of products and location, and inefficient in- vestment (Shleifer and Vishny 1996; La Porta and López-de-Silanes 1999). State-owned enterprises face soft budget constraints that allow them to implement such practices, since governments may not want to risk the political cost of firms going bust (Sheshinski and López-Calva 2003). The basic claims of the two strands of the literature have been validated by empirical research on state-owned enterprises and firm per- formance after privatization around the world (see Boardman and Vining 1989; Megginson, Nash, and van Randenborgh 1994; Ehrlich and others 1994; La Porta and López-de-Silanes 1999; Frydman and others 1999; Dewenter and Malatesta 2001; and Chong and López-de- Silanes 2004, among others). Motivated by the evidence on the failures of state-owned enterprises, governments in more than 100 countries have undertaken privatization programs since the mid-1980s (Megginson and Netter 2001). Through- out the world annual revenues from privatization soared during the late 1990s, peaking in 1998 at over $100 billion (OECD 2001).5 Industrial countries have pursued privatization less vigorously than have developing nations. Between 1984 and 1996 the participation of state-owned enter- prises in industrial countries declined from a peak of 8.5 percent to about 5.0 percent of gross domestic product (GDP), while production from state-owned companies declined more steeply in developing countries (figure 1.1). According to Sheshinski and López-Calva (2003), the activ- ities of state-owned enterprises as a percentage of GDP decreased from about 11 percent in 1980 to 5 percent in 1997 in middle-income coun- tries and from 15 to 3 percent in low-income economies. Developing 4 CHONG AND LÓPEZ-DE-SILANES Figure 1.1 Economic Activity of State-Owned Enterprises, 1978­97 Percent of GDP 25 20 Africa 15 10 Asia Latin America 5 Industrial countries 0 19781979198019811982198319841985198619871988198919901991199219931994199519961997 Note: Weighted average by country GDP. Source: World Bank 2001. countries also saw large reductions in employment among state-owned enterprises during the same period. In middle-income countries such em- ployment fell from a peak of 13 percent of total employment to about 2 percent, and in low-income countries it dropped from more than 20 per- cent to about 9 percent. These averages mask great regional variation in the size and economic importance of the remaining state-owned production. In Sub-Saharan Africa only a few governments have openly adopted an explicit divestment strategy for state-owned enterprises. The African privatization effort has been significant in only a handful of countries, and state production still accounts for more than 15 percent of GDP in the region.6 Asia also fea- tures large variation, in that several Asian countries have not consistently pursued a privatization strategy. China, for example, only recently com- mitted to privatizing all but the largest state enterprises. In India, where privatization has thus far not figured prominently in the agenda, the state still owns 43 percent of the country's capital stock. Many governments in THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 5 Table 1.1 Proceeds from Privatization in Developing Countries, 1990­99 (US$ billions) Eastern Middle East Europe East Asia and and Sub- and Latin Central North South Saharan Year Pacific America Asia Africa Asia Africa 1990 376 10,915 1,262 2 29 74 1991 834 18,723 2,551 17 996 1,121 1992 5,161 15,560 3,626 69 1,557 207 1993 7,155 10,488 3,988 417 974 641 1994 5,508 8,199 3,957 782 2,666 605 1995 5,410 4,616 9,742 746 916 473 1996 2,680 14,142 5,466 1,478 889 745 1997 10,385 33,897 16,537 1,612 1,794 2,348 1998 1,091 37,685 8,002 1,000 174 1,356 1999 5,500 23,614 10,335 2,074 1,859 694 1990­1999 44,100 177,839 65,466 8,197 11,854 8,264 Source: World Bank 2001. the region continue to hang on to their assets in sectors such as energy, telecommunications, transportation, and banking, although private equity funds and multinationals were expecting large state-owned fire sales after the Asian crisis of 1997.7 In contrast, transition economies and Latin American countries have been very active in privatization. Transition economies in eastern Europe and central Asia accounted for 21 percent of total privatization revenues in developing countries during the 1990s, second only to Latin America (table 1.1). To facilitate their shift to a market economy, most transition countries launched mass privatization programs that resulted in dramatic reductions of state ownership. These programs, however, have sometimes been unpopular, accused of corruption and foot-dragging on implement- ing corporate governance reforms, and thus affording poor protection for new minority investors. Even against the backdrop of massive economic transformations in transition economies, the privatization record of Latin America seems remarkable. Latin America accounted for 55 percent of total privatiza- tion revenues in the developing world in the 1990s. The decline in the economic activity of state-owned enterprises has been more substantial in Latin America than in Asia and Africa, bringing levels close to those of industrialized countries. In recent years, however, Latin America has virtually halted its privatization process. 6 CHONG AND LÓPEZ-DE-SILANES The privatization impetus has also faded in other regions, leaving the bureaucrats very much in business. State-owned enterprises still account for more than 20 percent of investment worldwide and about 5 percent of formal employment (Kikeri 1999). Governments may own or control much more than is apparent at first sight. A clear example is the case of government ownership of banks. Data for the late 1990s indicate that af- ter bank privatization programs had been completed in many countries, the world mean of government ownership of the top 10 banks was still 42 percent, 39 percent if former or current socialist countries were excluded (La Porta, López-de-Silanes, and Shleifer 2002). Thus, while government ownership has decreased with privatization, it has not fallen to negligible levels. Dramatic differences in the extent of privatization are also evident within regions. In Latin America, for example, countries with large state- owned sectors, such as Ecuador, Nicaragua, and Uruguay, barely priva- tized at all in the 1990s, while others, such as Argentina, Bolivia, Guyana, Panama, and Peru, raised revenues from comprehensive privatization programs that exceeded 10 percent of GDP (figure 1.2). The difference in the extent of privatization across countries and the large amount of assets in the hands of the state heighten the importance of understanding the privatization record so far and of developing lessons for future privatiza- tion programs.8 Which Firms Are Up for Sale? Concerns about What Is Privatized Privatization studies typically analyze the impact on firm performance by comparing firm-level data before and after privatization. This litera- ture has established worldwide evidence on the benefits of privatization from increased firm profitability driven primarily by increased effi- ciency (Megginson, Nash, and van Randenborgh 1994; Boubakri and Cosset 1998, 1999; Dewenter and Malatesta 2001). Critics suggest, however, that these results may reflect sample selection bias or result from the use of noncomparable data. Sample Selection Bias Sample selection bias can arise from five basic sources. First, politicians who conduct privatization have the incentive to sell only the healthiest firms--what critics refer to as the crown jewels. According to this hypothesis, politicians sell only viable assets and keep poor performers, allowing investors to engage in cherry-picking (Bayliss 2002). Second, several studies are based on information about firms privatized through public offerings on the stock exchange. Such samples are thus biased THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 7 Figure 1.2 Revenues from Privatization in Latin America, 1990­2000 Country Paraguay Peru Bolivia Panama Guyana Argentina El Salvador Brazil Guatemala Belize Venezuela Dominican Republic Mexico Jamaica Trinidad and Tobago Chile Nicaragua Honduras Colombia Barbados Ecuador Bahamas Costa Rica Uruguay 0 2 4 6 8 10 12 14 16 18 Percent of 1999 GDP Source: Lora 2001. 8 CHONG AND LÓPEZ-DE-SILANES toward the largest, and probably the best-performing, firms. A third source of sample selection comes from the greater availability of data from industrialized countries, which may have relatively better-performing firms. Cross-country firm-level analyses are therefore biased because their samples include a disproportionate share of well-performing firms.9 The fourth source emerges from the intense focus of the studies on oligopolistic or heavily regulated industries, where the gains from privatization may come from market power. Finally, survivorship bias is introduced when firms that went bankrupt after privatization are excluded from the sample that compares performance before and after privatization. Several early studies on firm performance after privatizations in Latin America suffer from these biases (table 1.2). Some of these papers are spe- cific case studies of a limited number of large firms (see, for example, Galal and others 1994 and Chong and Sánchez 2003). Others do not include econometric or statistical analysis (Sánchez and Corona 1993; Hachette and Lüders 1994; Birch and Haar 2000). Still others are econometric stud- ies of one or two heavily regulated sectors (Ramamurti and Vernon 1991; López-de-Silanes and Zamarripa 1995; Pinheiro 1996; Ramamurti 1996, 1997). Finally, some provide evidence from cross-country analysis of oligopolistic sectors such as telecommunications (Petrazzini and Clark 1996; Ramamurti 1996; Ros 1999; Wallsten 2000). Overcoming sample selection bias is empirically difficult and requires large amounts of pre- and postprivatization information for nearly com- plete cross-industry samples of privatized firms of all sizes. La Porta and López-de-Silanes (1999) deal with these issues by collecting information from 95 percent of nonfinancial firms privatized in Mexico in the period 1983­92.10 Mexico undertook a comprehensive privatization program in which the goal was to eliminate state ownership across the board, with the exceptions of electricity and oil. As a result, the sample contains large, medium-size, and small firms that span more than 40 sectors covering mining, manufacturing, agricultural products, and services as varied as night clubs and soccer teams. These characteristics make it a good sample for testing the validity of the concerns raised above. The study concludes that sample selection bias does not explain the positive results reached by privatization, as profitability of privatized firms increases across sectors and firm sizes, even considering bankrupt firms. The median firm experi- enced an increase in operating profitability of 24 percentage points. More- over, the Mexican government did not sell the crown jewels, given that this oil-rich nation retained petroleum and some petrochemicals as state assets.11 A recent research effort across Latin America expands the detailed pri- vatization analysis for the region, using comprehensive data and a methodology similar to that described above for Mexico to examine the programs of Argentina, Bolivia, Brazil, Chile, Colombia, and Peru and to update the data and findings for Mexico These studies, gathered together Table 1.2 Recent Studies on Firm Performance after Privatization in Latin America Study Sample, period, and methodology Summary of findings and conclusions Birch and Haar Uses a descriptive study of the privatization Finds sizable effects of privatization on short- and 2000 experience in the last two decades in Argentina, long-run macroeconomic conditions; shows a Brazil, Chile, Colombia, Mexico, Peru, Venezuela, positive effect of privatization on productivity and and some Caribbean countries. a negative effect on employment. Chisari, Estache, Assesses macroeconomic and distributional effects Concludes that privatization of utilities accounts for and Romero of privatization in Argentina's gas, electricity, total gains of about $3.3 billion (at 1993 prices) 1999 telecommunications, and water sectors using a or the equivalent of 1.25 percent of GDP. computable general equilibrium model. Privatization cannot be blamed for increased unemployment, which may be caused by ineffective regulation. Chong and Uses a detailed analysis of the contractual Concludes that clear, homogeneous, transparent, Sánchez 2003 arrangements of privatizations and concessions in and credible institutional processes during infrastructure in Brazil, Chile, Colombia, and Peru. privatization yield positive outcomes. Clarke and Cull Tests econometrically how political constraints Finds that provinces with high fiscal deficits were 1999 affect transactions during bank privatization, willing to accept layoffs and guarantee a larger based on evidence from the privatization program part of the privatized bank's portfolio in return of provincial banks in Argentina during the for a higher sale price. 1990s. Galal and others Compares postprivatization performance of 12 Finds net welfare gains in 11 of 12 cases covered, 1994 large firms (mostly airlines and regulated utilities) with average gains equal to 26 percent of the from Chile and Mexico. firms' predivestiture sales; uncovers no case in which workers were made worse off and three cases in which workers' conditions improved. 9 (Table continues on the following page.) 10 Table 1.2 (continued) Study Sample, period, and methodology Summary of findings and conclusions Hachette and Analyzes the difference in 10 performance Finds no significant differences in behavior among Lüders 1994 indicators of 144 private, public, and privatized public, private, and privatized firms that operate firms in Chile in 1974­87. under similar sets of rules and regulations. Petrazzini and Uses International Telecommunications Union data Deregulation and privatization are both associated Clark 1996 through 1994 to test whether deregulation and with significant improvements in the level and privatization affect the level and growth of growth of telephone density but have no telephone density, prices, service quality, and consistent impact on the quality of service. employment; sample covers 26 developing Deregulation is associated with lower prices and countries, including some Latin American increased employment; privatization has the nations. opposite effect. Pinheiro 1996 Analyzes the performance of 50 Brazilian firms Concludes that privatization has improved the before and after privatization, using data through performance of the firms; shows that the null 1994; variables used are net sales, net profits, net hypothesis of no change in behavior is rejected for assets, investment, employment, and indebtedness. the production, efficiency, profitability, and investment variables; and finds a significant negative impact on employment. Ramamurti 1996 Surveys four telecommunications, two airlines, and Concludes that privatization had positive results for one toll-road privatization program in 1987­91; telecommunications, partly owing to the scope for discusses political and economic issues and improvement of technology, capital investment, methods used to overcome bureaucratic and and attractiveness of offer terms; observes little ideological opposition to divestiture. improvement in airlines and toll road, which had less room for productivity enhancement. Ramamurti 1997 Examines the restructuring and privatization of Documents a 370 percent improvement in labor Ferrocarriles Argentinos in 1990, testing whether productivity and a 78.7 percent decline in productivity, employment, and the need for employment; an improvement and expansion in operating subsidies changed after divestiture. services, combined with a reduction in the cost to consumers; and the elimination of the need for operating subsidies. Ros 1999 Uses International Telecommunications Union data Countries with at least 50 percent private and panel data regressions to examine the effects ownership in the main telecommunications firm of privatization and competition on network have significantly higher telephone density levels expansion and efficiency in 110 countries in and growth rates. Both privatization and 1986­95. competition increase efficiency, but only privatization is positively associated with network expansion. Sánchez and Uses a descriptive case-study approach to analyze Finds great differences in the effects of privatization Corona 1993 the privatization experiences of Argentina, Chile, in the countries covered; concludes that firms, Colombia, and Mexico, focusing on the institutions, and regulations need sufficient time preparatory measures taken prior to privatization; to prepare for the privatization process to be valuation, sale mechanisms, regulation, and successful. supervision; and the fiscal and macroeconomic impact of privatization. Trujillo and Uses pooled and panel data with fixed and random Finds that private sector involvement in utilities and others 2002 effects to examine the macroeconomic effects of transport has minimal positive effects on GDP. private sector participation in infrastructure, Private investment is crowded out, and private based on a sample of 21 Latin American countries participation reduces recurrent expenditures-- in 1985­98. except in transport, where it has the opposite effect. The net effect on the public sector account is uncertain. (Table continues on the following page.) 11 12 Table 1.2 (continued) Study Sample, period, and methodology Summary of findings and conclusions Wallsten 2001 Explores the impact of privatization, competition, Indicates that competition is significantly associated and regulation on telecommunications firms' with increases in per capita access to telecom- performance in 30 African and Latin American munications services and with decreases in its countries in 1984­97. costs, while privatization is helpful only if coupled with effective, independent regulation. Concludes that competition combined with privatization is best and that privatizing a monopoly without regulatory reforms should be avoided. Comprehensive sample country studies in Latin America Study Country sample, period, and methodology Summary of findings and conclusions Galiani and Argentina. Covers 21 federal nonfinancial state- Profitability of nonfinancial firms increased 188 others 2005 owned firms plus all privatized banks in percent after privatization. Investment increased Argentina, which account for 74 percent of total at least 350 percent while employment decreased privatization revenues; tests whether performance approximately 40 percent; there was no impact on indicators of state-owned firms improved after prices. privatization. Period: 1991­2000. Capra and Bolivia. Covers 32 firms, which account for 60 Privatization had a significant impact in operating others 2005 percent of total transactions in Bolivia; tests efficiency as profitability increased by over 100 whether performance indicators of state-owned percent and costs per unit dropped by a third. firms improved after privatization. Period: Employment fell by 15 percent, but wages for 1992­99. remaining blue- and white-collar workers doubled. Anuatti-Neto Brazil. Includes 102 publicly traded firms Privatization improved the firms' profitability (14 and others (equivalent to 94 percent of total value of percent) and reduced their unit costs (33 percent) 2005 transactions in the country); tests whether and investment-to-sales ratio (41 percent). performance indicators improved after privatization. Period: 1987­2000. Fischer, Chile. Covers only 37 nonfinancial firms, owing to Profitability did not increase significantly after Gutiérrez and political and economic turbulence in the 1970s privatization, and productivity did not vary among Serra, 2005 and changes in accounting standards; tests regulated and unregulated sectors. Study finds no whether performance indicators improved after evidence that firms fired workers after privatization, privatization. Period: 1979­2001. although layoffs occurred prior to privatization. Pombo and Colombia. Analyzes 30 former firms in the Institute Firms were profitable before privatization. Labor Ramírez 2005 for Industrial Promotion program, which account productivity grew 13 percent and investment fell for 95 percent of the total accumulated from 5.9 to 2.5 percent per year owing to privatization sales; tests whether performance previous overinvestment; employment was indicators improved after privatization. Period: reduced by 23 percent. 1974­98. La Porta and Mexico. Assesses whether the performance of 218 The output of privatized firms increased 54.3 López-de- privatized firms improved after divestment; percent, while employment declined by half Silanes 1999; compares performance with industry-matched (though wages for remaining workers increased). Chong and firms; splits improvements documented between Firms achieved a 24 percentage point increase in López-de- industry- and firm-specific results. Period: operating profitability, eliminating the need for Silanes 2005 1983­1991. subsidies that amounted to 12.7 percent of GDP. Higher product prices explain 5 percent of improvements; transfers from laid-off workers, 31 percent; and incentive-related productivity gains, 64 percent. 13 (Table continues on the following page.) 14 Table 1.2 (continued) Study Country sample, period, and methodology Summary of findings and conclusions Torero 2005 Peru. This study covers 36 nonfinancial firms, Profitability, operational efficiency, and output which account for 90 percent of privatization increased after privatization. The ratio of sales to cases and 86 percent of total transactions. In employees increased by 50 percent in addition, it includes a separate analysis for the telecommunications, 69 percent in electricity, and financial sector. It tests whether performance 25 percent in the financial sector. After indicators improved after privatization. Period: privatization, 36 percent of employees retained 1986­2000. their jobs. Source: Megginson and Netter 2001; Chong and López-de-Silanes 2004. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 15 in this volume, compare firm performance before and after privatization, and they adjust for macroeconomic and industry effects with matching firms. Figure 1.3 summarizes the data collection efforts of this series of pa- pers. With the exception of Brazil, where access to preprivatization data for firms that are not publicly traded was denied, the coverage across firm sizes for all countries is enough to put to rest the main concerns regarding sample selection. The samples used for Bolivia and Chile are the smallest, around 66 percent in terms of value, while the samples for the rest of the countries cover 80, 90, and even 95 percent of transaction values and number of privatization contracts. Extensive groundwork and creative ways of accessing nonpublic infor- mation allowed researchers to collect comprehensive pre- and postprivati- zation data. In Peru, for example, Torero obtained preprivatization information from so-called White Books, or original privatization docu- ments that were available to prospective bidders when state-owned enter- prises were being privatized. He was able to collect comprehensive post- privatization data from privatization dossiers, as well as from the National Supervisory Commission of Firms and Securities and other regulatory agencies. All in all, Torero collected information for nearly 90 percent of privatized firms in Peru. For Argentina, Galiani, Gertler, Schargrodsky, and Sturzennegger drew a comprehensive sample based on information from individual companies, the Ministry of Economic Affairs, and regula- tory agencies. In Colombia, which has smaller privatization programs than those of Argentina and Peru, Pombo and Ramírez collected compre- hensive information on the privatization from the Institute for Industrial Promotion.12 They constructed an unbalanced panel data set with records from the Annual Manufacturing Survey starting in 1974 and ending in 1998. Their panel features over 140 variables covering 94 specific groups based on the International Standard Industrial Classification, together with survey information on about 6,000 establishments. For Mexico, Chong and López-de-Silanes use the same database as did La Porta and López-de-Silanes (1999), which combines information from the original privatization White Books with information collected from surveys sent to privatized firms and data from the various census bureaus. The informa- tion for Mexico basically covers the whole program, with 218 nonfinan- cial, state-owned enterprises privatized between 1983 and 1992. In Bolivia, information on privatized state-owned enterprises is par- ticularly difficult to gather owing to the relatively small size of firms and the lackadaisical record-keeping efforts in the country.13 Capra, Chong, Garrón, López-de-Silanes, and Machicado complement original infor- mation from government institutions with information collected through a survey sent to privatized firms. For Chile, Fischer, Serra, and Gutiérrez faced significant complications in collecting data owing to the long privatization period (1979­2001) and the change in accounting standards in 1982. Despite these problems, their data provide systematic 16 CHONG AND LÓPEZ-DE-SILANES Figure 1.3 Availability of Privatization Data on Latin America a. Value of transactions Argentina: Bolivia: Brazil: Colombia: 1990­2000 1992­2000 1991­2000 1974­98 Chile: Mexico: Peru: 1983­2000 1983­92 1986­2000 b. Privatization contracts Argentina: Bolivia: Brazil: Colombia: 1990­2000 1992­2000 1991­2000 1974­98 Chile: Mexico: Peru: 1983­2000 1983­92 1986­2000 Sample Note: On the pie charts, the gray area indicates sample coverage; the value of transactions is given as a percentage of the total value of privatization transactions in each country; the number of privatization contracts is given as a percentage of the total number of privatization contracts in the country. Source: Chapters 2­8. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 17 evidence that complements more descriptive work by others such as Lüders (1991) and Sáez (1992). Finally, Brazil proved to be the most dif- ficult case, since Anuatti-Neto, Barossi-Filho, Gledson de Carvalho, and Macedo were denied access to all preprivatization information for firms that are not publicly traded and were thus restricted to using informa- tion on firms traded on the stock exchange. Although their results may suffer from some sample selection bias, their study represents one of the most comprehensive data sets in Brazil, covering close to 95 percent of the total value of privatization transactions (chapter 4). Overall, the coverage and industry-matching techniques of this recent se- ries of privatization studies in Latin America demonstrate that the increased profitability of privatized firms is not the result of sample selection bias. Noncomparable Data There are two additional problems with data collection procedures relat- ing to the comparability of firms before and after the sale. In several coun- tries governments either split existing state-owned enterprises to sell them as independent units or grouped separate firms together to form packages to be sold as a unit. In both cases large amounts of data are needed to con- duct a firm-by-firm analysis of the pre- and postprivatization period. Hav- ing information disaggregated at the plant level and gaining access to financial statements prepared before the sale are essential for keeping units comparable across time. A second set of problems with the data emerges from changes in the sample after privatization, since the state-owned firm may be merged with the acquiring firm or with one of its subsidiaries. Such a merger creates a new entity and thus makes it difficult, if not impossible, to make meaningful comparisons. Table 1.3 summarizes the different problems faced by the researchers who recently undertook the comprehensive privatization analyses in seven Latin American countries. All countries presented the issues raised above to different degrees. In most cases, the problem was solved using detailed firm- or plant-level accounting information provided by audit- ing companies before privatization. That was the case for Argentina, Colombia, Mexico, and Peru. In Peru, the author also took advantage of privatization agreements that required firms to keep separate books for different units, thereby allowing data aggregation. Other methods in- cluded estimating proxy financial information or disassembling firms into their original constituents.14 When none of these efforts could be undertaken, firms were discarded from the sample to ensure clean estimates. The resulting samples typically excluded the following: · Cases of state-owned enterprises for which data from the preprivati- zation period were missing, often as a result of mergers or spinoffs 18 CHONG AND LÓPEZ-DE-SILANES Table 1.3 Reasons for Firm Exclusion from the Privatization Sample Sale of Merger small Firm Change with minority was Missing in private partici- liqui- informa- Recent accounting Country firm pation dated tion sale standards Argentina Yes Yes Yes Yes -- -- Bolivia Yes -- -- Yes Yes Yes Brazil Yes Yes -- Yes Yes -- Chile Yes -- Yes Yes -- Yes Colombia Yes -- -- Yes -- -- Mexico Yes -- Yes Yes -- -- Peru Yes -- Yes Yes -- -- Note: This table shows the main reasons for excluding some firms from the final sample in each country. "Yes" means some firms were excluded for that particular reason. -- means that the study does not suffer from the potential loss. Source: Chapters 2­8 of this volume. · A few instances of very small state ownership shares being sold (Argentina and Chile), firms that underwent changes in accounting (Bo- livia and Chile), and some very recent privatization cases (Bolivia and Brazil) · Firms that were liquidated after privatization, although robustness checks were applied to ensure results would not be significantly changed with their inclusion. To summarize, several early privatization studies suffered from biases introduced by incomplete samples and the use of poor data when the na- ture of the firm changed with privatization. Today these concerns have largely been put to rest thanks to the recent Latin American studies out- lined in this chapter and other efforts, mainly for eastern European coun- tries, that use comprehensive firm-level data across sectors and company sizes.15 Evidence from Comprehensive Data Samples on Privatization in Latin America This section outlines the evidence on performance changes after privati- zation emerging from the Latin American countries included in our compilation. As previously explained, the data are some of the most comprehensive and up-to-date for the region, allowing us to address THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 19 many of the concerns raised about privatization. We analyze profitabil- ity, operating efficiency, the behavior of inputs, output, and taxes. Latin American studies find improvements in firms' profitability, which is in line with earlier worldwide evidence (Megginson, Nash, and van Ran- denborgh 1994; Boubakri and Cosset 1998, 1999; D'Souza and Meg- ginson 1999). These increases are typically accompanied by reductions in unit costs, boosts in output, and reduced or constant levels of em- ployment and investment. The evidence suggests that higher efficiency, achieved through firm restructuring and productivity improvements, underpins profitability gains. The raw results on firm performance are followed by industry-adjusted information to verify their robustness. Whenever possible, we show the data for median firms, as they are less affected by outliers. Raw Data The evidence from Latin America shows substantial gains in profitability after privatization, measured by ratios of net income to sales and operat- ing income to sales (figure 1.4). For the countries in the sample, the me- dian net-income-to-sales ratio increased 14 percentage points, while the operating-income-to-sales ratio increased 12 percentage points. The largest gains are in Argentina and Peru, where median changes in the ra- tio of net income to sales reached about 20 percentage points, and in Bolivia, where the ratio of operating income to sales increased more than 15 percentage points. Brazil shows the second smallest gains, between 2 and 3 percentage points depending on the ratio. Colombian state-owned enter- prises, unlike their counterparts in other countries, were highly profitable before privatization, which is largely explained by the protective industrial policy implemented by the Colombian government during the 1980s. There is some evidence that profitability in Colombia dropped because firms were already efficient, and privatization was coupled with market liberalization, which brought increased competition. The data for Latin America suggest that the main reason behind the profitability gains is the improved operating efficiency brought about by privatization. In figure 1.5 we explore this issue using costs per unit, the ratio of sales to assets, and the ratio of sales to employment. Costs per unit plummet, with the median decline equivalent to about 16 per- cent for the countries with available data. The results are statistically significant at 1 percent for all countries except Chile. State-owned enterprises were highly unprofitable before privatization in five of the seven countries, with losses above 10 percent of sales in terms of net income over sales. The exceptions are Chile, whose state-owned enter- prises exhibited slightly positive profitability ratios, and Colombia, where the state-owned sector was very profitable compared with private competitors. 20 CHONG AND LÓPEZ-DE-SILANES Figure 1.4 Profitability Changes after Privatization in Latin America Percentage points 24 19 14 9 4 -1 Argentina Bolivia Brazil Chile Colombia Mexico Peru Net-income-to-sales ratio Operating-income-to-sales ratio Note: The components of the variables are defined as follows: net income is equal to operating income minus interest expenses and net taxes paid, as well as the cost of any extraordinary items; operating income is equal to sales minus operating expenses, minus cost of sales, and minus depreciation; and sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. For Bolivia, the net-income-to-sales ratio is not available. Source: Chapters 2­8. The sales-to-asset ratios similarly show a rising trend in four out of five countries. The median country increase in this ratio is 16 percent. Colombia and Peru are the only countries with a fall in sales to assets (about 30 and 20 percent, respectively); in both countries privatized en- terprises engaged in large investments that overtook output increases. Fi- nally, the impact on the sales-to-employment ratio is dramatic, with a median gain of 65 percent. Chile and Mexico show the most impressive results, in that sales per employee doubled. Information for Colombia suggests that state-owned enterprises also underwent restructuring with significant efficiency gains. The mean (median) manufacturing firm in Colombia experienced a 48 (65) percent gain in its sales-to-employment ratio and a 2.4 percent per year increase in its total factor productivity index. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 21 Figure 1.5 Operating Efficiency Changes after Privatization in Latin America Percentage change 120 100 80 60 40 20 0 -20 -40 Bolivia Brazil Chile Colombia Mexico Peru -60 Cost per unit Sales to assets Sales per employees Note: Cost per unit is defined as the ratio of cost of sales to sales. The components of the variables are defined as follows: cost of sales is equal to the direct expense involved in the production of a good (or provision of a service), including raw material expenditure plus total compensation paid to blue-collar workers; sales are equal to the total value of products and services sold nationally and internationally minus sales returns and discounts; employees corresponds to the total number of workers (paid and unpaid) who depend directly on the company; and assets are defined as property, plant, and equipment (PPE), which is equal to the value of a company's fixed assets adjusted for inflation. For Brazil, the sales-per-employees ratio is not available. Source: Chapters 3­8. As figure 1.6 shows, labor retrenchment is a significant component of the privatization experience in Latin America. Privatized firms reduced a substantial percentage of their work force in almost all countries. The exception to this trend is Chile, where the mean number of workers in pri- vatized firms increased by 15 percent and the median fell by 5 percent. In general, the median country reduced 24 percent of its work force. Priva- tized state-owned enterprises in Bolivia, Colombia, Mexico, and Peru show significant reductions: the median firm fired 13 percent, 24 percent, 57 per- cent, and 56 percent of the work force, respectively. The magnitude of em- ployment reductions in these countries speaks of state-owned firms with bloated work forces, providing evidence in line with the political economy view of the benefits of privatization. The evidence on labor cuts suggests 22 CHONG AND LÓPEZ-DE-SILANES Figure 1.6 Percentage Changes in Employment after Privatization in Latin America a. Mean values Percentage change 20 Argentina Bolivia Chile Colombia Mexico Peru 10 0 -10 -20 -30 -40 -50 -60 -70 Number of employees Industry-adjusted number of employees b. Median values Percentage change Bolivia Chile Colombia Mexico Peru 0 -10 -20 -30 -40 -50 -60 Number of employees Industry-adjusted number of employees Note: The number of employees corresponds to the total number of workers (paid and unpaid) who depend directly on the company. The industry-adjusted number of employees is computed by augmenting the preprivatization number by the difference between the cumulative growth rate of the number of employees of the firm and the cumulative growth rate of the number of employees of the control group in the postprivatization period relative to the average number of employees before privatization. For Argentina, the mean number of employees is not available; for Chile and Peru, the median industry-adjusted information is not available; for Bolivia, the industry-adjusted information is not available. Source: Chapters 2, 3, 5, 6, 7, and 8. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 23 that transfers from workers to shareholders may be a significant component of the success of privatization. We explore this issue later in the chapter. A priori, the impact of privatization on investment is not clear. One could expect privatized firms to avoid new investments since state-owned enterprises usually had ample idle capacity. At the same time, if the pro- duction process used by the state-owned firm is outdated, one could ex- pect a large increase in investment. The data for Latin America confirm the initial hypothesis, since investment exhibits modest gains or statistically insignificant changes. The exception is Argentina, where investment in- creased by more than 350 percent. Our analysis so far suggests that the profitability gains of privatized firms stem mostly from efficiency gains. Most countries show drastic cuts in employment and fairly consistent capital stocks. Perhaps the most strik- ing finding is that the output of privatized state-owned enterprises increased dramatically, despite dwindling employment and modest invest- ment (figure 1.7). The median firm in our sample increased output by over 40 percent, with the largest gains achieved by Mexico and Colombia, where median output increased 68 percent and 59 percent, respectively. The country with the lowest, albeit significant, increase in output is Brazil, where real sales went up 17 percent. Adjusted Ratios Latin America underwent major economic transformations in the 1990s as countries embraced liberal policies and opened up their borders. Most of these countries expanded and contracted at various points, leading to concerns about the interpretation of the evidence just discussed. In partic- ular, one might argue that the large profitability and output increases and the rapid growth in productivity can only be the result of macroeconomic and industry changes in the region. To isolate the role of privatization, the series of studies in our compilation present industry-adjusted measures, which support the patterns discussed so far. The data displayed in figure 1.7, for example, allow us to rule out macroeconomic factors as the driving force behind postprivatization out- put growth: median industry-adjusted sales grew 27.5 percent in the re- gion. In Brazil and Peru, matching private firms basically stagnated, while the median industry-adjusted output of privatized firms in those countries increased at about the same rate as the raw numbers. Meanwhile, the improved economic conditions and industry factors in Mexico and Colombia accounted for about one-fifth and three-fifths of output growth, respectively. Relative to industry benchmarks, the median (mean) employment of privatized firms fell roughly 20 (35) percent in the region (see figure 1.6). In contrast, relative investment behavior differs across countries. Ratios of median industry-adjusted investment to sales and investment to assets fell 24 CHONG AND LÓPEZ-DE-SILANES Figure 1.7 Median Changes in Output after Privatization in Latin America Percentage change 80 70 60 50 40 30 20 10 0 Argentina Bolivia Brazil Chile Colombia Mexico Peru Output Industry-adjusted output Note: Output is defined as the monetary value of sales. The industry- adjusted output is computed by augmenting the preprivatization value by the difference between the cumulative growth rate of output of the firm and the cumulative growth rate of output of the control group in the postprivatization period relative to the average level of output before privatization. For Colombia, the information corresponds to mean values; for Peru, industry-adjusted output information is expressed in mean values; for Argentina and Chile, output information is not available; for Bolivia, industry-adjusted output information is not available. Source: Chapters 2­8. considerably in Brazil and Mexico but showed a marked increase in Argentina, Chile, and Colombia. The second most important finding of this section involves the closing performance gap between privatized and comparable private firms after privatization (figure 1.8). Mexico offers the most dramatic example of con- vergence: the net-income-to-sales gap between state-owned and private firms disappeared with privatization and even turned slightly in favor of the privatized enterprises. The Argentine data, although not in a comparable format, also show a similar pattern of catching up. The industry-adjusted net-income-to-sales ratio increased 188 percent after privatization, while the operating-income-to-sales ratio rose 129 percent. The profitability gap THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 25 Figure 1.8 Net-Income-to-Sales Gap between Privatized and Private Firms before and after Privatization Percentage points 25 20 15 10 5 0 -5 -10 -15 -20 Brazil Chile Colombia Mexico Before privatization After privatization Note: Net income is equal to operating income minus interest expenses and net taxes paid, as well as the cost of any extraordinary items; sales are equal to the total value of products and services sold nationally and internationally, minus sales returns and discounts. For Colombia, information is from the energy sector. Source: Chapters 4, 5, 6, and 7. between Colombian privatized and private firms also closed, albeit from a different starting point.16 Before privatization, the median firm in manu- facturing was almost 4 percent more profitable than its private counter- parts, while in the state-owned energy sector, this difference was about 20 percent. Substantially lower levels of protection of these firms explain the narrowing gap with the private sector after privatization. Finally, the Brazilian and Chilean privatized samples also improved their relative prof- itability with respect to their industry competitors. In the case of Brazil, privatized state-owned enterprises became slightly more profitable than their private competitors, while the gap between Chilean privatized and private firms narrowed by about 20 percent. The gap between privatized and private firms also closed in terms of unit costs (figure 1.9). Brazilian privatized firms quickly reduced a gap of 26 CHONG AND LÓPEZ-DE-SILANES Figure 1.9 Cost-per-Unit Gap between Privatized and Private Firms before and after Privatization Percentage points 16 14 12 10 8 6 4 2 0 -2 Brazil Chile Mexico Before privatization After privatization Note: Cost per unit is defined as the ratio of costs of sales to net sales. Cost of sales is equal to the direct expense involved in the production of a good (or provision of a service), including raw material expenditure plus total compensation paid to blue-collar workers. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Source: Chapters 4, 5, and 7. 9 percentage points to about 2 percentage points. In Chile, this gap was 2­3 percentage points both before and after privatization. In Argentina, industry-adjusted unit costs for privatized firms declined 10 percent. Meanwhile, Mexico's privatized state-owned enterprises substantially cut costs to eliminate a large 14 percentage point gap with private competi- tors. The catching-up effect of privatization is explained by the large gains in operating efficiency that more than survive industry adjustments. Rela- tive to industry benchmarks, median sales per employee went up 9 percent in Argentina, 30 percent in Bolivia, and a massive 88 percent in Mexico. Similarly, median industry-adjusted sales-to-asset ratios increased 20 per- cent in Mexico, 34 percent in Brazil, and 49 percent in Chile. All of these numbers suggest that a large component of the higher profitability comes from improved efficiency, lining up with the rest of the evidence presented in the following section. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 27 Who Wins and Loses from Privatization? Concerns about Exploitation of Market Power, Workers, and the Government Some of the main criticisms of privatization are based on the belief that the gains in firm profitability are achieved at the expense of society. These gains are claimed to be extracted from consumers through the use of mar- ket power, from workers by means of lower salaries, and from the gov- ernment (Campbell-White and Bhatia 1998; Bayliss 2002). In this section we use the recent empirical evidence from Latin America and elsewhere to assess the sources of profitability gains of privatized state-owned enterprises. Government Revenues Critics of privatization often argue that the government, and thus soci- ety at large, loses from privatization because it gives up a positive stream of cash flows and puts it in the hands of private buyers. The ar- gument is extended to claim that the sale of state-owned enterprises is equivalent to a privatization of gains and a socialization of losses. In other words, well-connected groups are able to reap the profits of pri- vatized firms and receive government-sponsored bailouts when things go wrong. The evidence used to support these claims comes mostly from case studies of profitable state-owned enterprises that were privatized, unprofitable state-owned enterprises that turned out to be great mon- eymakers after privatization, and state-owned enterprises that became money losers and went into financial distress. This perception has swayed public opinion because of the excessive costs levied on society in some cases of botched privatizations. In Mexico, for example, the bailouts granted to keep banks and highways from going bankrupt in- creased public debt from less than 25 percent of GDP to over 50 percent (López-Calva 2004). The underlying logic of these arguments is similar to that undergirding the arguments for the economic benefits of state production, which in the 1950s and 1960s justified the existence of state-owned enterprises on the grounds that they help solve market failures by taking into account the so- cial costs of their actions. Today, academic evidence of the opposite abounds in at least three areas. First, systematic evidence shows that state- owned enterprises are less efficient than private firms in industrial and developing countries (Shleifer and Vishny 1994; Shleifer 1998). Second, the inefficiency of state-owned enterprises may be the natural result of politi- cal meddling when governments use them to achieve political objectives. This political use of state production leads to excessive employment, inef- ficient investments, and inadequate location of production sites, among 28 CHONG AND LÓPEZ-DE-SILANES other things (López-de-Silanes, Shleifer, and Vishny 1997). Finally, the large body of empirical work generated since the mid-1990s (reviewed in previous sections) shows that by and large privatization leads to substan- tial increases in the profitability of firms. Criticisms of privatization that center on what the government gives up disregard the fact that state-owned enterprises are typically money-losing entities before privatization. Moreover, the visible losses may underesti- mate the real bottom line, because their precise magnitude is obscured by large cross-subsidies from other state-owned enterprises and soft loans from the government. In fact, tax collection from state-owned enterprises improved after privatization in most Latin American countries analyzed here. Brazil, the country with the smallest gains in profitability, experi- enced a 1 percentage point decrease in its net-taxes-to-sales ratio, although that ratio was still positive after privatization (the difference is not statis- tically significant). In Mexico, the same ratio increased 7.6 percentage points. We do not have direct information for Argentina, Chile, and Peru, but given that net income over sales increased between 12 and 20 per- centage points, it is safe to assume that net taxes over sales also increased by a few percentage points. Increased fiscal revenues mean more resources that can be channeled to address pressing social needs, thereby benefiting society at large. Higher tax revenues, if managed appropriately, should allow govern- ments an increased capability for welfare-improving activities to benefit the poorest segments of society. Argentina, Mexico, and Peru are exam- ples of countries where privatization revenues and the increased tax receipts from firms that formerly did not make profits were probably large enough to offset the costs of job losses (Rama 1999; Chong and López-de- Silanes 2003). Privatization revenues need not be a blessing, however, if they are misused. Anuatti-Neto and his coauthors point out that in Brazil privatization brought about high macroeconomic costs because the rev- enues it produced may have delayed fiscal adjustment and helped prop up an overvalued currency. This is obviously not an argument against priva- tization but rather an argument against the political misuse of the resources it generates. Overall, the empirical literature on privatization shows that it affects the government's budget by reducing previous subsidies to state-owned enterprises, raising substantial revenue from the sale, and generating taxes on the increased profits. The benefits of a well-managed privatiza- tion program could be substantial not only for the privatized firm but also for society. Worker Exploitation The second potential source of gains after privatization is transfers from workers to shareholders, because cuts in labor costs may account for a THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 29 large fraction of reduced total costs. Labor cost reductions can come from two sources: fewer workers, or lower wages and benefits. As explained above, the set of papers presented here finds that direct employment by the median state-owned firm falls between 5 and 57 percent after privatization depending on the country (see figure 1.6). Layoffs explain part of the cost reduction and thus higher profits after privatization. The other potential component is cuts in wages and benefits. The hypothesis that privatization leads to redistribution of income from workers to the new owners predicts a reduction in real wages and benefits for those workers who remain in the firm. Data on wages at the firm level are scarce, but for those countries with available information (Argentina, Bolivia, Colombia, Mexico, and Peru), the evidence shows the exact opposite: an increase in the real and industry-adjusted wages of workers in privatized firms (figure 1.10). Both real and industry-adjusted wages for the median firm increased by about 100 percent in Mexico and Peru. Bolivia enjoyed real wage increases of almost 110 percent, while in Argentina the industry-adjusted increase was about 70 percent. Colombia shows the smallest increase, but even here workers in privatized firms increased their wages more than others in the private sector. The two components of the transfers from workers to profits move in opposite directions. The fraction of profitability changes that may be attributed to labor cost savings thus encompasses the lower costs stemming from layoffs and the higher costs from wage increases for the remaining workers. Following the methodology in La Porta and López-de-Silanes (1999), the studies on Argentina, Bolivia, Mexico, and Peru compute the impact on profits from lower labor costs after privatization. The evidence from these four countries shows that even with the extreme assumption that laid-off workers had zero productiv- ity, the median savings from labor costs is equivalent to 16 percent of the gains in net income to sales after privatization or 20 percent of the gains in operating income to sales. The range of calculations extends from close to 5 percent in Peru to 45 percent in Mexico (figure 1.11). If we assume that these workers are half as productive as those re- tained by the firm, the median savings from reduced labor costs for the countries with data falls to 8 percent of the gains in net income to sales and 10 percent of the gains in operating income to sales. Overall, the evidence indicates that labor cost reductions are a source of the gains after privatization, but these savings do not explain the bulk of the higher observed profitability. The welfare of displaced workers after privatization is another issue for consideration. The calculations above overstate the losses to workers to the extent that some of those laid off found alternative employment or attach some value to leisure. There is evidence that this is in fact the case; Galiani and his coauthors, for example, carried out a survey among displaced workers in Argentina. They found that the labor force 30 CHONG AND LÓPEZ-DE-SILANES Figure 1.10 Median Real and Industry-Adjusted Changes in Wages after Privatization Percentage change 200 180 160 140 120 100 80 60 40 20 0 Argentina Bolivia Colombia Mexico Peru Real wage increase Industry-adjusted wage increase Note: Real average wages are defined as the inflation-adjusted total compensation paid to the average worker. The Consumer Price Index was used as a deflator to calculate real wages. Industry-adjusted wages are computed by augmenting the preprivatization value by the difference between the cumulative growth rate of real wages per worker of the firm and the cumulative growth rate of real wages per worker of the control group in the postprivatization period relative to the average real wage per worker before privatization. For Bolivia, Mexico, and Peru, information is for a subsample of firms that have available wage evidence. Source: Chapters 2, 3, 6, 7, and 8. participation rate was high among such workers and that although unem- ployment rates were above those of the rest of the population, many dis- placed workers found alternative jobs in which they felt their situation was stable. Taking all factors into account, these authors estimate the welfare loss to displaced workers was equivalent to 39­51 percent of their earn- ings before privatization and that 40 percent thought they were not worse off after privatization. This is surprising, since most theories and evidence suggest that workers in state-owned enterprises are overpaid and have very low productivity. Further work is needed in this area to provide clearer THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 31 Figure 1.11 Transfers from Workers as a Percentage of Increased Profitability after Privatization Proportion of the gain (percent) 50 45 40 35 30 25 20 15 10 5 0 Argentina Bolivia Mexico Peru Net-income-to-sales ratio Operating-income-to-sales ratio Note: The figure shows the median gain in net income to sales and operating income to sales explained by savings in labor costs stemming from layoffs after privatization. Savings from layoffs are calculated as follows: WAGEbp(Lbp Lap) SALESap where WAGEbp is the average wage of employees in state-owned enterprises before privatization; Lbp is the number of workers employed before privatization; Lap is the number of workers employed after privatization; and SALESap is the monetary value of sales after privatization. The resulting number is thus expressed as a fraction of sales. We then divide by the percentage point increase in the operating-income-to-sales ratio to determine the percentage of the increase that results from transfers from workers. For Bolivia, Mexico, and Peru, information is for a subsample of firms that have available wage evidence. For Bolivia, net-income-to-sales data are not available. For Peru, data on savings in labor costs as a percentage of operating income to sales are not available. Source: Chapters 2, 3, 7, and 8. 32 CHONG AND LÓPEZ-DE-SILANES evidence on the extent of welfare losses to workers, but the available evidence thus far suggests that while laid-off workers do lose in this process, the losses may not be as large as previously thought. Finally, privatization could also have compositional effects on the la- bor force and hurt unskilled workers disproportionately. The empirical evidence on this issue is inconclusive for the two Latin American coun- tries with disaggregated wage and employment data, but it suggests that blue-collar workers actually fare better than their white-collar counter- parts. In Bolivia, only 5 percent of blue-collar workers were laid off, while over 27 percent of white-collar workers were fired by the median firm. Moreover, unskilled workers who remained saw their real wages increase 103 percent, compared with a 99 percent rise for skilled work- ers. In Mexico, blue-collar workers suffered higher layoffs than white- collar ones in the median firm: 61 percent (32 percent industry-adjusted) for blue-collar workers versus 46 percent (31 percent industry-adjusted) for white-collar workers. Wages again exhibited the same trend as in Bolivia, with sharp rises in blue-collar real and industry-adjusted wages (148 percent and 122 percent, respectively) and smaller, though still sub- stantial, wage increases for white-collar workers (100 percent real and 48 percent industry-adjusted). Therefore, for neither of these countries can we conclude that unskilled workers fared worse than skilled labor as a result of privatization. Abuse of Market Power and Consumer Exploitation The last concern about the sources of postprivatization gains is that the increase in firm profitability may come at the expense of consumers through weak regulation and abuse of market power. The papers pre- sented here provide useful data for assessing these claims. If market power is a significant determinant of the gains, we should expect firms in noncompetitive sectors to experience large gains in operating income ow- ing to higher product prices. Since profits are likely to be higher in non- competitive sectors than in competitive sectors both before and after pri- vatization, the relevant comparison for establishing the facts for this section is relative changes among privatized firms in competitive and non- competitive sectors. For the Latin American countries with data disaggregated by compet- itive and noncompetitive sectors, we find that changes in profitability are generally larger in the competitive sector than among noncompetitive industries. This evidence goes against the hypothesis that market power explains most of the gains. As figure 1.12 shows, the median ratio of op- erating income to sales in Mexico increased 14.5 percentage points for privatized firms in the competitive sector and only 7.5 points for firms in noncompetitive industries. Competitive firms in Colombia performed relatively better than their noncompetitive counterparts: their median THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 33 Figure 1.12 Median Changes in Profitability of Privatized Firms in Competitive and Noncompetitive Industries in Latin America Percentage points 35 30 25 20 15 10 5 0 -5 -10 -15 Chile Colombia Mexico Peru -20 Competitive industries Noncompetitive industries Note: Profitability is defined as the median ratio of operating income to sales, except for Peru where it is the mean net-income-to-sales ratio. For Chile, firms are classified as noncompetitive if they are in telecommunications, electricity, or social services sectors; and as competitive if they are not. For Colombia, noncompetitive firms are those in the energy sector; all other sectors are considered competitive. For Mexico, firms are classified into competitive and noncompetitive based on the description of the industry provided by the privatization prospectus of the firm. For Peru, the noncompetitive sectors are electricity, financial, and telecommunications, and the data for the competitive industries show aggregate information for the whole sample. For Peru, the information is expressed in mean values. Source: Chapters 5­8. profitability decreased by only 2 percentage points, compared with the 13 point drop for noncompetitive sectors, which underwent severe deregulation. Data for Peru reinforce this trend. Firms in noncompetitive sectors increased their profitability by an average of 27 percentage points, while the mean increase in the whole sample was 32 percentage points. In Chile, although the noncompetitive sectors' profitability in- creased more (8.5 percentage points), the increase is not statistically dif- ferent from the 5.5 percentage point increase in competitive sectors. 34 CHONG AND LÓPEZ-DE-SILANES Regression analysis for Peru using concentration proxies also con- tributes to assessing the role of market power. Confirming the trend above, market concentration in Peru was found not to be a significant de- terminant of profits. Finally, information on firms' product prices before and after privatization in Mexico also suggests that market power is not a large source of gains. Cumulative price increases in the noncompetitive sector in Mexico were only 6 percent higher than the growth of the in- dustry-matched producer price index over the postprivatization period. La Porta and López-de-Silanes (1999) use these product price data to draw a quick calculation of the contribution of changes in prices to the observed change in profitability of the whole sample of privatized firms. Their data show that price increases accounted for only 5 (7) percent of the change in mean (median) operating income to sales after privatization.17 If market power were an important source of profits for privatized firms, those in noncompetitive sectors would show lower growth in em- ployment, investment, and output than firms in competitive sectors (La Porta and López-de-Silanes 1999). Available evidence for Latin America does not support these claims (figure 1.13). In Mexico and Colombia, em- ployment dropped 46 percent and 24 percent, respectively, for firms in the competitive sector, and it decreased only 19 percent and 10 percent for noncompetitive firms. In Chile, the pattern is even more striking: employ- ment increased in both sectors, rising 16 percent in competitive industries and 32 percent in noncompetitive sectors. For Peru, employment data show no divergence in results between competitive and noncompetitive sectors, as employment fell 50 percent in noncompetitive sectors and 51 percent for the whole sample. Output growth data for Mexico and Peru reinforce this trend. In Peru, output growth for both sectors was very sim- ilar, with noncompetitive firms increasing sales 47 percent and the sales of the whole sample going up 50 percent. Similarly, in Mexico, output of competitive firms increased 56 percent, while sales in the noncompetitive sector went up 78 percent. Additional evidence comes from investment patterns. Investment per employee grew 49 percent and 154 percent in the noncompetitive sectors of Mexico and Colombia, respectively. Meanwhile, the same ratio grew only 29 percent in competitive sectors of Mexico and stagnated in Colom- bia's competitive industries. The evidence for Chile here runs in the oppo- site direction, but it is hardly conclusive of market power abuse. Although investment per employee grew 74 percent in Chile's competitive sectors, it also grew almost 50 percent in noncompetitive industries. Overall, the Latin American evidence presented in this section does not support the claim that consumer exploitation is a significant source of pri- vatization gains. These studies suggest that a large source of the gains may lie in deep firm restructuring that leads to lower costs and higher effi- ciency. Evidence from Chile and Mexico is suggestive of this pattern. Unit costs in the competitive sector fell 3 percent in Chile and 13 percent in THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 35 Figure 1.13 Changes in Employment and Output of Privatized Firms in Competitive and Noncompetitive Industries in Latin America a. Employment Percentage change 40 Chile Colombia Mexico Peru 30 20 10 0 -10 -20 -30 -40 -50 -60 Competitive industries Noncompetitive industries b. Output Percentage change 90 80 70 60 50 40 30 20 10 0 Chile Colombia Mexico Peru Competitive industries Noncompetitive industries Note: Employment corresponds to the total number of workers (paid and unpaid) who depend directly on the company; output is the monetary value of sales. For definitions of competitive and noncompetitive firms, see figure 1.12. For Chile, output information is not available. For Peru, the information is expressed in mean values. Source: Chapters 5­8. 36 CHONG AND LÓPEZ-DE-SILANES Mexico, while those of noncompetitive industries decreased 8 percent and 24 percent, respectively. Abuse of market power may be an issue for some firms, but the bulk of the evidence suggests it is not the main explanation of privatization gains across the board. Dimensions of Consumer Welfare beyond the Effect on Prices Beyond its effect on prices, privatization may affect consumer welfare through decreased access, worsened distribution, and reduced quality of goods and services (Akram 2000; Bayliss and Hall 2000; Bayliss 2001; Birdsall and Nellis 2002; Freije and Rivas 2002). These concerns are significant because the poorest segments of society are generally the main consumers of goods and services previously produced by state- owned enterprises. The evidence presented earlier on increased output, firm restructuring, and prices should alleviate some of these concerns, particularly for the case of standardized goods and products. Output and price are suitable proxies for measuring the availability of most of these goods. In the area of public utilities and services, however, access and distribution may still be a concern, since some segments of the popula- tion may lack access to the network and may thus be unable to purchase these services regardless of their price. Similarly, the quality of services such as water, electricity, telecommunications, or transportation may be reduced to try to meet price regulations. In all of these circumstances, consumer welfare may suffer as a result of privatization. Some reviews of privatization cases are pessimistic about its success in the utilities sector. Bayliss (2002) points to examples of botched privati- zations in Puerto Rico and Trinidad and Tobago, where water privatiza- tion led to price hikes and no apparent improvement in provision. Simi- larly, the privatization of the electricity sector in the Dominican Republic is claimed to have led to frequent blackouts and increases in utility prices, culminating in civil unrest and the deaths of several demonstrators (Bayliss 2002). One can always find cases of failure and cases of success. Therefore, the only way to address this question fully is to gather data that allow a systematic and economically robust analysis. A first generation of privatization studies sheds light on this subject by analyzing case studies in several countries. Galal and others (1994), for ex- ample, analyze 12 privatization cases in Chile, Malaysia, Mexico, and the United Kingdom, including firms in sectors such as airlines and telecom- munications. Their results indicate that privatization led to welfare gains of about 25 percent of preprivatization sales in 11 of the 12 cases. Early work on the privatization experience in Argentina also shows significant gains in access to services such as water, power, and port infrastructure (Carbajo and Estache 1996; Crampes and Estache 1996; Estache and Ro- dríguez 1996). Ramamurti (1996, 1997) concludes that privatization had THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 37 a positive effect on Latin American telecommunications and railroad in- frastructure because it led to a technological overhaul of the sectors and increased both access and the quality of service. Similarly, Ros (1999) ex- amines the effect of privatization on the telecommunications sector in 110 countries and finds that the transfer of control from the public to the pri- vate sector led to significantly higher telephone density levels. Although the level of competition had a positive effect on industry efficiency, only privatization was related to network expansions. A new generation of studies based on more detailed data and new econometric approaches corroborates the early results in terms of access and quality. For instance, Torero and Pasco-Font (2001) show that the number of telephone lines in Peru increased from 2.9 to 7.8 per 100 in- habitants and that the electrification coefficient jumped from 48 to 70 percent between 1993 and 1998. A study by Torero, Schroth, and Pasco- Font (2003) tests the impact of the privatization of telecommunications on the welfare of urban consumers in Peru; the authors find significant welfare gains and dramatic improvements in terms of efficiency, access, and quality of service. Similarly, Fischer, Gutiérrez, and Serra, in chapter 5, find improvements in access and service quality in the telecommunica- tions sector in Chile, where the number of phone lines in operation in- creased sixfold, bringing density levels from 4.7 lines per 100 inhabitants in 1987 to 23.1 lines in 2001. The average length of the waiting period for a new phone line dropped from 416 days in 1993 to only 6 days in 2001, while the waiting list for a phone dropped from a peak of 314,000 households in 1992 to only 32,000 by 2001.18 The region offers a number of similar examples of improvements in ac- cess to water, electricity, telecommunications, and other services that have created benefits beyond lower prices. Nonetheless, one may still be con- cerned about the distributional impacts of the increased coverage, as it may not be reaching the poorest sectors of society. Bayliss (2002) recog- nizes that privatization has the potential for welfare-enhancing outcomes if it allows low-income households to gain access to the service network. However, her review of cases suggests that the drive to seek higher profits in the private provision of services will almost invariably lead to a loss for the poor. Birdsall and Nellis (2002) also argue that privatization may lead to improvements in efficiency and profitability while worsening income distribution and wealth.19 They conclude that the gains in profitability are probably not worth the distributive effects they create. Recent detailed econometric analyses with better samples provide some answers to these concerns. In chapter 2 Galiani and his coauthors offer some of the best data available for the municipal level in Argentina, where about 30 percent of localities privatized water delivery services. Their re- sults show a significant increase in the proportion of households con- nected to water services in municipalities that privatized compared with those that did not. Their regression estimates suggest that the number of 38 CHONG AND LÓPEZ-DE-SILANES households connected to the water network increased by 11.6 percent as a result of privatization (with the exception of Buenos Aires, where 98 per- cent of households were already connected). Using less comprehensive data from Bolivia, Barja, McKenzie, and Urquiola (2002) find that priva- tization increased access to water relative to both the existing trend and nonprivatized areas. They further report that the relative benefits of water privatization are greatest for the poorest segments of the population, who gained from the largest increases in access. For Argentina, Galiani and his coauthors cleverly design tests that map water delivery to infant mortality in an effort to directly address the concerns about quality after privatization. Their regressions show that, controlling for other factors, child mortality fell by 5 to 7 percent more in areas of Argentina that privatized water services than in those that did not. The effect was largest in the poorest municipalities that privatized, where child mortality fell 24 percent. Privatization translated into 375 child deaths prevented per year. McKenzie and Mookherjee (2003) provide an overview of four studies from Argentina, Bolivia, Mexico, and Nicaragua that use household surveys to measure the im- pact of privatization on welfare. They conclude that the sale of state- owned enterprises brought positive welfare effects and that the poorest segments of the population appear to be relatively better off. In Ar- gentina, for example, they report falling electricity prices that improved the welfare of all income deciles. For Bolivia, they report welfare gains from increased electricity access for all but the top income deciles; the gains exceeded 100 percent for the lowest deciles despite real price in- creases. The price of electricity increased in Nicaragua, but the welfare loss to households that already had access was less than 1 percent of their per capita expenditure, because the budget share allocated to elec- tricity is typically low. At the same time, the value of gaining access to electricity was positive and of a larger magnitude for lower-income deciles that had relatively less access before privatization. The net posi- tive impact of electricity privatization for these low-income groups reached nearly 16 percent of per capita expenditure. So far, we have provided evidence that counters most of the criticisms of privatization. What remains unaddressed, however, is how to make sense of the cases of privatization failures pointed out by several authors (see Bayliss 2002 and Birdsall and Nellis 2002 for reviews). It will always be possible to find instances of failed privatizations, but analysts should not distort this information and turn it into an argument against privati- zation itself. The overwhelming evidence showing that it can be done right suggests that we should look for the reasons why it failed in certain instances. In the next two sections, we argue that many of these failures have two roots: the role of politicians in the privatization process, which may lead to corruption, renegotiation, and opportunistic behavior; and the lack of an appropriate postprivatization regulatory and corporate THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 39 governance framework that sets the boundaries for nonabusive corporate behavior and facilitates investment. What Is the Best Approach for Selling? Concerns about the Privatization Process Privatization requires heavy government involvement because the politi- cians involved are frequently setting up the method and running the sale process. This may lead to favoritism and nepotism (Perotti 1995; Bor- tolotti, Fantini, and Scarpa 2001; Biais and Perotti 2002; Earle and Gehlbach 2003). Looking at the privatization process in this light shows the relevance of understanding the impact of the process's characteristics and the opportunities for corruption they may provide. Privatization may be the last chance for politicians to appropriate cash flows or de- liver favors that further their political objectives. The role of politicians in privatization is central in three areas: the method of privatization cho- sen, the restructuring of firms before they are sold off, and the types of contracts written. The Method of Privatization The way the privatization process is carried out is of utmost impor- tance. A successful program can increase social welfare and bring about efficiency gains across the board, while a botched process may create opportunities for inefficiency and corruption. In Argentina, as in other countries, an obscure bidding process raised suspicions of corruption and political favoritism. When governments fail to ensure a crystal clear process, the perception of corruption can breed unease among the pub- lic and may lead to a backlash against privatization. In principle, a clear and homogeneous privatization process should be established from the start, and special emphasis should be placed on making the auction re- sults as transparent as possible. In reality, however, only a handful of countries have followed this path. Many fail to establish clear guidelines because their privatization programs were originally planned as small affairs or because they lack the necessary skills to do so. Alternatively, politicians may have strong incentives to create obscure and arbitrary privatization mechanisms that allow them to extract high rents for themselves or their constituencies. To analyze the validity of such claims empirically, one could use systematic evidence of the impact of the pri- vatization process on sale prices and on subsequent firm performance. This is difficult although not impossible. The existing empirical literature has taken two approaches to address these issues. The first approach uses cross-country comparisons. Chong and Riaño (2003), for example, analyze 285 privatizations in industrial 40 CHONG AND LÓPEZ-DE-SILANES and developing countries and find that bureaucratic quality, lack of cor- ruption, and privatization prices are positively related. Their results show that when they control for macroeconomic conditions and firm character- istics, a 1 point increase in their 10 point index of bureaucratic quality is associated with a 10.2 percent increase in the price paid per dollar of assets in privatizations, while a similar increase in the 10 point lack-of-corrup- tion index results in a 9.6 percent rise in the price paid per dollar of assets. Bortolotti, Fantini, and Scarpa (2001), who analyze data for 49 countries, conclude that strong legal institutions and adequately developed capital markets substantially contribute to successful privatizations. Finally, Chong and Sánchez (2003) provide data for infrastructure privatization contracts in Brazil, Chile, Colombia, and Peru to show that establishing a clear and transparent contractual arrangement helped achieve the privati- zation objectives set out by these governments. These results together sug- gest that the success or failure of privatization programs is influenced by the honesty and efficiency of the government and by the simplicity and transparency of contractual agreements. The second approach to analyzing the impact of the method of privati- zation is to use within-country data. López-de-Silanes (1997) for Mexico and Arin and Okten (2002) for Turkey are able to control for potentially omitted variables and therefore provide a full analysis of the impact of sev- eral restructuring measures and privatization mechanisms on the net price of state-owned enterprises.20 The case of Mexico is a good illustration of the impact of specific differences in the privatization process, since the program lasted for more than a decade and was executed by different administrations. An additional benefit of this sample is that although the general method of a first-price sealed-bid auction was the rule throughout the period, certain firms were privatized with specific requirements that provide useful variations to analyze. Between 1982 and 1988 privatization was not conducted as a centralized program, but rather each ministry was allowed to sell enterprises in its realm of operations. This resulted in a plethora of requirements for bidders and methods of payment. The administration that took power in 1988 established a centralized privati- zation office and developed a homogeneous process, which improved transparency by mandating public disclosure of the bidding stages through the press. Econometric estimations show that once the analyst controls for macroeconomic and firm-level characteristics, firms privatized during the second period sold at a premium of about 15 percent (López-de-Silanes 1997). The gains in efficiency owing to improved coordination and the presumably reduced room for corruption and political meddling have a clear mapping in the price received for enterprises sold. Econometric work with firm-level data from Mexico also shows that different auction requirements make a substantial difference in the net price received by the government for state-owned enterprises. Firms sold under restrictions banning foreign bidders, requiring a prequalification THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 41 stage, or asking for cash-only payments brought significantly lower prices per dollar of assets sold. Such requirements thus have an effect that is in- dependent of the fact that they reduce competition in the auction; this evidence suggests that idiosyncratic and arbitrary privatization processes come at a direct cost to the government in terms of the price paid for state- owned enterprises. The speed at which each privatization takes place may also have an im- pact on net prices raised. The theoretical literature is split between the ben- efits and costs of a short process. While rushing a sale carries potential costs such as not attracting enough bidders or not having enough time to set up an appropriate regulatory framework, the advocates of a speedy process point to the benefits of quickly disposing of money-losing firms and avoiding costly restructuring (Coes 1998). The recent literature addresses this issue by measuring the impact of the length of the privati- zation process on the price paid for the specific state-owned enterprise. Some believe that a lengthy privatization process should come at no cost, either because managers' concern for their reputation will lead them to run the firm efficiently or because the announcement of privatization may im- prove stakeholders' incentives and thus boost company performance (Caves 1990; Bolton and Roland 1992). Conversely, to the extent that the privatization process is similar to the situation of a firm in financial dis- tress, the privatization announcement may be followed by a deterioration of incentives and performance (Altman 1984; Wruck 1990). Within-country firm-level panel data are ideally set up for resolving this dispute. Evidence from Mexico and Turkey shows that after one controls for firm and industry characteristics, lengthy privatization processes come at a substantial cost to the government. The announce- ment of privatization in these countries brought a considerable deterio- ration in performance, which is probably attributable to the collapse of managers' incentives and to the performance of disgruntled workers who see their futures as highly uncertain.21 Restructuring Firms before Privatization Government restructuring of state-owned enterprises before their sale is an issue likely to be fraught with political difficulties given that this is probably the last chance for government officials to extract benefits. As with other policies, restructuring programs can be defended rationally on grounds that they may increase revenues from the sale or ensure that firms are sent out to the market in the best condition to minimize layoffs and secure their survival (Nellis and Kikeri 1989; Kikeri, Nellis, and Shirley 1992; Kikeri 1999). As a result, there is great ambivalence about the opti- mal policy approach toward restructuring before privatization. López-de-Silanes (1997) summarizes the theoretical arguments for and against various measures of prior restructuring and suggests that the issue 42 CHONG AND LÓPEZ-DE-SILANES should be resolved empirically. This is not a straightforward proposition, however, even with firm-level data. Restructuring measures are not un- dertaken randomly but are selectively targeted to firms that need them most. We would expect the government to absorb the debt of highly indebted state-owned enterprises, to fire workers when firms face serious overem- ployment, and to invest in new machinery when production processes are outdated. If the endogenous nature of these measures is not considered, we run the risk of reaching the wrong conclusions because regression coeffi- cients would capture not only the effect of the restructuring measure, but also the negative effects of being in distress or having a bloated work force. Available empirical evidence strongly suggests that restructuring policies do not lead to better net prices per dollar of assets sold. For the case of Mexico, López-de-Silanes (1997) shows that, after he controls for endogeneity, the optimal policy seems to be to refrain as much as possible from engaging in the restructuring of state-owned enterprises. Some of the most popular measures, such as debt absorption, do not in- crease net prices, while measures such as the establishment of invest- ment and efficiency programs actually reduce net prices. These facts may be the result of politicians themselves carrying out the restructur- ing programs and emphasizing their political preferences when deciding what to invest in and what to do with existing infrastructure. It is disin- genuous to think that the government can satisfy the desires of the new owners better than they could themselves. In Mexico's case, a few changes to the privatization mechanism could have yielded large bene- fits: emphasizing speed, firing the chief executive officer before privati- zation, and refraining from costly restructuring measures would have increased net prices by 135 percent. A similar study by Chong and Galdo (2004) analyzes a cross-country sample of telecommunications firms that were privatized between 1985 and 2000; the authors' ordinary least squares (OLS) and instrumental variables (IV) regressions yield no evidence that streamlining before privatization is linked to higher net prices. Finally, evidence from Turkey also supports the conclusion that restructuring measures are either useless or counterproductive in raising net prices (Arin and Okten 2002). One of the most sensitive topics in the area of firm restructuring be- fore privatization is that of labor force retrenchment. To analyze the im- pact of such retrenchment policies beyond their effects on privatization prices, we construct (in an earlier paper, Chong and López-de-Silanes 2003) a worldwide privatization database containing detailed prepriva- tization firm and labor force characteristics, labor restructuring meas- ures undertaken by the government, and information on postprivatiza- tion labor rehiring policies, among other things. Table 1.4 shows that despite heavy unionization rates, most governments around the world downsize the labor force of state-owned enterprises before privatiza- tion. Labor retrenchment occurred in 78 percent of the sample, while THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 43 only 33 percent of all firms experienced voluntary downsizing pro- grams. Employment guarantees were established as part of privatization in 28 percent of the cases, while pay cuts before privatization were in- frequent (7.5 percent). Asia is the only region of the world with a sig- nificantly lower frequency of labor downsizing before privatization. Governments in Latin America deviated little from the general pattern; the only notable exception is the low frequency of employment guaran- tees, which was used in only 8 percent of all firms privatized in the re- gion. Table 1.4 also shows that state-owned enterprises in Latin Amer- ica were heavily unionized and active: two-thirds of state-owned enterprises privatized in the region experienced labor strikes in the three years before privatization. Following our earlier methodology, we ran OLS and instrumental vari- ables regressions for the 94 state-owned enterprises privatized in Latin America to test whether labor restructuring policies in this region trans- lated into higher net prices per dollar of firm sales. The first column of table 1.5 shows the OLS results, which suggest that labor downsizing be- fore privatization has a significant negative impact, equivalent to 28 per- cent of the average net price per dollar of sales. The instrumental variables results in column 2 show that once we control for endogeneity, the coeffi- cient drops essentially to zero and loses all significance.22 The results for Latin America reflect those for other regions: labor downsizing before pri- vatization is not priced by the buyers. From the point of view of increased government revenues, if a state-owned enterprise is overstaffed, it is prob- ably best for governments to wait and let the new owners make the deci- sions after they buy the firm. The other two regressions in table 1.5 focus on the effect of labor retrenchment in the form of voluntary downsizing programs in which governments offer monetary incentives for workers to quit. Even after controlling for endogeneity, voluntary downsizing leads to a marginally significant discount in the net price paid by private buyers. This negative effect might be explained by adverse selection, in that workers with the highest productivity or the best chances of finding alternative work are more likely to leave. Voluntary downsizing may therefore hurt firms, since it tends to result in the termination of valuable workers and the re- tention of less productive ones (Fallick 1996; Rama 1999). Despite the fact that voluntary separation programs are politically palatable, the findings here show that these programs may weaken firms and distort the composition of the work force, as predicted by theoretical models (Kahn 1985; Diwan 1994; Jeon and Laffont 1999). To shed further light on the "quality of firing" carried out by gov- ernments before privatization, we collected data on the hiring policies of state-owned enterprises after privatization (Chong and López-de-Silanes 2003). While hiring new workers probably responds to the legitimate business needs of privatized firms, rehiring previously fired workers 44 Table 1.4 Labor Restructuring before Privatization, by Region Latin Africa and Developed Transition Indicator America Asia Middle East countries economies All Sample size, unionization, and strikes before privatization Number of firms 101 24 64 77 42 308 Firms with unions before privatization 92.1 58.3 81.2 83.1 88.1 84.4 Firms with strikes before privatization 66.3 29.2 45.3 29.8 47.6 47.4 Type of restructuring measure before privatization Downsizing 82.2 58.3 79.7 79.2 76.2 78.2 Voluntary downsizing 32.5 12.5 45.3 28.6 14.3 32.5 Employment guarantee 8.4 20.1 51.6 13 52.4 28.2 Pay cut 8.9 0 1.6 13 7.1 7.5 Note: The table shows the number of firms included for each region, the regularity of unions and strikes, and the frequency of restructuring measures undertaken before privatization. The variables are defined as follows: (1) firms with unions before privatization is the percentage of privatized state-owned enterprises that had a union up to three years before privatization; (2) firms with strikes before privatization is of state- owned enterprises that suffered any kind of protest such as picketing or strikes during the three years before privatization; (3) downsizing is a dummy variable equal to 1 if the firm undertook any downsizing of the labor force up to three years before privatization, and 0 otherwise; downsizing may be classified as voluntary or compulsory, and may be neutral (no particular group targeted) or targeted according to age (age- biased downsizing), skills (skill-biased downsizing), or gender (female-biased downsizing); (4) voluntary downsizing is a dummy variable equal to 1 if the state-owned enterprise reduced its labor force in an exclusively noncoercive manner during the three years before privatization, and 0 otherwise; the most common methods of voluntary downsizing are incentive-based measures such as severance packages and pension enhancements; (5) employment guarantee is a dummy variable equal to 1 if the state-owned enterprise made any promise regarding the employment status of workers during the three years before privatization, and 0 otherwise; (6) pay cut is a dummy variable equal to 1 if there were any reductions in the salary or wage of workers during the three years before privatization, and 0 otherwise. Source: Chong and López-de-Silanes (2003). Table 1.5 Labor Restructuring and Privatization Prices in Latin America Dependent variable: net price OLS IV OLS IV Variable (1) (2) (3) (4) Firm and privatization characteristics Net total liabilities 0.0176 0.0168 0.0216 0.0153 (0.041) (0.043) (0.040) (0.042) Mining 0.3265*** 0.3406*** 0.293*** 0.3466*** (0.071) (0.067) (0.074) (0.061) Industry 0.2580*** 0.2711*** 0.2104*** 0.277*** (0.076) (0.074) (0.075) (0.065) Services 0.4106*** 0.4232*** 0.3565*** 0.4177*** (0.069) (0.066) (0.072) (0.057) Foreign 0.0561* 0.0737** 0.0666** 0.0856** (0.033) (0.036) (0.033) (0.038) Labor characteristics Unions 0.1592 0.1821 0.1878 0.1814 (0.131) (0.149) (0.122) (0.143) Labor policies Downsizing 0.1683*** 0.0201 (0.044) (0.027) Voluntary downsizing 0.1213*** 0.0558* (0.038) (0.032) 45 (Table continues on the following page.) 46 Table 1.5 (continued) Dependent variable: net price OLS IV OLS IV Variable (1) (2) (3) (4) Macroeconomic variable Gross domestic product 0.0673*** 0.0681* 0.0687*** 0.0713*** (0.010) (0.010) (0.010) (0.011) Constant 1.2120*** 1.3512* 1.2715*** 1.4746*** 0.334 0.341 0.311 0.350*** Observations 94 94 94 94 R2 0.47 0.38 0.53 0.41 F 11.36 10.32 11.59 12.35 Prob F 0.000 0.000 0.000 0.000 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: IV instrumental variables and OLS ordinary least square. The dependent variable is the net privatization price/sales, which is defined as the amount that accrues to the government from the sale of the state-owned enterprise after all privatization and restructuring costs are taken into account, adjusted by the percentage of company shares sold and divided by the average net sales of the state-owned enterprise during the three years before its privatization. The present value of the resulting number as of 2000 is used. The independent variables are defined as follows: (1) net total liabilities is a dummy variable equal to 1 if net total liabilities of the firm were greater than 0 up to three years before privatization, and 0 otherwise; (2) dummy variables for sectors (mining, industry, and services) are equal to 1 if the state-owned enterprise is part of that sector, and 0 otherwise; (3) foreign is a dummy variable equal to 1 if foreign firms were allowed to bid on the sale of the state-owned enterprise, and 0 otherwise; (4) unions is a dummy variable equal to 1 if the state-owned enterprise had a union up to three years before privatization, and 0 otherwise; (5) downsizing is a dummy variable equal to 1 if the firm undertook any downsizing of the labor force up to three years before privatization, and 0 otherwise; downsizing may be classified as voluntary or compulsory, and may be targeted according to age (age-biased downsizing), skills (skill-biased downsizing), or gender (female-biased downsizing) or may be neutral (no particular group targeted); (6) voluntary downsizing is a dummy variable equal to 1 if the state-owned enterprise reduced its labor force in an exclusively noncoercive manner during the three years before privatization, and 0 otherwise; the most common methods of voluntary downsizing are incentive-based measures such as severance packages and pension enhancements; (7) gross domestic product is the log of the average GDP in the country (in U.S. dollars at purchasing power parity) during the three years before privatization. All regressions include firm size controls. Columns 1 and 3 provide estimates from OLS regressions, while columns 2 and 4 show the second stage of the two-step instrumental variables procedure used in order to account for the endogenous nature of the labor downsizing variables. The instrumental variables approach is carried out according to the procedure outlined in Chong and López-de-Silanes (2004). Robust standard errors are given in parentheses. Source: Chong and López-de-Silanes 2003. 47 48 CHONG AND LÓPEZ-DE-SILANES Figure 1.14 Rehiring after Privatization, by Region Percentage 60 50 40 30 20 10 0 Latin America Asia Africa and Industrial Transition All Middle East countries economies Rehired Rehired same Note: Variables are defined as follows: rehired is a dummy variable equal to 1 if the privatized firm rehired previously fired workers up to 18 months after privatization, and 0 otherwise; rehired same is a dummy variable equal to 1 if the privatized firm rehired previously fired workers and placed them in the same department from which they were fired up to 18 months after privatization, and 0 otherwise. Previously fired workers are those who were terminated during the three years before privatization. Source: Chong and López-de-Silanes 2003. could mean that the downsizing programs before privatization went too far. After all, why else would a firm rehire a worker who was deemed expendable a relatively short time before? Figure 1.14 shows that close to 45 percent of all firms that underwent labor retrenchment programs in the three years before privatization hired back some of the fired work- ers after privatization. Across countries, only 10 percent of firms with government-run retrenchment programs ended up hiring back some of those workers to their previous positions within 18 months after priva- tization. Latin America is the region with the highest percentage of firms rehiring workers (53 percent) and rehiring to the same jobs that they had previously held (20 percent). THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 49 Table 1.6 analyzes the determinants of the probability that the privatized state-owned firm with labor retrenchment programs before privatization would hire new workers (new hires) or former workers previously fired by the government (rehires). Results show that the existence of a voluntary downsizing program before privatization does not predict a higher proba- bility of firms hiring new workers after privatization (column 1), but it in- creases by 34 percentage points the probability that the private buyer will rehire some of the workers who were previously fired by the government (column 2).23 The hiring behavior of firms in the postprivatization period says a great deal about the quality of the firing process and provides further evidence against the wisdom of government restructuring before privati- zation. Based on the evidence in this section, governments should think hard before restructuring the work force of state-owned enterprises in- tended for privatization. The political costs are high, the impact on net prices is low, and the firm could end up losing some of its most valuable employees. Type of Privatization Contract The type of privatization contract written is another potential area that may leave room for opportunistic behavior from politicians and private buyers. The simplest contracts are straightforward, outright sales of assets in which the government disconnects itself completely from the opera- tional future of the privatized firm. Other types of contracts may actually lead to a perverse relationship between the privatized firm and the state as managers and bureaucrats collude to serve their own interests at the expense of consumers and taxpayers. These contracts could take the form of the provision of services, the construction of infrastructure projects, or the establishment of joint ventures between private companies and the government. The common element in all of these cases is that the umbili- cal cord between the government and the firm has not been severed, leav- ing ample room for a complex set of problems. Shleifer and Vishny (1994) develop a theoretical model to help understand the incentives faced by firms in instances of partial privatization. When privatized firms depend significantly on the state, they may not restructure as expected because it is easier for them to extract rents from the government than to undergo painful reforms. At the same time, politicians have incentives to keep them afloat by subsidizing them and shielding them from competition. These arrangements persist because they are beneficial for both parties, although they reduce social welfare. As Bayliss (2002) points out, water privatiza- tion programs in Guinea and Côte d'Ivoire are examples of poor deals in which the private sector was able to make substantial profits controlling the distribution and fee collection of the service, while the government spent resources maintaining the infrastructure. 50 CHONG AND LÓPEZ-DE-SILANES Table 1.6 New Hires and Rehires in Privatized Firms in Latin America Dependent variable: Dependent variable: new hires rehires Probit Probit Indicator (1) dF/dX (2) dF/dX Voluntary 0.6035 [0.1600] 0.9004** [0.3370] downsizing (0.3835) (0.3826) Strikes 0.6026 [0.1408] 1.0382** [0.3961] (0.431) (0.423) Foreign 0.3092 [ 0.0852] 0.2469 [ 0.0943] participation (0.4074) (0.3879) Collective 0.2898 [ 0.0767] 0.8634* [ 0.3340] relations (0.4835) (0.5221) laws Constant 0.1453 10.1961** (4.2973) (4.3490) Observations 76 76 Log likelihood 29.49 33.99 Wald chi2 6.58 13.60 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: The dependent variable in the first regression is new hires, which is a dummy variable equal to 1 if the privatized firm hired new workers up to 18 months after privatization, and 0 otherwise; in the second regression, it is rehires, which is a dummy variable equal to 1 if the privatized firm rehired previously fired workers up to 18 months after privatization, and 0 otherwise. The independent variables are defined as follows: (1) voluntary downsizing is a dummy variable equal to 1 if the state-owned enterprise cut its labor force in an exclusively noncoercive manner during the three years before privatization, and 0 otherwise; the most common methods of voluntary downsizing are incentive-based measures such as severance packages and pension enhancements; (2) strikes is a dummy variable equal to 1 if there were any protests, picketing, or strikes up to three years before privatization, and 0 otherwise; (3) foreign participation is a dummy variable equal to 1 if foreign firms were allowed to bid for the state-owned enterprise, and 0 otherwise; (4) the collective relations laws index ranges from 0 to 3 and measures the level of protection granted to workers by labor and employment laws (higher values of the index represent more stringent laws regarding worker protection); it measures the areas of collective bargaining, worker participation in management, and collective disputes. All regressions include a partial privatization dummy, sectoral dummies, and country macroeconomic controls. Standard errors and marginal effects are given in parentheses and brackets, respectively. Source: Data collected by the authors; Chong and López-de-Silanes 2003; Botero and others 2005. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 51 To find a solution to the complications that these relationships gener- ate, Engel, Fischer, and Galetovic (1999, 2001) analyze the Chilean infra- structure concessions of the 1990s and note that franchising programs can provide a better alternative to the traditional approach of full state fi- nancing for infrastructure projects, particularly for governments that are financially and politically constrained. The regulatory framework, how- ever, must be effective if governments are to reap the potential benefits of franchising and avoid falling into hold-up problems in which firms under- bid to get the contracts but then threaten bankruptcy if a renegotiation is not granted. Guasch (2001) provides empirical evidence that renegotiations in con- cessions are fairly common. He analyzes more than 1,000 concessions granted in Latin American countries during the 1990s and finds that more than 60 percent of them were substantially renegotiated within three years. Infrastructure projects are usually very risky because of the diffi- culty inherent in forecasting demand. Firms therefore press for income guarantees and other explicit or implicit insurance mechanisms that end up costing the government too much. It may occasionally be in a country's best interest to give out these guarantees, but they should be explicit and transparent, and they should ideally be made in exchange for a fee (Engel, Fischer, and Galetovic 2003). In all of these situations, the solution should also include very clear dis- closure and monitoring mechanisms to avoid related-party transactions at unfair terms. Such transactions may end up bankrupting the joint venture or the asset that the government has an interest in keeping afloat to the benefit of the private corporation, as happened in the case of highways and commercial banks in Chile and Mexico (Ramírez 1998; Johnson and others 2000; La Porta, López-de-Silanes, and Zamarripa 2003). These are not easy issues to solve, and many of the failures of privatization can be linked to perverse incentives provided by misguided privatization conces- sion contracts. The evidence in this section can be understood from a political econ- omy perspective. Privatization involves politicians with incentives and objectives. Therefore, the design of the privatization process, the contracts ultimately written, the restrictions attached to the sale of state-owned en- terprises, and the restructuring measures adopted before privatization should be understood as opportunities for politicians to extract rents and hand out favors. This perspective helps rationalize instances in which cor- ruption in privatization leads to disastrous results. The policy lesson is clear: a transparent and expeditious privatization process leaves less room for corruption and collusion among politicians and businessmen who may try to benefit from opaqueness. One must also consider the time needed to set up an effective privatizing agency and build the regulatory framework that should be in place before state-owned enterprises with market power are sold. We turn to this topic in the next section. 52 CHONG AND LÓPEZ-DE-SILANES Complementary Policies: Reregulation and Corporate Governance The previous section analyzed some of the main privatization failures emerging from policies or decisions taken before or at the time of privati- zation. In this section, we turn to the impact of the regulatory and institu- tional framework after privatization. Privatization should not be looked at in isolation. Its success is likely to depend on at least two sets of comple- mentary policies. The first is deregulation and reregulation of sectors with market power or in which government ownership represented a substan- tial percentage of total assets before privatization. The second is the establishment of a set of institutions that promote good corporate gover- nance, which facilitates access to capital and allows recently privatized firms to finance their growth without dependence on the state. Many pri- vatization failures can be explained by a lack of careful consideration of these two complementary sets of policies. Privatization, Reregulation, and Deregulation An appropriate regulatory framework after privatization is a key compo- nent of the success or failure of the program, particularly in utilities and services. A common element across many failed examples of privatization is inadequate regulation leading to suboptimal levels of competition or sit- uations in which producers are allowed to keep the gains from privatiza- tion without sharing them with consumers (Boubakri and Cosset 1999; Megginson and Netter 2001). The classic position of critics is to turn this into an argument against further privatization. However, the ample em- pirical evidence surveyed here shows that privatization can be done cor- rectly and can lead to social gains. This evidence should be enough to dis- card a simplistic interpretation of cases of failures. Regulation should be carefully revised in conjunction with privatiza- tion in two prominent situations: industries characterized as natural monopolies or by the presence of oligopolistic markets, and industries in which the government owns most of the assets in the industry even if no individual firm has substantial market power. Sectors with heavy state presence tend to be protected by a web of regulations originally instituted to cut the losses of state-owned firms and reduce fiscal deficits. In some of these cases, the necessary regulatory effort can be best understood as deregulation to eliminate protective structures that shield companies from competition and allow privatized firms to make extraordinary gains at the cost of consumers. As explained in both the early and more recent litera- ture, competition and deregulation should be carefully considered in pri- vatization (Yarrow 1986; Allen and Gale 2000). Winston (1993) argues that deregulation has the power to produce efficiency improvements, THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 53 which can benefit consumers and producers. There is no reason to believe that deregulation should lead to different outcomes in the case of privati- zation of overprotected industries.24 In sectors with oligopolistic power, the deregulation effort needs to be complemented by a reregulation that clearly establishes a new package of rules and disclosures that will enhance supervision and reduce abuse of market power. Reregulation of oligopolistic sectors is complicated because of weak- nesses in regulatory governance. As Fischer and Serra (2002) explain, regulators are often subject to pressures from populist politicians and in- dustry lobbyists, and their low salaries make them susceptible to cap- ture. Moreover, regulatory systems often operate within the context of an inefficient and perhaps even corrupt judicial system. Deregulation complements privatization in two ways (La Porta and López-de-Silanes 1999). First, product market competition provides a tool for weeding out the least efficient firms. This process may take too long--or not work at all--if regulation inhibits new entry or makes exit costly. Wallsten (2001) undertakes an econometric analysis of the effects of telecommunications privatization and regulation in a panel of 30 countries in Latin America and Africa. His results show that competition from mobile operators and privatization combined with the existence of a separate regulator are significantly associated with increases in labor efficiency, mainlines per capita, and connection capacity. A casual inter- pretation of his results suggests that privatization of oligopolistic indus- tries without concurrent reforms may not necessarily improve welfare. Second, deregulation may also complement privatization by raising the cost of political intervention. Whereas an inefficient monopoly can squander its rents without endangering its existence, an inefficient firm in a competitive industry would have to receive a subsidy to stay afloat. The introduction of competition forces politicians to pay firms directly to en- gage in politically motivated actions, whereas previously the costs of these measures were absorbed by a state-owned firm that did not have to worry about market performance. In fact, competition is often restricted precisely because it raises the costs of political influence. Colombia and Mexico provide good examples of deregulatory policy actions that, when coupled with privatization, can be used as a lever to transform the eco- nomic landscape and reduce political interference in the economy. In the early 1990s Colombia began an economic openness program through the promotion of market competition and deregulation. As Pombo and Ramírez describe in chapter 6, privatization was perceived as an instru- ment for economic deregulation and the promotion of market competi- tion. A decade earlier Mexico started to transform its previously closed economy characterized by capital controls, price regulation, restrictions on foreign direct investment, high tariffs, import quotas, and a large state-owned public sector. As in the case of Colombia, privatization cou- pled with deregulation played a key role in the drive to restructure the 54 CHONG AND LÓPEZ-DE-SILANES economy and help privatized state-owned enterprises catch up to their private peers (La Porta and López-de-Silanes 1999). Generally speaking, reregulation or deregulation can take place at three different moments: before privatization, at the time of privatization, or af- ter the state-owned enterprise has been sold. The literature emphasizes the importance of having efficient regulation at an early stage. Reregulation or deregulation before privatization of the industry may increase the pace of divestiture and help sell companies at a higher price if it reduces regulatory risk.25 Wallsten (2002) finds that countries that established a separate reg- ulatory authority in telecommunications before privatization not only benefited from increased telecommunications investment and telephone penetration but also gained from investors' willingness to pay more for the telecommunications firms.26 Establishing effective preprivatization regu- lation is not easy, however, for at least three reasons. First, changes to the regulatory regime before privatization are likely to lower the profits of state-owned enterprises, which translates into higher financial needs for the government at a very difficult time. Second, the political will for a true regulatory reform might not materialize without the pressure of imminent privatization. Finally, governments with little experience in privatization often find it difficult to carry out an effective preprivatization regulatory reform. Deregulation and reregulation at the time of privatization solve the first two problems and reduce regulatory risk discounts. As long as a suit- able regulatory framework is in place at or before the time of privatiza- tion, consumers and the government should benefit from the process. Chisari, Estache, and Romero (1999) use a computable general equilib- rium model for Argentina to show that the gains from efficient regulation are nontrivial. Their model estimates the gains from the private operation of utilities at about 0.9 percent of GDP and those of effective regulation at an additional 0.35 percent of GDP. Moreover, the distribution of the gains across income classes is driven by the effectiveness of the regulators. In short, they claim that clear reregulation is good for the poor. Lack of regulatory capabilities at the time of privatization, coupled with a desire to maximize price at the time of the sale, has led several gov- ernments to postpone full and clear reregulation. Establishing an ade- quate regulatory scheme after privatization, however, may be problem- atic from a political economy perspective. Since the agency in charge of enforcing and regulating the contracts is often the same as or subordi- nated to the agency that carried out the privatization, the people involved have an incentive to implement lax enforcement to avoid exposing past mistakes. Chong and Sánchez (2003) document that for a broad number of concessions in infrastructure projects, the private sector was able to bargain and keep protective regulation after privatization because of the threat of bankruptcy, withdrawal, or desertion of future investment com- mitments. All of these affect the reputation and credibility of privatizing THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 55 politicians. According to Guasch (2001), concession contracts in devel- oping countries often led to renegotiations over the last 15 years. In Latin America and the Caribbean, 40 percent of all concession contracts were renegotiated just over 2.2 years after they were signed. Engel, Fischer, and Galetovic (2003) argue that opportunistic renegotiations of concessions are common because of a "privatize now, regulate later" approach. Cost overruns in concessions and unclear rules governing contingencies pro- vide private owners with the opportunity to extract economic rents from the government. Finally, attempting to substantially alter the regulatory framework after the sale is further complicated by the fact that new con- stituencies against reregulation are created at the time of privatization. Shareholders and managers of privatized state-owned enterprises are joined by workers and even consumers who could benefit from the pro- tective regulatory status of firms. The political economy approach explains why it is hard to bring about changes in regulation after privatization and why privatized firms are frequently able to renegotiate their contracts on more favorable terms. It is therefore advisable to push for changes in the regulatory framework at the time of privatization or earlier, if possible. Perfecting the new regulatory framework may take a lot of time, however, and this should not be used as an excuse for postponing the privatization of money-losing entities. Privatization and Corporate Governance The last issue we address in this paper is the connection between the suc- cess of privatization and the establishment of an institutional framework that promotes good corporate governance. The absence of this framework increases the cost of capital and thus prevents privatized firms from un- dertaking the investments needed to operate in a more competitive envi- ronment. Access to alternative sources of finance at a low cost allows firms to survive and grow without state help. The development and appropriate functioning of stock and credit mar- kets need a solid regulatory framework that promotes investor protection and disclosure. Recent research shows a strong link between a firm's access to capital and efficiently enforced laws (La Porta and others 1997, 1998, 2000b, 2002; La Porta, López-de-Silanes, and Shleifer 2003). In countries where large numbers of firms have been sent out to the private market and deregulation has increased competition and lowered trade barriers, there is an urgent need for institutions that can efficiently chan- nel resources to the new private sector. The old laws and institutions might have been efficient in covering the needs of state-owned enterprises, but private enterprises and privatized firms require different services and stand to benefit from the development of deep stock and credit markets. Ariyo and Jerome (1999) argue that the absence of developed capital markets 56 CHONG AND LÓPEZ-DE-SILANES and the lack of appropriate legal and judicial structures have hindered the success of privatization in Africa. Before privatization, government banks are typically used as a source of financing. Yet in most privatization programs, the banking sector is one of those turned over to private hands. If financing for privatized state-owned enterprises is expected to come from privatized banks--or from any other private credit institution--then creditor rights, embedded in bankruptcy laws, and the efficiency of courts must be strengthened and streamlined. Without proper bankruptcy procedures that allow for the expedient recov- ery of assets, financial institutions will be reluctant to lend for fear of po- tential losses, and they may end up failing to satisfy the financial needs of the private sector. The banking system itself is rendered more vulnerable to crises without effective creditor rights, since it loses its ability to repossess collateral expediently (La Porta, López-de-Silanes, and Zamarripa 2003). The development of large stock markets where firms can access long- term funds is also an important complementary measure to privatization. In some cases governments have provided a boost to stock markets by pri- vatizing state-owned enterprises through initial public offerings. This is not enough, however, to ensure the development of the market and its usefulness as a source of future financing for these firms. Privatization without a commitment to improve shareholder rights in corporate and se- curities laws will probably lead to widespread abuse and appropriation of benefits by managers or those in control, with only small gains for mi- nority investors in the form of dividends, for example (La Porta and oth- ers 2000a; López-de-Silanes 2002; La Porta, López-de-Silanes, and Shleifer 2003). The failure to institute appropriate securities laws and ef- fective enforcement may be responsible for many of the scandals that are now blamed on privatization in countries such as the Czech Republic (Dyck 2001; Glaeser, Johnson, and Shleifer 2001). An additional benefit of corporate governance reform is that the improvement in disclosures and accounting standards facilitates the work of regulators. As Carey and others (1994) and Campos-Méndez, Trujillo, and Estache (2001) argue, postprivatization regulators end up relying on standard accounting data instead of imposing specific regulatory accounting needs. If this is the case, enhanced accounting standards should be of great benefit to regula- tors of privatized firms, particularly in the area of disclosure of related- party transactions and conflicts of interest. The reform of corporate governance institutions through the establish- ment and enforcement of effective securities, corporate, and bankruptcy laws should become an essential complementary policy to prevent expro- priation by controlling investors and to promote the development of sta- ble sources of funds to which privatized firms can turn to finance their growth. Bear in mind that many financially troubled private firms became state owned in the last 50 years when limited access to capital pushed them to seek government financing (López-de-Silanes 1994). THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 57 Conclusions The push for privatization and the drive to restructure the role of the state in production have lost their appeal. A large political backlash to privatiza- tion has been brewing for some time, and public opinion and policymakers in Latin America and other regions of the world have now turned against privatization. The goal of this chapter is to help set the privatization record straight by analyzing systematic evidence emerging from comprehensive studies around the world. In this quest, we benefit from a recent series of ac- ademic papers focusing on the Latin American experience. Given their ex- tensive coverage and systematic econometric approach, these papers are able to address the series of concerns voiced against privatization. The evidence lines up: countries that privatize benefit, and the gains not only are kept by firm owners--they are also distributed to society. These findings do not mean that failures do not occur, but rather that they are not the norm. Most instances of failure can be explained by three factors. First, opaque processes with heavy state involvement open the door to corruption and opportunistic behavior. Second, poor contract de- sign and regulatory capture are linked to a lack of deregulation and in- adequate reregulation. Third, deficient corporate governance institutions raise the cost of capital, hamper restructuring efforts, and may throw firms back into the hands of the state. The understanding of the political economy mechanisms behind the causes of failure should be used to im- prove privatization, not to stop it. Notes The authors gratefully acknowledge comments from Eduardo Bitrán, Eduardo Engel, Luis Felipe López-Calva, John Nellis, Máximo Torero, and Andrés Velasco. They would also like to thank Patricio Amador, José Caballero, Cecilia Calderón, Virgilio Galdo, Magdalena S. López-Morton, Alejandro Ponce, and Alejandro Riaño for outstanding research assistance. 1. On the poor performance of state-owned enterprises, see Boardman and Vining (1989) and Mueller (1989); on improvements after privatization, see Meg- ginson, Nash, and van Randenborgh (1994); Ehrlich and others (1994); Frydman and others (1999); La Porta and López-de-Silanes (1999); Dewenter and Malatesta (2001); Megginson and Netter (2001); Sheshinski and López-Calva (2003); and Chong and López-de-Silanes (2004). 2. See Bayliss (2002) and Birdsall and Nellis (2002) for recent cross-country reviews of privatization failures. Criticism about specific countries or industries includes Coes (1998), Nellis (1999), Harper (2000), Wallsten (2001), and Stiglitz (2002). 3. Polls show that privatization is becoming less popular even in the United Kingdom, which led the privatization effort in the 1980s. In 1983 around 43 percent of people wanted more privatization, but that number was down to 24 58 CHONG AND LÓPEZ-DE-SILANES percent by 1992, and it barely reached 19 percent in 2002 ("The End of Priva- tization," The Economist, June 13, 1998, 53­54). 4. Earle and Gehlbach (2003) provide a framework that rationalizes why pol- icymakers may pay too much attention to public sentiment and thus refrain from potentially welfare-improving actions. 5. All dollar amounts are in U.S. dollars unless otherwise specified. 6. Recent research shows that the privatization effort in Africa may have been highly underestimated. Bennell (1997) argues that most papers studying privatization in Africa are based on low-quality or outdated samples. Based on a comprehensive survey of privatization transactions that spans 16 years (1980­95) and includes over 2,000 privatizations, he concludes that African privatization programs are larger than previously thought and that they increased substantially in the 1990s. 7. "State-Owned Stockpiles," The Economist, March 31, 2001, 58­59. 8. The analysis in this paper covers only the privatization experience at the country or federal level--that is, assets sold by the central or federal government-- which accounts for the majority of assets sold around the world so far. A different sample and experience is that of the privatization of services at the local, municipal, or county level, where local governments "privatize" the public provision of services. These programs have taken place in only a few nations, such as the United States (López-de-Silanes, Shleifer, and Vishny 1997) and England, where public service pro- vision by the private sector has become a central issue. 9. Differences in accounting procedures may also be problematic in determin- ing adequate measures of operating performance (Megginson and Netter 2001). 10. Financial firms privatized in Mexico are analyzed in a separate paper (López- de-Silanes and Zamarripa 1995). 11. Sheshinski and López-Calva (2003) make similar claims after they analyze privatization programs and remaining state-owned assets around the world. 12. The role of the Institute for Industrial Promotion in creating new manufac- turing enterprises was central during the 1950s and 1960s. The largest private capi- tal enterprises in the steel, chemical, paper, fertilizer, metalworking, and automobile sectors today were companies formerly associated with the institute. 13. Not surprisingly, chapter 3 represents the first formal empirical study of the impact of privatization on firm performance in Bolivia. 14. For example, Pombo and Ramírez used the first method in Colombia, whereas Anuatti-Neto and and his coauthors (Brazil) and Fischer, Serra, and Gutiér- rez (Chile) employed both methods. 15. Comprehensive privatization studies for eastern European countries also find higher profitability results, although the accounting data for such countries are more problematic. Some examples are Claessens, Djankov, and Pohl (1997) for the Czech Republic; Dyck (1997) for East Germany; and Frydman and others (1999) for the Czech Republic, Hungary, and Poland. For most of these cases, accounting differences before and after privatization are of greater concern than in Latin America, where the state-owned enterprises filed and collected information similar to that of private firms. 16. Data for Colombia are from the energy sector. 17. To isolate the contribution of changes in relative prices as a factor behind the observed profitability gains, the calculation compares the observed percentage point increase in operating income to sales with what would have taken place if privatized firms had increased output but left real prices unchanged at preprivatization levels. Specifically, the formula used for the price contribution is SALESap COSTap S [SALESap ^ Q 1 R T COSTap PRICECONTRIB , SALESap SALESap^(1 ) THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 59 where SALESap represents sales in the postprivatization period, COSTap repre- sents operating costs in the postprivatization period, and is the increase in real prices. 18. Trujillo and others (2002) provide evidence for 21 Latin American countries between 1985 and 1998 and find that private sector involvement in utilities and transportation yielded marginally positive results on per capita GDP. 19. Birdsall and Nellis (2002) indicate that these results are less valid for Latin America than for transition economies and less relevant for utilities than for banks or oil. 20. The net price in these studies is defined as the net privatization price (after the costs of privatization and restructuring are deducted) divided by the dollar value of the firms' assets. The benefit of focusing on this measure is that it provides a useful framework for comparing across firms and gives a benchmark against which to think about the relative price of other privatization goals pursued by the government. Privatization programs are typically designed with the aim of pursu- ing revenue generation, to get out of a fiscal crisis, or to serve redistributive pur- poses. For Brazil, Colombia, Mexico, and Peru, the price paid was a crucial moti- vation in selecting winners for almost all privatized state-owned enterprises (see López-de-Silanes 1997, Torero 2002, and the chapters in this volume on Brazil and Colombia). Furthermore, economists generally endorse the goal of maximizing rev- enues. Bolton and Roland (1992) show that a policy of maximizing net sales rev- enue is likely to be consistent with a policy of maximizing social welfare since the proceeds from the sale can be used to subsidize employment, investment, a social safety net, and other public goods. 21. López-de-Silanes 1997; Arin and Okten 2002. The evidence for the case of Turkey should be regarded as tentative since the lack of data has thus far prevented a robust instrumental variables analysis for this country. 22. We apply a two-step instrumental variables approach by estimating a non- linear reduced-form equation that describes the probability that a particular labor re- structuring policy will be implemented. The instruments used are classified in two groups: firm-level and macroeconomic-level determinants. The firm-level variables included the presence of a leading agent bank, involvement of a ministry before pri- vatization, the political affiliation of unions, and sectoral dummies. The macroeco- nomic variables include the average GDP growth rate and the degree of openness in the three years before privatization, as well as the legal origin of the country. None of these variables are statistically significant when included in the price equation. The F statistic for the excluded instruments is statistically significant at 1 percent in all cases. 23. Regressions control for labor rigidities coming from the collective relations laws from each country, as the incidence of rehires after privatization could also re- flect the firing costs and rigidities of the labor market. 24. For the case of Mexico, La Porta and López-de-Silanes (1999) find that deregulation--particularly the removal of price or quantity controls and trade barri- ers--is linked to faster convergence of privatized firms to industry benchmarks. 25. See Bortolotti, Fantini, and Siniscalco (2001) for the case of the electricity sector. 26. Chong and Galdo (2004) find similar results. References Akram, Tanweer. 2000. "Publicly Subsidized Privatization: A Simple Model of Dysfunctional Privatization." Applied Economics 32 (13): 1689­99. 60 CHONG AND LÓPEZ-DE-SILANES Allen, Franklin, and Douglas Gale. 2000. "Corporate Governance and Competi- tion." In Xavier Vives, ed., Corporate Governance: Theoretical and Empirical Perspectives. Cambridge, U.K.: Cambridge University Press. Altman, Edward. 1984. "A Further Empirical Investigation of the Bankruptcy Cost Question." Journal of Finance 39 (4): 1067­89. Arin, Kermin, and Cagla Okten. 2002 "The Determinants of Privatization Prices: Evidence from Turkey." Working Paper. Louisiana State University, Baton Rouge, La. Ariyo, Ademola, and Afeikhena Jerome. 1999. "Privatization in Africa: An Ap- praisal." World Development 27 (1): 201­13. Barja, Gover, David McKenzie, and Miguel Urquiola. 2002. "Capitalization and Privatization in Bolivia." Cornell University, Ithaca, N.Y. Bayliss, Kate. 2001. "Privatization of Electricity Distribution: Some Economic, So- cial and Political Perspectives." Report 2001-04-E-Distrib. University of Green- wich, Public Services International Research Unit, Greenwich, U.K. ------. 2002. "Privatization and Poverty: The Distributional Impact of Utility Pri- vatization." Annals of Public and Cooperative Economics 73 (4): 603­25. Bayliss, Kate, and David Hall. 2000. "Privatization of Water and Energy in Africa." Report 2000-09-U-Afr. University of Greenwich, Public Services Inter- national Research Unit, Greenwich, U.K. Bennell, Paul. 1997. "Privatization in Sub-Saharan Africa: Progress and Prospects during the 1990s." World Development 25 (11): 1785­803. Biais, Bruno, and Enrico Perotti. 2002. "Machiavellian Privatization." American Economic Review 92 (1): 240­58. Birch, Melissa, and Jerry Haar, eds. 2000. The Impact of Privatization in the Amer- icas. Boulder, Colo.: North-South Center Press. Birdsall, Nancy, and John Nellis. 2002. "Winners and Losers: Assessing the Dis- tributional Impact of Privatization." Working Paper 6. Center for Global De- velopment, Washington, D.C. Boardman, Anthony, and Aidan R. Vining. 1989. "Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises." Journal of Law and Economics 32 (1): 1­33. Bolton, Patrick, and Gerald Roland. 1992. "Privatization Policies in Central and Eastern Europe." Economic Policy 6 (15): 275­303. Bortolotti, Bernardo, Marcella Fantini, and Carlo Scarpa. 2001. "Privatization: Politics, Institutions and Financial Markets." Emerging Markets Review 2 (2): 109­36. Bortolotti, Bernardo, Marcella Fantini, and Domenico Siniscalco. 2001. "Regula- tion and Privatization: The Case of Electricity." In Giuliano Amato and Laraine L. Laudati, eds., The Anti-Competitive Impact of Regulation. Northampton, U.K.: Elgar Press. Botero, Juan, and others. 2005. "The Regulation of Labor." Quarterly Journal of Economics, forthcoming. Boubakri, Narjess, and Jean Claude Cosset. 1998. "The Financial and Operating Performance of Newly Privatized Firms: Evidence from Developing Countries." Journal of Finance 53 (3): 1081­110. ------. 1999. "Does Privatization Meet the Expectations? Evidence from African Countries." Working Paper. École des Hautes Etudes Commerciales, Montreal. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 61 Campbell-White, Oliver, and Anita Bhatia. 1998. "Privatization in Africa." Direc- tions in Development Report 17972. World Bank, Washington, D.C. Campos-Méndez, Javier, Lourdes Trujillo, and Antonio Estache. 2001. "Processes, Information and Accounting Gaps in the Regulation of Argentina's Private Railways." Working Paper 2636. World Bank, Washington, D.C. Carbajo, José, and Antonio Estache. 1996. "Competing Private Ports: Lessons from Argentina." Viewpoint Note 100. World Bank, Washington, D.C. Carey, Alan, and others. 1994. "Accounting for Regulation in U.K. Utilities." Lon- don: Institute of Chartered Accountants of England and Wales. Caves, Richard. 1990. "The Concentration-Margins Relationship Reconsidered: Comments." Brookings Papers on Economic Activity: Microeconomics 322­24. Chisari, Omar, Antonio Estache, and Carlos Romero. 1999. "Winners and Losers from the Privatization and Regulation of Utilities: Lessons from a Gen- eral Equilibrium Model of Argentina." World Bank Economic Review 13 (2): 357­78. Chong, Alberto, and Virgilio Galdo. 2004. "Streamlining and Privatization Prices in the Telecommunications Industry." Economica. Chong, Alberto, and Florencio López-de-Silanes. 2003. "Privatization and Labor Restructuring around the World." Yale University, School of Management, New Haven, Conn. ------. 2004. "Privatization in Latin America: What Does the Evidence Say?" Economía 4(2): 37­94. Chong, Alberto, and Alejandro Riaño. 2003. "Institutions and Privatization Prices." Inter-American Development Bank, Washington, D.C. Chong, Alberto, and Jose Miguel Sánchez, eds. 2003. Medios privados para fines publicos: arreglos contractuales y participacion privada en infraestruc- tura en America Latina. Washington, D.C.: Inter-American Development Bank. Claessens, Stjin, Simeon Djankov, and Gerhard Pohl. 1997. "Ownership and Cor- porate Governance: Evidence from the Czech Republic." Policy Research Paper 1737. World Bank, Washington, D.C. Clarke, George, and Robert Cull. 1999. "Bank Privatization in Argentina: A Model of Political Constraints and Differential Outcomes." Working Paper 2636. World Bank, Washington, D.C. Coes, Donald. 1998. "Beyond Privatization: Getting the Rules Right in Latin America's Regulatory Environment." Quarterly Review of Economics and Finance 38 (3): 525­32. Crampes, Claude, and Antonio Estache. 1996. "Regulating Water Concessions: Lessons from the Buenos Aires Concession." Viewpoint Note 91. World Bank, Washington, D.C. D'Souza, Juliet, and William L. Megginson. 1999. "The Financial and Operating Performance of Newly Privatized Firms in the 1990s." Journal of Finance 54 (4): 1397­438. Dewenter, Kathryn L., and Paul H. Malatesta. 2001. "State-Owned and Privately- Owned Firms: An Empirical Analysis of Profitability, Leverage and Labor In- tensity." American Economic Review 91 (1): 320­34. Diwan, Ishac. 1994. "Public Sector Retrenchment and Severance Pay: Nine Propo- sitions." In Shahid Chaudhry, Gary Reid, and Waleed Malik, eds., Civil Service 62 CHONG AND LÓPEZ-DE-SILANES Reform in Latin America and the Caribbean: Proceedings of a Conference. Technical Paper 259. Washington, D.C.: World Bank. Dyck, Alexander. 1997. "Privatization in Eastern Germany: Management Selec- tion and Economic Transition." American Economic Review 87 (4): 565­97. ------. 2001. "Privatization and Corporate Governance: Principles, Evidence and Future Challenges." World Bank Research Observer 16 (1): 59­84. Earle, John, and Scott Gehlbach. 2003. "A Spoonful of Sugar: Privatization and Popular Support for Reform in the Czech Republic." Economics and Politics 15 (1): 1­32. Ehrlich, Isaac, and others. 1994. "Productivity Growth and Firm Ownership: An Empirical Investigation." Journal of Political Economy 102 (5): 1006­38. Engel, Eduardo, Ronald Fischer, and Alexander Galetovic. 1999. "The Chilean Infrastructure Concessions Program: Evaluation, Lessons, and Prospects for the Future." Working Paper 60. Centro de Economía Aplicada, Santiago, Chile. ------. 2001. "Least-Present-Value-of-Revenue Auctions and Highway Franchis- ing." Journal of Political Economy 109 (5): 993­1020. ------. 2003. "Privatizing Highways in Latin America: Fixing What Went Wrong." Economía 4 (1): 129­64. Estache, Antonio, and Martín Rodríguez. 1996. "Regulatory Lessons from Ar- gentina's Power Concessions." Viewpoint Note 99. World Bank, Washington, D.C. Fallick, Bruce. 1996. "A Review of the Recent Empirical Literature on Displaced Workers," Industrial and Labor Relations Review 50 (1): 5­16. Fischer, Ronald, and Pablo Serra. 2002. "Regulating the Electricity Sector in Latin America." Economía 1 (1): 155­98. Freije, Samuel, and Luis Rivas. 2002. "Privatization, Inequality and Welfare: Evi- dence from Nicaragua." Instituto de Estudios Superiores de Administración, Centro de Desarrollo Humano y Organizaciones, Caracas, Venezuela. Frydman, Roman, and others. 1999. "When Does Privatization Work? The Impact of Private Ownership on Corporate Performance in Transition Economies." Quarterly Journal of Economics 114 (4): 1153­91. Galal, Ahmed, Leroy Jones, Pankay Tandon, and Ongo Vogelsang. 1994. Welfare Consequences of Selling Public Enterprises. Oxford, U.K.: Oxford University Press. Glaeser, Edward, Simon Johnson, and Andrei Shleifer. 2001. "Coase versus the Coasians." Quarterly Journal of Economics 116 (3): 853­99. Guasch, José Luis. 2001. "Concessions and Regulatory Design: Determinants of Performance--Fifteen Years of Evidence." Policy Research Paper. World Bank, Washington, D.C. Hachette, Dominique, and Rolf J. Lüders. 1994. Privatization in Chile: An Eco- nomic Appraisal. San Francisco, Calif.: ICS (International Center for Economic Growth) Press. Harper, Joel T. 2000. "The Performance of Privatized Firms in the Czech Repub- lic." Working Paper. Florida Atlantic University, Boca Raton, Fla. Jeon, Doh-Shin, and Jean-Jacques Laffont. 1999. "The Efficient Mechanism for Downsizing the Public Sector." World Bank Economic Review 13 (1): 67­88. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 63 Johnson, Simon, and others. 2000. "Tunneling." American Economic Review 90 (May, Papers and Proceedings, 1999): 22­27. Reprinted in Klaus J. Hopt, ed., Capital Markets and Company Law. Oxford, U.K.: Oxford University Press, 2002. Kahn, Charles. 1985. "Optimal Severance Pay with Incomplete Information." Journal of Political Economy 93 (3): 435­51. Kikeri, Sunita. 1999. "Privatization and Labor: What Happens to Workers When Governments Divest?" Technical Paper 396. World Bank, Washington, D.C. Kikeri, Sunita, John Nellis, and Mary Shirley. 1992. "Privatization: The Lessons of Experience." Policy Views. World Bank, Country Economics Department, Washington, D.C. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." Quarterly Journal of Economics 114 (4): 1193­242. La Porta, Rafael, Florencio López-de-Silanes, and Andrei Shleifer. 2002. "Govern- ment Ownership of Banks." Journal of Finance 57 (1): 265­302. ------. 2003. "What Works in Securities Laws?" Working Paper 9882. National Bureau of Economic Research, Cambridge, Mass. La Porta, Rafael, Florencio López-de-Silanes, and Guillermo Zamarripa. 2003. "Related Lending." Quarterly Journal of Economics 118 (1): 231­68. La Porta, Rafael, Florencio López-de-Silanes, Andrei Shleifer, and Robert Vishny. 1997. "Legal Determinants of External Finance." Journal of Finance 52 (3): 1131­50. ------. 1998. "Law and Finance." Journal of Political Economy 106 (6): 1113­55. ------. 2000a. "Agency Problems and Dividend Policies around the World." Journal of Finance 55 (1): 1­33. ------. 2000b. "Investor Protection and Corporate Governance." Journal of Financial Economics 58 (1): 3­27. ------. 2002. "Investor Protection and Corporate Valuation." Journal of Finance 57 (3): 1147­70. López-Calva, Luis F. 2004 "Comments to Privatization in Latin America: What Does the Evidence Say?" Economía 4(2): 95­111. López-de-Silanes, Florencio. 1994. "A Macro Perspective on Privatization: The Mexican Program." In Santiago Levy and Lars E. O. Svensson, eds., Macro- economic Aspects of Privatization. Washington, D.C.: World Bank. ------. 1997. "Determinants of Privatization Prices." Quarterly Journal of Eco- nomics 107 (4): 965­1025. ------. 2002. "The Politics of Legal Reform." Economía 2 (2): 91­152. López-de-Silanes, Florencio, and Guillermo Zamarripa. 1995. "Deregulation and Privatization of Commercial Banking." Revista de Análisis Económico 10 (2): 113­64. López-de-Silanes, Florencio, Andrei Shleifer, and Robert W. Vishny. 1997. "Priva- tization in the United States." Rand Journal of Economics 28 (3): 447­71. Lora, Eduardo. 2001. "Structural Reforms in Latin America: What Has Been Re- formed and How to Measure It." Working Paper W-466, Inter-American De- velopment Bank, Research Department, Washington, D.C. Lüders, Rolf J. 1991. "Chile's Massive SOE Divestiture Program, 1975­1990: Fail- ures and Successes." Contemporary Policy Issues 9 (4): 1­19. 64 CHONG AND LÓPEZ-DE-SILANES McKenzie, David, and Dilip Mookherjee. 2003. "The Distributive Impact of Pri- vatization in Latin America: Evidence from Four Countries." Economía 3 (2): 161­218. Megginson, William, and Jeffry Netter. 2001. "From State to Market: A Survey of Empirical Studies on Privatization." Journal of Economic Literature 39 (2): 321­89. Megginson, William, Robert Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An Interna- tional Empirical Analysis." Journal of Finance 49 (2): 403­52. Mueller, Denis G. 1989. Public Choice. Cambridge, U.K.: Cambridge University Press. Nellis, John. 1999. "Time to Rethink Privatization in Transition Economies?" Dis- cussion Paper 38 International Finance Corporation, Washington, D.C. Nellis, John, and Sunita Kikeri. 1989. "Public Enterprise Reform: Privatization and the World Bank." World Development 17 (5): 659­72. OECD (Organisation for Economic Cooperation and Development). 2001. "Re- cent Privatization Trends in OECD Countries." Paris. Perotti, Enrico. 1995. "Credible Privatization." American Economic Review 85 (4): 847­59. Petrazzini, Ben, and Theodore H. Clark. 1996. "Costs and Benefits of Telecom- munications Liberalization in Developing Countries." Working Paper. Hong Kong University of Science and Technology, Hong Kong, China. Pinheiro, Armando C. 1996. "Impactos microeconômicos da privatização no Brasil." Pesquisa e Planejamento Econômico 26 (3): 357­97. Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Re- forms in Eastern Europe and Latin America. Cambridge, U.K.: Cambridge Uni- versity Press. Rama, Martin. 1999. "Efficient Public Sector Downsizing." World Bank Eco- nomic Review 13 (1): 1­22. Ramamurti, Ravi. 1996. Privatizing Monopolies: Lessons from the Telecommuni- cations and Transport Sectors in Latin America. Baltimore: Johns Hopkins Uni- versity Press. ------. 1997. "Testing the Limits of Privatization: Argentine Railroads." World Development 25 (12): 1973­93. Ramamurti, Ravi, and Raymond Vernon. 1991. "Privatization and Control of State-Owned Enterprises." EDI (Economic Development Institute) Develop- ment Studies. World Bank, Washington, D.C. Ramírez, Miguel D. 1998. "Privatization and Regulatory Reform in Mexico and Chile: A Critical Overview." Quarterly Review of Economics and Finance 38 (3): 421­39. Ros, Agustin. 1999. "Does Ownership or Competition Matter? The Effects of Telecommunications Reform on Network Expansion and Efficiency." Journal of Regulatory Economics 15 (1): 65­92. Sáez, Raul. 1992. "An Overview of Privatization in Chile: The Episodes, the Re- sults, and the Lessons." Consultancy report, CIEPLAN (Corporación de Inves- tigaciones Económicas para América Latina), Santiago, Chile. Sánchez, Manuel, and Rossana Corona. 1993. "Privatization in Latin America." Inter-American Development Bank, Centers for Research in Applied Econom- ics, Washington, D.C. THE TRUTH ABOUT PRIVATIZATION IN LATIN AMERICA 65 Sheshinski, Eytan, and Luis F. López-Calva. 2003. "Privatization and Its Benefits: Theory and Evidence." In K. Basu, P. Nayak, and R. Ray, eds., Markets and Governments. Oxford, U.K.: Oxford University Press. Shleifer, Andrei. 1998. "State versus Private Ownership." Journal of Economic Perspectives 12 (4): 133­50. Shleifer, Andrei, and Robert W. Vishny. 1994. "Politicians and Firms." Quarterly Journal of Economics 109 (4): 995­1025. ------. 1996. "A Survey of Corporate Governance." Working Paper 5554. Cam- bridge, Mass.: National Bureau of Economic Research. Stiglitz, Joseph. 2002. Globalization and Its Discontents. New York: Norton. Torero, Máximo. 2002. "Impacto de la privatización sobre el desempeño de las empresas en el perú." Documento de Trabajo 41. Grupo de Analisis para el De- sarrollo, Lima, Perú. Torero, Máximo, and Alberto Pasco-Font. 2001. "El impacto social de la privati- zación y de la regulación de los servicios públicos en el Perú." Working Paper 35. Grupo de Análisis para el Desarrollo, Lima, Perú. Torero, Máximo, Enrique Schroth, and Alberto Pasco-Font. 2003. "The Impact of Telecommunications Privatization in Peru on the Welfare of Urban Con- sumers." Economía 4 (1): 99­128. Trujillo, Lourdes, Noelia Martin, Antonio Estache, and Javier Campos. 2002. "Macroeconomic Effects of Private Sector Participation in Latin America's Infra- structure." Policy Research Working Paper 2906. World Bank, Washington, D.C. Vickers, John, and George Yarrow. 1988. Privatization: An Economic Analysis. Cambridge, Mass.: MIT Press. Wallsten, Scott. 2000. "Telecommunications Privatization in Developing Coun- tries: The Real Effect of Exclusivity Periods." SIEPR Policy Paper 99-21, Stan- ford, Calif. ------. 2001. "An Empirical Analysis of Competition, Privatization, and Regula- tion in Africa and Latin America." Journal of Industrial Economics 49 (1): 1­19. ------. 2002. "Does Sequencing Matter? Regulation and Privatization in Telecommunications Reforms." Policy Research Working Paper 2817. World Bank, Washington, D.C. Winston, Clifford. 1993. "Deregulation: Days of Reckoning for Microecono- mists." Journal of Economic Literature 31 (3): 1263­89. World Bank. 2001. World Bank Privatization Database. Washington, D.C. Wruck, Karen. 1990. "Financial Distress, Reorganization, and Organizational Ef- ficiency." Journal of Financial Economics 27 (2): 419­44. Yarrow, George. 1986. "Privatization in Theory and Practice." Economic Policy 2 (4): 324­64. 2 The Benefits and Costs of Privatization in Argentina: A Microeconomic Analysis Sebastián Galiani, Paul Gertler, Ernesto Schargrodsky, and Federico Sturzenegger SINCE THE BEGINNING OF THE 1980S, THE world has undergone a major shift in thinking about the appropriate economic role of the state. Privatization of state-owned enterprises (SOEs) has been at the core of this change ever since Britain and France initiated privatization planning. In the last two decades, several countries have launched ambitious privatization pro- grams. Although the extent, form, and pace of change have varied from country to country, the general trend has been similar: the state has grad- ually withdrawn from directly producing goods and services. Despite the importance of this experience, we still have little empirical knowledge about how well privatization works in practice. Few studies have analyzed the impact of privatizations. Early empiri- cal research found mixed results about the relative performance of private versus public firms (Caves 1990; Vining and Boardman 1992). More re- cent research finds private ownership to be generally more efficient than public ownership (see, for example, Megginson, Nash, and van Randen- borgh 1994). These studies focus only on the question of productive effi- ciency. Recently, however, La Porta and López-de-Silanes (1999) studied Mexico's experience with privatization in the 1980s and early 1990s. They analyzed how privatization changed the performance of SOEs over a broad set of outcomes. Additionally, these authors considered the possibility that the increased profitability of privatized companies came at the expense of society through higher prices or layoffs. In this chapter, we follow La Porta and López-de-Silanes (1999) to evaluate the Argentine privatization program. Thus, we study the effects 67 68 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER of privatization on profitability, operating efficiency, productivity, out- put, investment, employment, wages, and prices. The structure of the Argentine public firms, however, was very different from the privatized Mexican public sector. In Argentina the state primarily owned a few large natural monopolies. In Mexico the state ran a large number of firms across several productive sectors. Thus, although both privatiza- tion programs were massive, the Mexican experience was richer in the number of cases compared with the Argentine experience, while the Argentine privatization program was enormous relative to the size of the economy. Mexico privatized around 1,000 firms of various sizes in sectors throughout the economy, but some of the largest public compa- nies such as PEMEX, the oil monopoly, or the electricity companies were not privatized. In contrast, Argentina privatized a smaller number of firms of much larger average size (Lustig 1992; Galiani and Petrec- olla 1996, 2000). The particular features of the Argentine privatization process allow us to study the direct impact of privatization in sectors in which, because the state was a monopolist, the whole industry was transferred to the private sector. In such cases, laid-off workers may not be able to use their sector- specific human capital in other sectors of the economy, or consumers may have no choice in their suppliers. We go beyond the impact of privatiza- tion on firms to measure the direct impact of privatization on consumers' and workers' welfare. We propose two direct measures of the welfare impact of privatiza- tions. First, the Argentine program involved the privatization of local water and sewerage firms. Changes in the health of the population associ- ated with these privatizations would provide a measure of the impact of privatization that goes beyond transfers of consumer surplus. We evaluate how the privatization of local water and sewerage firms affected both ac- cess to these services and child mortality. Second, the Argentine program involved massive layoffs. Profitability gains in privatized firms may have been obtained at the expense of workers (Shleifer and Summers 1988). We measure the effect of privatizations on workers' wages by comparing the wages of a random sample of workers before and after they were laid off from the former state oil company (Yacimientos Petrolíferos Fiscales, or YPF) with a matched counterfactual group built up using data gathered from an ongoing household survey. In short, in this chapter we address three questions: · How did privatization affect the performance of firms and through which channels--market power or productivity gains? · Are there direct welfare impacts of privatization that can be rigor- ously identified in an econometric sense? In particular, has the privatiza- tion of water and sewerage services improved or worsened the health of the population? THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 69 · Part of the efficiency gains of privatized firms may have come from the breach of explicit and implicit contracts between workers and firms. What is the evidence of this for Argentina? What has been the effect of the privatization of YPF on the earnings of laid-off workers? This chapter assesses both the efficiency and some significant distribu- tional impacts of the Argentine privatization program. This is done by considering privatization as a policy instrument and by exploiting the fact that the group of economic units (that is, SOEs, public banks, households, and workers) exposed to privatization varied both by unit and by year. That enables us to use a similar statistical identification strategy to docu- ment some of the benefits and costs of privatization. Although we are not able to identify all the efficiency and distributional impacts of privatiza- tion by applying this treatment-and-control-group approach, our main contribution to the literature is to document causal effects of privatization on measures of efficiency and distribution.1 Our results show that the profitability of the nonfinancial firms increased after their privatization. Both the ratio of operating income to sales and the ratio of net income to sales increased significantly as a result of privatiza- tion. Large increases in operating efficiency underpin these gains in prof- itability. Thus, we find a huge overall increase in the operating efficiency of the privatized firms in Argentina. Employment cuts are a big part of the story, however. Employment decreased approximately 40 percent as a result of privatization. Labor productivity increased not only because employment decreased, but also because privatized firms increased production. Privati- zation also had a big impact on investment; all the measures of investment analyzed were positively and significantly affected by privatization. Invest- ment itself increased at least 350 percent as a result of privatization. Finally, we do not find any statistically significant effect of privatization on prices. Nevertheless, after the privatization, prices did not decrease. The efficiency gains we documented imply that prices should have fallen if the quality im- provements were not large enough. Contrary to the case of nonfinancial firms, we do not find large over- all increases in operating efficiency after the privatization of public banks. However, some indicators of efficiency performed well because of privati- zation. Output per employee increased 20 percent, while the average number of employees per branch decreased 37 percent as a result of pri- vatization. As in the case of nonfinancial firms, employment cuts are a big part of the story. Employment decreased approximately 36 percent be- cause of privatization. Thus, on several indicators, the privatized banks seem to be more efficient after the privatization than they were before. Finally, the average capitalization ratio (the ratio of net worth to assets) increased 5 percent because of privatization. The higher capitalization rate of the privatized banks means a more solvent system, which is quite important in countries as vulnerable to external shocks as Argentina. 70 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER On the direct measures of welfare analyzed, we find a negative and sta- tistically significant effect of the privatization of water services on child mortality. The estimated coefficient implies a decrease of approximately 5 percent in child mortality rates induced by the privatization of water provision. Our estimates of earnings losses of displaced workers indicate that a huge redistribution cost is associated with the privatizations of SOEs: displaced workers incur substantial earnings losses. We estimate that after taking unemployment into account, the earnings losses because of privatization were approximately 50 percent of the real earnings of the workers before privatization. In this chapter, we limit ourselves to analyzing the effects of privatiza- tion on several measures of firm performance, and on consumers' and em- ployees' welfare in a partial equilibrium analysis. An important caveat is that we do not evaluate general equilibrium effects of the massive privati- zation program implemented in Argentina.2 Indeed, it would be possible to argue that the deep macroeconomic crisis that the country is suffering is to some extent related to the previous privatization policies. For example, if the privatization package distorted the equilibrium path of the exchange rate, it could have induced a severe and unsustainable misallocation of re- sources in the economy. Moreover, the debt financing associated with ac- quisition of the privatized firms contributed to the large increase in the country indebtedness. In addition, the massive layoffs that accompanied privatization may have contributed to the severe increase in unemploy- ment. Rather than a macroeconomic study, however, we conduct an ex- clusively microeconomic analysis in the industries and markets in which privatization took place. The chapter is organized as follows. In the next section, we document the Argentine privatization program, followed by a sample of privatized firms. The next two sections present the results of the effect of privatiza- tion on the performance of both financial and nonfinancial firms. We then study the impact of the privatization of water and sewerage companies on both access to these services and child mortality, followed by an analysis of the impact of privatization on the earnings of long-term displaced workers. The last section presents our conclusions. The Argentine Privatization Program In 1989 Argentina was in the midst of an acute hyperinflation driven by the monetization of large fiscal deficits. The administration of recently elected president Carlos Menem launched an ambitious privatization program. A remarkable characteristic of the Argentine privatization program was its extent and speed. The program included most SOEs as well as other state assets that had not previously been operated as inde- pendent firms (Galiani and Petrecolla 1996, 2000; Heymann and THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 71 Kosacoff 2000). The privatization program was launched together with other structural reforms, such as financial and trade liberalization, the implementation of a monetary currency board in 1991 (Plan de Con- vertibilidad), the independence of the central bank (Banco Central de la Republica Argentina, or BCRA), the decentralization of health and ed- ucation services, and other pro-market actions such as a general dereg- ulation of economic activities. The fiscal revenues from the privatization of the SOEs constituted a crucial instrument of the stabilization programs launched by the Menem government. According to Gerchunoff (1992), the main objective of the privatization program, at least at the beginning, was to solve Argentina's (intertemporal) fiscal problems. Company-specific reasons also drove the privatization process. After a long period of negative net investments, the SOEs needed high levels of capital investment to improve both the quality of and access to their services. The public sector had no means to fund those investments. In addition to its direct effects, the privatization pro- gram signaled a clear change in the direction of the economic development of the country. For the most part, SOEs in Argentina were large, vertically integrated, natural monopolies. These characteristics implied that Argentina priva- tized a small number of very large firms and that the Argentine privati- zation program was huge relative to the size of the economy. Under the priority of raising privatization revenues, in many sectors the authorities decided to maintain a monopolistic structure in order to make the new private companies more attractive to the new buyers. With the same ob- jective, prices for the services or products were raised in the immediate preprivatization period. The tax structure to which the new companies were going to be exposed was simplified. Moreover, the state absorbed the companies' liabilities before transferring the companies to private hands. Finally, the new companies enjoyed considerable regulatory free- dom at the beginning of the program. The creation of regulatory agen- cies was delayed or neglected during the early years of the privatization program. The transfer of companies and assets to private control took several forms, such as total sale through open international auctions, conces- sions, public offer of shares, licensing, leases with or without purchase options, management contracts, and the issue of exploration permits. The government obtained revenues in the form of cash and external debt bonds. A positive fiscal impact also resulted from a reduction in current losses, which were previously financed by the public budget, and a positive flow of taxes from the privatized companies. Table 2.1 presents the revenues from privatizations by sector in federal and provincial transfers according to the Ministerio de Economía (2000). The table shows the income for every sale (annual fees paid for conces- sions are not included). 72 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Table 2.1 Privatization Revenues in Argentina, by Sector (US$ millions, unless otherwise indicated) Percent Bonds, Bonds, Total of market nominal Other Sector income total Cash value value revenues Federal privatizations Petroleum 7,594 39.1 6,716 878 1,271 n.a. and gas Electricity 3,908 20.1 1,989 1,451 2,586 468 Communications 2,982 15.4 2,279 703 5,150 n.a. Gas 2,950 15.2 1,553 1,397 3,116 n.a. Transportation 756 3.9 284 183 1,314 290 Petrochemical 438 2.3 418 20 132 n.a. Banks and 394 2.0 394 n.a. n.a. n.a. financial services Steel 158 0.8 143 14 30 n.a. Derivatives from 116 0.6 116 n.a. n.a. n.a. petroleum and gas Pipelines 77 0.4 77 n.a. n.a. n.a. Construction 20 0.1 20 n.a. n.a. n.a. Other 11 0.1 11 n.a. n.a. n.a. manufacturing industries Hotels and 8 0.0 3 5 13 n.a. restaurants Chemical 5 0.0 3 2 3 n.a. Electronics 2 0.0 1 n.a. 1 1 Agriculture 2 0.0 2 n.a. n.a. n.a. Total federal 19,422 100.0 14,009 4,653 13,615 759 Provincial privatizations Electricity 2,085 47.1 2,068 n.a. n.a. 18 Petroleum and 1,703 38.5 1,703 n.a. n.a. n.a. gas Water and 589 13.3 589 n.a. n.a. n.a. sewerage Paper 50 1.1 50 n.a. n.a. n.a. Total provincial 4,427 100.0 4,410 n.a. n.a. 18 n.a. Not applicable. Note: Income from every privatization sale is included, but annual concession fees are not. Other revenues includes the use of trusts and liabilities assumed by the companies. Source: Ministerio de Economía 2000. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 73 Figure 2.1 Percentage of Accumulated Income from Privatizations, 1990­98 Percent 100 80 60 40 20 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 Note: Four-fifths (80.9 percent) of total privatization revenue was obtained between 1990 and 1994. Source: Ministerio de Economía 2000. Figure 2.1 shows the accumulation of privatization revenues during the 1990s. Nearly 81 percent of total revenues from privatizations was ob- tained between 1990 and 1994. Mostly small companies and some resid- ual shares of large companies were sold in the second half of the decade. The Sample According to official statistics (CEP 1998 and the Central Bank), 154 pri- vatization contracts were signed during the 1990s. However, the sample of privatized SOEs analyzed in our study is smaller than the number of signed contracts for several reasons: · Several SOEs were split vertically and horizontally into smaller units or assets and privatized separately. In the majority of these cases, it is not possible to obtain preprivatization financial statements and performance indicators reported separately according to the criteria used to break up the SOEs.3 This reduces the number of observations, since our unit of analysis has to be the SOE and not the private companies that emerged out of the process. 74 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER · Concessions of roads, freeways, and docks cannot be analyzed because no financial statements are available for the preprivatization period. Roads, freeways, and docks were not organized as companies under public owner- ship. · The sale of state minority participation in private companies is not considered in our study since the privatization itself did not imply a change in the management objectives of those firms. · Within the oil sector, some contracts involved exploration permits of areas where the state oil company had not previously operated (áreas petroleras marginales). · Several SOEs were liquidated or ceased operation. · In a few cases, data for the SOEs are not available. Given these constraints, we have been able to collect data for 21 federal nonfinancial SOEs and for all the privatized banks. Our database for non- financial SOEs accounts for 81.7 percent of the revenues from the sale of companies that continued operating as separate companies after being privatized and for 72.4 percent of the total privatization revenues.4 Ap- pendix 2A describes the industrial structure and data sources for the sec- tors included in our study. As stated above, our units of analysis are the SOEs. Therefore, we ag- gregate the information from all the companies that resulted from the pri- vatization of each SOE, with the exception of Ferrocarriles Argentinos, the one company where it was possible to find data by business unit for the preprivatization period. Table 2.2 presents the set of nonfinancial compa- nies included in our study. The smaller nonfinancial privatized firms not considered in our study are listed in table 2.3, along with the reason they were excluded. Finally, we construct a separate database for the banking sector, where for regulatory reasons we have monthly data for an extended number of variables, and for which we have a control group comprised of nonpriva- tized public banks and private banks. The privatized provincial banks are listed in table 2.4 Results for Nonfinancial Firms With the objective of analyzing the benefits and costs of privatization, in this section we report the effects of privatization of nonfinancial firms on profitability, operating efficiency, productivity, output, investment, em- ployment, wages, and prices. In doing so, we document the effects of pri- vatization on several measures of firm performance. Suppose one is interested in estimating the influence of a policy in- strument on an outcome for a group--in our case, for example, the effect of privatization on productivity. Thus, the group consists of state-owned Table 2.2 Nonfinancial Companies Included in the Database Years with data State-owned company name Private company name Public Private Obras Sanitarias Aguas Argentinas 1988­92 1993­97 ENTEL Telefónica 1985­90 1991­99 Telecom Trenes de Buenos Aires Metrovías Ferrovías Transp. Metropolitanos Gral. Roca Cargo: Cargo: Transp. Metropolitanos Gral. San Martín 1989­92 1993­99 Ferrocarriles Argentinos Transp. Metropolitanos Belgrano Sur Ferroexpreso Pampeano Urban passenger: Urban passenger: Ferrosur Roca 1991­92 1993­99 Ferrocarril Mesopotámico Nuevo Central Argentino Buenos Aires al Pacifico Aerolíneas Argentinas Aerolíneas Argentinas 1986­89 1992­94 TGS TGN Gas del Estado Dist. de gas Metropolitana 1987­92 1993­99 Dist. de gas Buenos Aires Norte Dist. de gas Noroeste 75 Dist. de gas del Centro (Table continues on the following page.) 76 Table 2.2 (continued) Years with data State-owned company name Private company name Public Private Dist. de gas del Litoral Dist. de gas Cuyana Gas del Estado 1987­92 1993­99 Dist. de gas Pampeana Dist. de gas del Sur YPF YPF 1987­90 1991­99 Transener Hidroeléctrica Piedra del Aguila Hidronor Hidroeléctrica Cerros Colorados 1986­91 1993­99 Hidroeléctrica Alicura Hidroeléctrica El Chocón Edenor Edesur SEGBA Edelap 1986­91 1992­95 Central Costanera Central Puerto SOMISA SIDERAR 1987­91 1995­98 Encotel Correo Argentino 1989­96 1997­2000 Tandanor Tandanor 1988­91 1994­99 Note: For full names of companies and other information about the database, see appendix 2A. Source: Authors' data. Table 2.3 Nonfinancial Privatizations in Argentina Not Included in the Database Divested assets Ceased to operate Not operating as Information from YPF or liquidated companies in public period not found 86 oil marginal areas Astillero Domecq Garcia Administración General de Altos Hornos Zapla Puertos (AGP) 6 Docks Area Petrolera Aguaragüe Carboquímica Argentina Elevador Terminal del Canal 11 Puerto de Quequen Area Petrolera El Huemul­ Fabricaciones Militares Elevadores Puerto de Buenos Canal 13 Koluel Kaike (Acido Sulfúrico, de armas Aires Area Petrolera Palmar Largo Matheu, Pilar, Río Tercero Elevadores Puerto Diamante Fabricaciones Militares Cargas, San Francisco) (San Martin, ECA, Area Petrolera Puesto ELMA Elevadores Terminales de Tolueno Sintético, Área Hernández Rosario Militar Córdoba) Area Petrolera Santa Cruz I Empresa de Desarrollos Highways Hipódromo Argentino Especiales Area Petrolera Santa Cruz II Entesa Hotel Llao Llao Interbáires Area Petrolera Tierra del Forja Argentina Navigation waterways Petroquímica Bahía Blanca Fuego Area Petrolera Tordillo Hipasam Unidad Portuaria San Pedro Radio Belgrano Area Petrolera Vizcacheras Induclor Buques Tanque (YPF) Intesa Destilería Dock Sud (YPF) Radio Excelsior Destilería San Lorenzo (YPF) Sateena Oleoductos del Valle (YPF-- Sidinox 77 70 percent) (Table continues on the following page.) 78 Table 2.3 (continued) Divested assets Ceased to operate Not operating as Information from YPF or liquidated companies in public period not found Planta de Aerosoles Dock Sisteval Sud (YPF) Puerto Rosales (YPF--70 Sitea percent) Refinería Campo Durán Tanque Argentino Mediano Tecnología Aeroespacial Note: Divested assets from YPF are assets that were part of YPF (Yacimientos Petrolíferos Fiscales) before the privatization but not afterward. Includes areas that were not operated before their concession. No data were available for companies that ceased to operate or were liquidated after the privatization process or for privatized companies that were not operated as independent organizations before privatization. Records for the com- panies in the last column could not be found. Source: Authors' data. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 79 Table 2.4 Privatized Banks in Argentina Included in the Database Number of available monthly observations Before After Bank Privatization date privatization privatization Caja de Ahorro March 1994 8 62 Chaco November 1994 14 58 Entre Ríos January 1995 17 57 Formosa December 1995 29 45 Misiones January 1996 23 23 Río Negro March 1996 30 41 Salta March 1996 31 43 Tucumán July 1996 35 39 San Luis August 1996 37 25 Santiago del Estero September 1996 38 36 San Juan November 1996 34 34 Previsión Social November 1996 41 18 de Mendoza Mendoza November 1996 41 27 Jujuy February 1998 47 20 Municipal de August 1998 60 14 Tucumán Santa Cruz December 1998 53 10 Santa Fe January 1999 50 9 Note: Privatized provincial banks are included in the sample. In May 1998 the Bank of Mendoza acquired the privatized Banco de Previsión Social de Mendoza. Source: Authors' data. enterprises i 1 . . . N observed over a sample horizon t 1 . . . T. Sup- pose further that the policy instrument (that is, the privatization of a firm) changes in a particular period t for a segment of the group (or, as in our case, that it changes for all the SOEs but at different points in time). Let dPit be a 0­1 indicator that equals unity if the privatization was operative for firm i in period t. Firms of the group that experience privatization react according to a parameter . The standard statistical model to estimate is the following two-way fixed effect error compo- nent model:5 yit dPit t i it (2.1) where iis a time-invariant effect unique to firm i that also captures in- dustry differences, t is a time effect common to all firms in period t, and it is an individual time-varying error distributed independently across 80 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER individuals and time and independently of all iand t (Chamberlain 1984; Heckman and Robb 1985). The difference-in-difference estimator of answers the question we are interested in: how the expected value of a specific variable y (that is, the dependent variable in equation 2.1) changes in any period if the SOE is privatized. Thus, E(yit | dPit 1) E(yit | dPit 0) for all i and t. This estimator assumes that, in the absence of privatization, the mean change in the privatized and nonprivatized firms would have been the same. Thus, the change in the outcome measured in the comparison group serves to benchmark common period effects among SOEs.6 As explained in the previous section, the privatizations in our data set occurred at different points in time.7 The Argentine privatization program induced some exogenous variation in the transfer of enterprises across SOEs and time. Thus, our identification strategy exploits the fact that ex- posure to privatization varied both by firm and by year.8 As we do not have information for the whole period for every com- pany, our 21 nonfinancial firms compose an unbalanced panel: Year 1985 1986 1987 1988 1989 1990 1991 1992 Public 1 4 7 10 15 14 18 14 Private 0 0 0 0 0 0 2 4 Year 1993 1994 1995 1996 1997 1998 1999 2000 Public 1 1 1 1 0 0 0 0 Private 18 19 19 18 19 18 16 1 We are interested in analyzing the change in performance of our sam- ple of firms following privatization. We rely on six broad indicators of performance: profitability, operating efficiency, employment and wages, capital investment, total output, and prices. Appendix tables 2B.1 and 2B.2 describe our variables. We express nominal variables in 1999 pesos, deflating them by the aggregate consumer price index (CPI).9 Following La Porta and López-de-Silanes (1999), we calculate two profitability ratios: operating income to sales and net income to sales. Evaluating changes in operating income offers a superior measure of effi- ciency gains, whereas evaluating changes in net income provides a useful summary statistic of the full impact of privatization on the performance of the SOEs. We could also have evaluated the impact of privatization on the ratio of operating (net) income to fixed assets, but our measure of fixed as- sets (property, plant, and equipment, or PPE) is not reliable because of the difficulties in consistently measuring PPE in periods of extreme price in- stability. We observed that PPE adjusts dramatically downward after privatizations. Moreover, it is likely that PPE could have been overstated THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 81 in the SOEs' balance sheets because the cost of public investment projects used to be extremely high. We also examine three indicators of operating efficiency to capture changes in the ability of firms to produce output from any given level of inputs (compare La Porta and López-de-Silanes 1999). We compute the logarithm of unit costs (defined as the ratio of the cost of labor and inter- mediate inputs to sales), the logarithm of sales per employee, and the log- arithm of output per employee. We also analyze the impact of privatization on labor variables: the log- arithm of total employment and the logarithm of real average wages. To assess the impact of privatization on capital formation, we examine the level of investment.10 Here, we consider the logarithm of investment, the logarithm of investment to sales, the logarithm of investment to total em- ployment, and the logarithm of investment to fixed assets (PPE). Finally, we examine the behavior of output and prices. Before analyzing the impact of privatization on these indicators, we need to discuss some econometric issues. Although it is customary to study the influence of a policy instrument on a (conditional) mean outcome, it is advisable in our case to also study the influence of privatization on the (conditional) median of the distribution of the firm performance indica- tors studied. Let: yit dPit t i it (2.2) with Q (yit | dPit, ) t dPit t i, where Q (yit | dPit, ) denotes t the th conditional quantile of y given dP for a given unit i and period t. This is the quantile regression model of Koenker and Basset (1978). The quantile regression model is concerned with the distribution of a scalar random variable y conditional on a vector of covariates x when the -quantile of y conditional on x is a linear function in x. For exam- ple, consider the case where equals 0.5. This is the median regression. This estimator is obtained by minimizing the sum of absolute errors and is referred to as the least absolute deviation (LAD) estimator. The LAD estimator is a robust alternative to the ordinary least squares (OLS) estimator for estimating the parameters of a linear regression func- tion.11 A potentially serious problem in our data set is the presence of se- vere outliers in many of the measures of firm performance that we analyze. The LAD estimator protects us against outliers in the dependent random variable y and is preferable over the OLS estimator in this respect. Moreover, the impact of privatizations on any outcome considered in our study is likely to be heterogeneous across SOEs, and this heterogene- ity is unlikely to be successfully parameterized. Thus, the OLS estimate of in equation 2.1 is likely to estimate a mixture of different population parameters (different privatization impacts across industries) with a 82 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER severely skewed distribution. The median impact of privatization on any outcome will be less influenced by extreme observations (impacts) than the mean impact of privatization on any outcome. Thus, the impacts of pri- vatization on any performance indicator are likely to be better represented by its median than by its mean. In that case, equation 2.2 could instead be estimated with equal to 0.5 and the consequent redefinition of the pa- rameter of interest. The heterogeneity of impacts of privatization across SOEs also leads us to study the percentage change of any variable y with respect to privatiza- tion instead of the level impact of privatization on these variables whenever that is practical.12 It is reasonable to assume that the former parameter is much less heterogeneous across industries than the latter one. Turning now to the results, note first that the profitability of privatized firms increased dramatically after the privatization period. The SOEs in our sample were highly unprofitable during the preprivatization period. La Porta and López-de-Silanes (1999) find similar results for Mexico. Table 2.5 shows simple comparisons before and after privatization for the mean and median operating-income-to-sales and net-income-to-sales ratios. Both profitability performances show statistically significant in- creases after privatization. The huge differences that exist between the mean and median statistics, especially in the preprivatization period, sug- gest that the parameter of interest in equation 2.2 is more appealing than the one in equation 2.1. In table 2.6 we report the estimate of the impacts of privatization on both the conditional mean and the conditional median of the set of indi- cators we propose to analyze. Thus, we report the difference-in-difference estimates of the impact of privatization on the set of indicators proposed. Table 2.5 Changes in Profitability for the Sample of Nonfinancial Privatized Firms Mean Mean Median Median before after t-statistic before after Z-statistic privati- privati- for change privati- privati- for change Variable zation zation in mean zation zation in median Operating 0.579 0.158 2.59*** 0.100 0.055 4.32*** income to sales Net 0.479 0.030 3.49*** 0.157 0.040 15.90*** income to sales *** Significant at the 1 percent level. Note: See appendix table 2B.1 for definitions of variable. The number of (firm-year) observations is 170. Source: Authors' calculations. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 83 Table 2.6 Changes in Performance for the Sample of Nonfinancial Privatized Firms Number of Mean Median Variable observations regressions regressions Profitability Operating income/sales 168 0.75* 0.83*** (0.41) (0.01) Net income/sales 168 1.03** 1.06*** (0.45) (0.01) Operating efficiency Log(unit cost) 126 5.63*** 0.1*** (1.46) (0.06) Log(sales/employment) 145 1.02 0.09*** (0.69) (0.02) Log(production/employment) 111 4.53*** 0.38*** (1.33) (0.07) Labor Log(employment) 148 0.65** 0.50*** (0.31) (0.04) Log(average real wage per 72 0.096 0.34*** employee) (0.35) (0.07) Investment Log(investment) 88 1.7* 1.51*** (1.03) (0.22) Log(investment/sales) 88 1.3 0.29*** (0.87) (0.09) Log(investment/employment) 71 4.25*** 2.21*** (1.5) (0.07) Log(investment/noncurrent 86 2.07*** 1.8*** assets) (0.79) (0.06) Output Log(production) 150 6.4*** 0.22*** (0.96) (0.03) Prices Log(prices) 155 0.11 0.02 (0.19) (0.03) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: See appendix table 2B.1 for definitions of variables. All regressions include year and firm fixed effects. Standard errors are in parentheses. Obviously, the observations for 1985 and 2000 are excluded from the estimated regression functions since the models include year effects. The number of observations varies across regressions because information is not available for all variables for every firm during the sample period. For each firm, we exclude from the sample the observation for the year in which the company was privatized due to the lack of reliable data during the transition period. Results are similar if we also exclude the two years before privatization. None of the results changes qualitatively if we exclude the data for 1989 from the analysis. In that case the estimates become more precise. Source: Authors' calculations. 84 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER The distinction between these estimates and the before-and-after esti- mates reported in table 2.5 is that the difference-in-difference estimates also control for the common aggregate effects (year effects) on the de- pendent variables. We confirm the significant increase in profitability af- ter privatization. Both operating income to sales and net income to sales increased significantly as a result of privatization. This result and, indeed, all the results reported in table 2.6 are qualitatively robust to the param- eter analyzed (mean or median). Large increases in operating efficiency underpin these gains in prof- itability. The impact of privatization on the (conditional) median unit costs shows a reduction in the latter of 10 percent. This effect is close to the effect found by La Porta and López-de-Silanes (1999) for Mexico. The impact of privatization on (conditional) mean unit costs, however, is implausibly large. Most likely, it shows the pervasive upshot of ex- treme effects in some SOEs. Because this occurs with other variables as well, we emphasize the report of the impacts of privatization on the con- ditional median of the performance measures studied, although none of the reported results changes qualitatively if we concentrate on the impact of privatization on the conditional mean of the performance measures. The ratio of median sales to employment also increases 10 percent because of the privatization of the SOEs. Finally, the impact of privati- zation on labor productivity, measured by the ratio of production to total employment, is dramatic. The impact of privatization on the me- dian level of productivity shows an increase of 46 percent. Thus, we find a huge overall increase in the operating efficiency of the privatized firms in Argentina. As in the Mexican case, employment cuts explain a large part of the in- crease in profitability. Employment decreased approximately 40 percent as a result of privatization. It is likely that this figure underestimates the layoffs experienced by privatized firms because in some SOEs, employ- ment was already falling during the immediate preprivatization period. For example, a significant proportion of the layoffs at YPF occurred two years before the privatization of the big oil company. Nevertheless, our re- sults show that a substantial proportion of the layoffs occurred after the firms were privatized. Labor productivity increased not only because employment decreased but also because privatized firms increased production. The median level of production increased 25 percent because of privatization. The impact of privatization on the real average wage for a pool of work- ers is unlikely to be identified because the composition of workers' human capital probably changed with the layoffs accompanying the privatizations. On the one hand, many workers were laid off in early retirement plans and hence, on average, SOE workers had more tenure than the average re- maining worker in the privatized firms.13 Additionally, positions created through nepotism and cronyism disappeared because of privatization, and THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 85 these positions had likely been rewarded above the average wage. On the other hand, casual evidence shows that managers' real wages increased substantially because the privatized firms had to pay competitive wages to attract skilled executives to replace the previous, politically appointed SOE directors. Thus, when we consider the impact of privatization on average wages at the firm level, the fixed effect assumption of the difference-in- difference estimator breaks down because the composition of the workers' human capital likely changed with the privatization of the SOEs. Never- theless, the estimated impact of privatization on average real wages seems to be negative or null.14 This impact appears to vary greatly across firms, reaffirming our suspicion that the identified effect of privatization on wages is mainly driven by composition effects instead of productivity effects (see also La Porta and López-de-Silanes 1999). Following La Porta and López-de-Silanes (1999), we evaluate the con- tribution of layoffs to the changes in profitability. We compute operating income for the postprivatization period for each SOE maintaining the em- ployment at the preprivatization level. Then, we estimate equation 2.2 for the operating-income-to-sales ratio. The coefficient of the privatization dummy variable drops to 0.67.15 Only 20 percent of the estimated increase in the ratio of median operating income to sales seems to be attributable to workers' layoffs, a figure considerably lower than the one estimated by La Porta and López-de-Silanes (1999) for Mexico.16 Regarding the impact of privatization on investment, all the measures analyzed are positively and significantly affected by privatization. Invest- ment itself increased at least 350 percent as a result of privatization. This effect is enormous and well above the one found in Mexico by La Porta and López-de-Silanes. This result is consistent with the view that one of the main motives for privatizing the SOEs in Argentina was to reestablish investment. Finally, we consider the behavior of prices. The main difficulty in iden- tifying the impact of privatization on prices is that prices were usually in- creased before the firms were privatized, substantially so in some cases, to make the companies attractive to private investors. Moreover, prices were not increased immediately before every privatization but rather when the privatization package was launched at the beginning of the 1990s. Thus, there is not enough variability across both firms and time in the changes in prices as a result of privatization to identify the effect of the latter on the former. Furthermore, we lack enough data in the immediate prepriva- tization period to document this effect. Additionally, the quality of several products supplied by the privatized firms increased significantly after pri- vatization. These changes in quality are difficult to measure but well known in several sectors such as telecommunications and electricity.17 Under these restrictions, we do not find any statistically significant ef- fect of privatization on prices. Nevertheless, prices did not decrease in the postprivatization period, when the efficiency gains we documented imply 86 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER that prices should have fallen if the improvements in quality were not large enough. These results thus suggest that there is an important pending reg- ulatory mission to accomplish in Argentina.18 Results for Banks The Argentine banking sector went through an important transformation after the Tequila financial crisis of 1995 following the devaluation of the Mexican peso in December of 1994. Under the currency board, the central bank, BCRA, faced severe limits on its ability to act as a lender of last re- sort. Thus, it could not bail out banks that were having solvency problems. Instead, BCRA helped these banks to be acquired, merged, or, in the case of public banks, privatized. This process led to a significant reduction in the total number of banks operating in the country, from 168 in December 1994 to 122 two years later (Burdisso, D'Amato, and Molinari 1998). The data set used in this study was compiled by BCRA and contains monthly financial information for all the entities that participated in the Argentine Financial System from June 1993 to September 1999. It in- cludes the basic balance sheet accounts, the net income structure, and some physical data such as the number of employees and branches for each bank. Although the data set covers the period during which almost all privatizations took place, not all the information is available for every bank variable at every moment. In particular, more disaggregated data are available for the more recent periods. These data have the advantage of being perfectly comparable across in- stitutions, as well as before and after the privatizations, as BCRA, acting as regulator of the financial system, requires the entities to present their balance sheets with the same accounts and criteria. In 1991 there were 35 public banks in Argentina, 27 owned by the provinces and 8 owned by national and municipal governments. Between 1992 and 1999, 19 of these public banks were privatized and 2 others were merged with privatized banks, leaving only 14 banks under public ownership by September 1999. We include 17 of the 19 privatized banks in our study, since no preprivatization information was available for two banks in the data set. These were the Banco de La Rioja (privatized in July 1994) and Banco de Corrientes (May 1993). The privatization of the Banco Hipotecario Nacional is also not covered here because it was done after September 1999. The variables used in our study are detailed in ap- pendix table 2B.2. When an SOE is privatized, the government usually tries to make the firm more attractive to buyers and ends up selling it--after a restructur- ing process--without the "undesired" assets and liabilities. In the case of the Argentine public banks, most of the provincial governments formed a residual entity with the low-quality assets and liabilities. To be THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 87 able to cope with the short-term liabilities, the Argentine government, the Inter-American Development Bank, and the World Bank created the Fondo Fiduciario para el Desarrollo Provincial (FFDP) to lend money and give technical assistance to the provinces for privatizing their banks. In short, the privatization of provincial banks involved the creation of residual entities with the purpose of keeping the low-quality assets and liabilities that would not be attractive to potential buyers. Because of these ownership changes, stock variables such as total assets and de- posits are worthless for detecting changes in performance. Instead, we consider performance ratios. We rely on five broad indicators of performance: profitability, oper- ating efficiency, employment, capitalization, and loan growth. We ex- press nominal variables in 1999 pesos, deflating them by the aggregate CPI index. The results show, first, that most profitability indicators of privatized banks increased dramatically in the postprivatization period. Table 2.7 shows simple before-and-after comparisons for the mean and median profitability indicators. Looking at the profitability ratios, almost every indicator is negative before the privatization period and turns positive af- ter it. The median increase in the profit margin, operating margin, interest margin, return on equity (ROE), and return on assets (ROA) are all sta- tistically significant.19 However, the median operating income per branch decreased after privatization. We now analyze the influence of the privatization of banks on the set of selected performance indicators estimating the model described in equation 2.1. In contrast with the case of nonfinancial firms, not all the public banks were privatized during the 1990s. Following our definition of the parameter in the previous section, we include only public banks in the control group. In table 2.8, we present two different estimates of the impact of the privatization of banks on their performance. The data one year before and one year after the privatizations are not included in the analysis shown in the first column, whereas they are in the second column. The data just before privatization could be misleading since the government could be trying to restructure the banks before privatization to increase their attractiveness or, in case of corruption, could have mod- ified the financial records to favor its friends in the auctions (the "cooked-books" hypothesis). The year after privatization could be con- sidered as one dedicated to the restructuring process (see La Porta and López-de-Silanes 1999). We report the difference-in-difference estimates of the impact of priva- tization on the set of indicators proposed. Most results are similar across samples. Thus, we find no evidence to support the claim that results are at- tributable to misleading accounting conducted before privatization. The overall positive impact of privatization on profitability is not con- firmed. Even though we find a statistically significant increase on ROA 88 Table 2.7 Changes in Profitability for the Sample of Privatized Banks t-statistic Median Median Z-statistic Mean before Mean after for change before after for change Variable privatization privatization in mean privatization privatization in median Profit margin (percent) 27.07 22.32 1.09 15.51 7.53 100.62*** Operating margin (percent) 37.56 15.52 0.56 22.67 5.96 109.99*** Interest margin (percent) 0.83 0.79 3.41*** 0.20 0.55 101.71*** Operating income per branch 142 123 0.61 145 107 11.47*** ROA (percent) 0.007 0.002 4.93*** 0.002 0.001 167.20*** ROE (percent) 2.305 1.224 0.66 1.141 1.149 98.91*** *** Significant at the 1 percent level. Note: ROA return on assets and ROE return on equity. See appendix table 2B.2 for definitions of variables. Number of (firm-month) observations: profit margin 987; operating margin 985; interest margin 955; operating income per branch 723; ROA 1,148; ROE 1,148. Source: Authors' calculations. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 89 Table 2.8 Changes in Performance for Privatized Public Banks Column 1 Column 2 Number of Number of observa- observa- Variable tions Coefficient tions Coefficient Profitability ROA 2,007 0.01*** 2,246 0.01*** (0.002) (0.001) ROE 2,007 0.031 2,246 0.04 (0.03) (0.05) Profit margin 1,724 1.22 1,923 1.24* (0.94) (0.77) Operating margin 1,666 0.30 1,864 0.06 (1.08) (0.84) Interest margin 1,675 0.02*** 1,874 0.03*** (0.004) (0.01) Operating income/ 1,311 227.67*** 1,462 199.45*** branch (73.73) (59.90) Operating efficiency Log(average cost) 1,859 0.01 2,089 0.02 (0.03) (0.03) Log(administrative 1,996 0.31*** 2,226 0.44*** expenses) (0.02) (0.02) Log(output/ 1,392 0.19*** 1,581 0.14*** employee) (0.03) (0.03) Log(employees/ 1,507 0.46*** 1,694 0.55*** branch) (0.02) (0.02) Employment Log(employees) 1,513 0.45*** 1,702 0.59*** (0.02) (0.02) Other Capitalization 2,010 0.07*** 2,249 0.08*** (0.01) (0.01) Loan growth 1,922 0.03*** 1,462 0.07*** (percent) (0.01) (0.01) * Significant at the 10 percent level. *** Significant at the 1 percent level. Note: ROA return on assets and ROE return on equity. See appendix table 2B.2 for definitions of variables. Column 1 does not include data for one year before and one year after privatization. Column 2 includes all observations. Source: Authors' calculations. 90 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER upon privatization, we do not find any statistically significant impact on ROE, and the impact on operating income per branch is negative and sta- tistically significant. The interest margin has also increased in association with the privatization of banks. The evidence suggests, however, that the privatization impact on both the profit and operating margins is not sta- tistically significant. Contrary to the case of nonfinancial firms, we do not find large in- creases in operating efficiency after the privatization of public banks. The impact of privatization on the (conditional) mean average costs is null. In addition, the impact of privatization on the mean administrative expenses is positive and statistically significant. They have increased 36 percent as a result of privatization. However, other indicators of efficiency per- formed better because of privatization. Output per employee increased 20 percent, while the average number of employees per branch decreased 37 percent. As in the case of the nonfinancial firms, employment cuts are a big part of the story. Thus, on several indicators, the privatized banks seem to be more efficient after the privatization than before. This result is in line with the results found in Burdisso, D'Amato, and Molinari (1998). The privatization process also implied a supply increase in the credit market. Finally, solvency improved in privatized banks. The average cap- italization ratio (net worth to assets) increased 5 percent because of priva- tization. This increase in capitalization is statistically significant. The average capitalization of the banks in the year before privatization was 10 percent, a fact that helps explain the reasons for undertaking priva- tization. The higher capitalization rate of the privatized banks means a more solvent system, which is quite important in countries as vulnerable to external shocks as Argentina. This is in line with the consensus that the reforms taken by the BCRA after the Tequila crisis regarding the approval of mergers, liquidation of bankruptcy banks, and privatizations helped to strengthen the financial system. Privatization of Water and Sewerage Companies: Access to Services and Welfare In this section, we study the impact of the privatization of water and sew- erage companies on both access to service and child mortality. There are three reasons to focus our analysis on the privatization of water and san- itation services. First, access to water supply and sanitation is a fundamental need. The significance of water, as distinct from other infrastructure industries, de- rives from the fact that human survival depends on access to water that is free of pollutants. The health and economic benefits of water and san- itation supply to households and individuals (and especially to children) THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 91 are well documented. The lack of improved domestic water supply leads to disease through two principal transmission routes: waterborne disease that occurs from drinking contaminated water, and water-washed disease that occurs from a lack of water. Diarrhea, which can be both waterborne and water-washed, is the most important public health problem affected by the quality of and access to water and sanitation. Approximately 4 bil- lion cases of diarrhea each year cause 2.2 million deaths throughout the world, mostly among children under the age of five. These deaths repre- sent approximately 15 percent of all deaths of children under the age of five in developing countries (WHO 2000). Water, sanitation, and hygiene interventions can reduce diarrheal disease, on average, by between one- fourth and one-third (Esrey and others 1991). Thus, of all privatizations, the transfers of the provision of water and sanitation services are the ones that potentially could have the highest impact on a direct measure of welfare like health. Second, the proportion of people in the world with access to water and sanitation facilities remained constant during the 1990s despite all the ef- forts and programs to increase the access of the poor to these services (WHO 2000). Thus, it is of special interest to test whether the privatiza- tion of water and sanitation services increased access to those services. Third, water and sanitation services are a natural monopoly, where declining long-run average costs mean that it is most efficient for only a single firm to serve the market. Moreover, water differs from other nat- ural monopolies in the importance of the externalities present. Both the natural monopoly feature and the health effects of water and sanitation force high public interest in the sector (Shirley 2000). Between 1991 and 2000, several provincial privatizations in the water sector occurred, in addition to the privatization of the federal Obras Sani- tarias de la Nación (OSN), which transferred the responsibility for water and sanitation service in the Buenos Aires metropolitan area to a private company, Aguas Argentinas, in May 1993.20 The provision of water has been privatized in localities covering approximately 60 percent of the pop- ulation of the country (as of the 1991 census). Water and sanitation priva- tizations were dispersed throughout the decade. Thus, there are localities in Argentina where privatization has not taken place. In those localities where privatization occurred, there is variability across both localities and time. This exogenous variation in the provision of water and sanitation services across time and space can be exploited to identify the causal effect of water and sanitation privatization on both access to water and child mortality. Table 2.9 shows the access to connection to both water and sanitation services in urban areas in 1991. Connection to the water network is high (approximately 70 percent of the population) but certainly far from full coverage like the one achieved in the Argentine capital, Capital Federal (Ciudad de Buenos Aires). Connection to the sewerage network is much lower (approximately 37 percent of the population). Localities do not 92 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Table 2.9 Access to Water and Sewerage Services, 1991 (percent) Water service Sewerage service Urban Urban Urban Urban population households population households with with with with Sample connection connection connection connection Total 70 73 37 41 Localities privatized 71 74 40 45 between 1990 and 1999 Localities not 69 70 29 32 privatized between 1990 and 1999 Capital Federal 98 98 94 94 (privatized in 1993) Note: The data are obtained from the 1991 Census of Households and Popula- tion. Urban population is the population living in all localities with more than 5,000 inhabitants in 1991. A locality is in the privatized group if the privatization of water services occurred between 1990 and 1999. Source: 1991 Census of Households and Population. show substantial differences in the proportion of the population (house- holds) with access to connection to the water network, whether it is pub- lic or private. Artana, Navajas, and Urbiztondo (1999) analyze the first two privati- zations of water and sewerage services in Argentina: Aguas Argentinas and Aguas de Corrientes. Using official data for the period 1991 to 1995, they report that the number of water connections in the area covered by Aguas de Corrientes rose by 22 percent and the number of sewerage con- nections by 50 percent. That translates into an additional 7 percentage points of the population covered with water services and 12 percentage points of the population covered with sewerage services. These increases in coverage are outstanding by any standard. For Aguas Argentinas, we obtained similar data from the regulator for the period 1980 to 1999, and we estimated the following regression function (where the notation is self- explanatory, and t equals 1, 2, 3, . . . , 20): Log (Population served) const. 0.0064 t 0.042 (t 14) I(t 14) (0.001)a (0.006)a R2 0.94 The results of the estimation are shown in figure 2.2. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 93 Figure 2.2 Logarithm of Population Connected to the Water Network and Fitted Values, Aguas Argentinas, 1980­99 Logarithm 15.9 15.8 15.7 15.6 15.5 15.4 1980 1985 1990 1993 1995 1999 year Logp observed Fitted values Note: The fitted values come from a simple regression including a constant, a linear time trend, and an indicator variable for the postprivatization period, beginning in 1993, as explained in the text. Fitted values closely follow the actual data. Source: Authors' calculations. The increase in the access to water services in the area covered by Aguas Argentinas after privatization also seems to be exceptional. After 1993 the population with access to water services increased by approximately 3 per- cent annually.21 These figures are not estimates of the causal effect of privatization of water on access to service, however. First, there is a measurement error problem since the firms know exactly where connections were expanded, but this knowledge translates noisily into population served. We have only household data on access to service from the 1991 census and from a ran- dom survey conducted in 1996­97 (by Encuesta de Desarrollo Social, or EDS) that covered all urban localities with more than 5,000 habitants and that asked the same questions about access to water connections as the 94 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER 1991 census did. Second, and more important, connections could have also expanded without privatization. Thus, to identify the causal effect of privatization on access to water, we exploit the fact that between 1991 and 1997, the two dates for which we possess household access informa- tion, some privatizations affecting several localities had already occurred. Thus, for the localities that were randomly chosen in the EDS survey, we perform a test of the difference in difference of the proportion of house- holds with access to the water network in 1991 and 1997 between urban localities where water provision had been privatized and those where it had not. The results are reported in table 2.10. Thus, we find a statistically significant increase in the access to water services caused by the privatization of firms. Similar results are found in Table 2.10 Difference in Difference of the Proportion of Households with Access to Water Connection, 1991­97 (percent) Water All localities Excluding Localities with access Capital Federal Localities not privatized: 1991 86.6 86.6 census data Localities not privatized: 1997 89.8 89.8 EDS, survey data Localities not privatized: 3.2 3.2 Difference 1997 1991 Localities privatized: 1991 73.0 64.0 census data Localities privatized: 78.0 71.4 1997 EDS, survey data Localities privatized: 5.0 7.4 Difference 1997 1991 Difference-in-difference 1.8 4.2 estimate Z-test for difference in the 2.83*** 5.78*** changes in proportions *** Significant at the 1 percent level. Note: Difference in difference of the proportion of households with access to water between 1991 and 1997, between the localities where water services were privatized and those where it was not. Only the localities randomly chosen in the EDS (Encuesta de Desarrollo Social) survey are included in the 1991 estimates. Sampling weights are used to estimate the proportions reported in the table. A locality is in the privatized group if the privatization of water services occurred between 1990 and 1996. Source: Authors' calculations. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 95 terms of population. Note that the increase is higher when we exclude Capital Federal where access was already full before the privatization. Also, note that the increase in the access to water services in the privatized regions is approximately 11 percent (from 64.0 percent to 71.4 percent, excluding Capital Federal). This increase is consistent with the increase in access to water services that we estimated for OSN for 1993­97. Because the network expansion induced by privatization is phased over a longer period (1991­97) than the one covered by the test we conduct, it is rea- sonable to conclude that the causal impact of privatizations on access to water services is greater than our estimate. Finally, we evaluate the impact of the privatization of water services on child mortality. Mortality data, compiled in Argentina by the Minis- terio de Salud, are constructed at the local level disaggregated by age. Al- though the data are not publicly available, we have been able to access the data for the 66 localities in the country with 100,000 or more in- habitants. These localities account for 58.6 percent of the total popula- tion and for 64.4 percent of the urban population. We focus on child mortality (mortality of children below five years of age) because children are more vulnerable to water-related diseases such as diarrhea (WHO 2000). We divide the number of deaths by the number of children of that age to obtain mortality rates, the dependent variable of the following analysis. Consider the evaluation of the impact of the privatization of water pro- vision on child mortality.22 The difference-in-difference estimator of the impact of privatization on mortality, , is obtained by estimating the fol- lowing regression function: Mortality Ratesit dPrivit xit t i + it (2.3) where Mortality Ratesit are the mortality rates of children below five years of age in locality i and year t, xit is a set of control variables--income and inequality23--that vary across both localities and time, and dPrivit is a 0­1 indicator that equals unity if in locality i and period t the main provider of water services is a private firm. When dPrivit is 0, the main provider of wa- ter services in locality i and period t may be a public firm or a cooperative. Finally, i is a time-invariant effect unique to locality i, tis a time effect common to all localities in period t, and it is a locality time-varying error distributed independently across locality and time, and independently of all iand t. We report the results of this exercise in table 2.11. We find that priva- tization of water services has a negative and statistically significant effect on child mortality. The estimated coefficient implies that privatization in- duces a decrease of approximately 5 percent in child mortality rates. Thus, we find that the privatization of water services both increased access to water services and reduced child mortality. 96 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Table 2.11 The Effect of Privatization on Child Mortality Rates, 1990­99 Impact Dependent variable: mortality rates dPrivit 0.21* (0.12) Number of observations 658 Number of localities 66 * Significant at the 10 percent level. Note: Standard errors are in parentheses. dPrivit is a 0­1 indicator that equals unity if in locality i and period t the main provider of water services is a private firm. When dPrivit is 0, the main provider of water services in locality i and period t may be a public firm or a cooperative. The regression includes year and province fixed effects. It also includes local income and inequality as control variables. Source: Authors' calculations. Privatization and Worker Displacement: Wages and the Distribution of Welfare A great deal of attention in recent years has been devoted to the conse- quences of worker displacement for individual labor market perfor- mance (see, among others, Hamermesh 1989 and Hall 1995). Displaced workers are generally defined as those workers who were permanently laid off without cause. This type of involuntary rupture in employment relationships is usually associated with structural change, sectoral real- location, or technological innovation. Displacement is usually followed by a period of slow rebuilding of employment relationships, as workers displaced from long-term jobs require time to find new acceptable jobs (Hall 1995). Therefore, the emphasis of our study is on long-term wel- fare losses after displacement. The workers displaced by the privatiza- tion of SOEs constitute a valuable alternative source for studying the consequences of worker displacement for individual labor market per- formance. One good reason to study worker displacement is that its consequences can yield insights into the wage determination process. Human capital the- ory predicts that to the extent that experience and skills acquired on the job are general, displaced workers should not suffer large wage losses upon reemployment. In contrast, workers with accumulated industry or firm-specific capital or workers extracting industry or firm-specific rents are likely to sustain large pay cuts when changing firms or sectors.24 Thus, even if the privatization of SOEs results in a socially efficient reduction in the level of employment in the privatized firms, the laid-off workers may still be badly harmed. If that is the case, then part of the efficiency gains of THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 97 privatized firms may have come from the breach of implicit and explicit contracts between workers and firms. In this section, we explore to what extent workers displaced by privatization suffered long-term earnings losses. We rely on a survey of a random sample of displaced workers from the former state oil company YPF. Privatization and Worker Displacement One of the salient characteristics of the Argentine privatization program is the huge reduction in employment associated with it. Employment fell approximately 40 percent as a result of privatization. A very important question in itself but also as part of a broader study of the microeconomic benefits and costs of privatization, is the following: What has been the ef- fect of the privatizations on the earnings of laid-off workers? To assess this question, we use a survey of a random sample of workers displaced from YPF in 1991 as part of the restructuring process of the firm before its privatization. The frame of the survey was a list of all displaced work- ers from YPF during 1991. The survey was conducted during the first week of August of 2001. The sample size is 504 observations. It is likely that workers in state-owned enterprises were extracting rents and that their wages therefore did not reflect their productivity. In such a case, a long-term earnings loss because of privatization estimates a dimension of the distributive cost of privatization and is not necessarily a destruction of workers' specific human capital. Thus, our concern is with the distribution of the costs of what otherwise appears to be an effi- cient reform. Nonetheless, the impact of privatization on long-term earn- ings is not a minor point in society's perceptions of the benefits and costs of privatization. Certainly, the welfare of workers depends not only on their earnings but also, among other things, on their fringe benefits, health insurance, and the stability of their jobs. Thus, we think it is informative to consider the overall subjective impact of privatization on the displaced workers' welfare. We find that approximately 60 percent of the displaced workers in our sample consider that they were adversely affected by displace- ment.25 Nevertheless, even if revealing, this is only a subjective appraisal of the overall costs of displacement. In addition to the earnings losses, the jobless rate for laid-off workers in the United States is higher than it is for unemployed workers who were not laid off (Ruhm 1991). In our sample, even in the long term, we also find that this is the case. The primary factor to consider is the age distribution of the displaced workers in our sample. Displacement because of privatization was con- centrated among workers older than 40 years; 30 percent of the surveyed individuals were older than 59 years at the time of the survey. Indeed, 60 percent of the sampled individuals were between 39 and 59 years, and 98 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER none of the displaced workers was younger than 28 years at the time of the survey. To draw inferences about this sample of displaced workers, we con- struct comparison groups from the ongoing household survey. The house- hold survey is conducted twice a year, in May and October, in the main 28 urban agglomerates of the country. The male labor force participation rate among the individuals displaced from YPF younger than 60 years is 90 percent.26 Preferably, we shall com- pare the statistics obtained from the survey of YPF displaced workers with the same statistics estimated from the May 2001 wave of the household survey in all urban agglomerates.27 Unfortunately, at the time of writing, only the data tapes of Greater Buenos Aires were available for May 2001. It is preferable to contrast estimates of 2001 with statistics obtained using data from the same year since the recession deepened during 2001. Thus, we compare the statistics obtained from the sample of displaced workers with the same statistics estimated using the data of Greater Buenos Aires for May 2001.28 The labor force participation rate among males between 28 and 59 years in Greater Buenos is 96.3 percent. The participation rate in this control group and in the displaced workers' sample differs at the 1 percent level of significance. Similarly, we find that the unemployment rate among males displaced from YPF is 26.4 percent, while unemployment in the control group is 13.9 percent. Thus, we find that even though the labor force participation rate of the displaced workers, 10 years after dis- placement, is slightly below the labor force participation rate of a compa- rable group in the population, their unemployment rate is twice as high as the population unemployment rate.29 The Long-Term Impact of Job Displacement on Earnings The bulk of the evidence on worker displacement comes from the United States; there is no evidence at all from Latin America. Even using different methods and data sets, the evidence from the United States is unambiguous: in addition to the direct income loss associated with unemployment, work- ers face large and persistent earnings losses after displacement (see, for ex- ample, Jacobson, LaLonde, and Sullivan 1993). Most of the U.S. research on the impact of job displacement on earn- ings assumes that workers' earnings at a given date depend on the time since displacement through a set of dummy variables for the number of quarters after (and possible before) displacement (see, for example, Jacobson, LaLonde, and Sullivan 1993). Consider the simplest case where we observe earnings at time t0 before displacement and at time th. At time tj, where h j 0, a group of work- ers was displaced from their jobs. If a longitudinal data set were available, we would estimate the displaced workers' earning losses as the difference between their actual and expected earnings had the events that led to their THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 99 job losses not occurred. Thus, we would estimate a two-way fixed-effect error component model like the following: yit dDPit txit t i it (2.4) where t 0 or h; yit is the logarithm of earnings of worker i in period t, xit is a set of control variables--the standard human capital variables in- cluded in a earnings equation--that vary across both individuals and time, and dDPit is a 0­1 indicator that equals unity in period h if individual i in period j was displaced from his or her job. Finally, iis a time-invariant effect unique to individual i, tis a time effect common to all individuals in period t, and it is an individual time-varying error distributed inde- pendently across individuals and time, and independently of all iand t. Note that, in general, is not allowed to be a time-varying parameter. Indeed, by default, the regression equation 2.4 models the "returns to ed- ucation" as time-invariant. This assumption is not free of problems since it assumes that the growth rate of earnings is not affected by the change in relative prices. The control group would be the workers not displaced. Thus, it is crit- ical that displacement represents an event exogenous to the wage profile of the displaced individuals and hence, the expected wages of the control group are the same as the expected wages of the displaced workers had they instead stayed in their jobs. Although, in principle, displacement caused by privatization seems to be an event exogenous to workers' deci- sions, in practice, it is not. Displaced workers are older than workers who are not displaced, and displacement is dominated by selection based on in- dividual characteristics. The main problem with the counterfactual group, however, is that their wages are not the same as the expected wages of the displaced workers if privatization had not taken place; that is because pri- vatization affects the functioning of the entire firm including productivity and real wages. Furthermore, since our parameter of interest measures long-term earnings losses because of displacement from privatization, the control group would be made up of individuals who have remained in their jobs since privatization occurred; the control group would thus com- prise an unusual group of workers that is not likely to be comparable to the group of workers displaced by privatization. In this study, therefore, we focus on a family of alternative parame- ters--namely, the parameters of interest in a study of the benefits and costs of privatization. First, we define displaced workers' earnings losses (DWEL) to be the difference between the actual and expected earnings of a displaced worker where the expected earnings are assumed to be those taken from the population of similar individuals in terms of observable socioeconomic variables (DWEL I) instead of from the workers not dis- placed from a privatized firm. We argue that, in general, this parameter is the appropriate evaluation of the earnings costs of displacement caused by 100 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER privatization. Argentina, however, has a relatively generous system of sev- erance pay (Galiani and Nickell 1999). Thus, when displaced, workers received a nontrivial severance payment that they could invest and from which they could obtain a flow of income. Naturally, workers may invest their severance differently and have different returns as a result. Nonethe- less, they could invest the severance in secure coupon bonds with a fixed interest rate and constant, regular repayment of interest (U.S. Treasury bonds, for example). Thus, an alternative estimate of the displaced work- ers' earnings losses is the difference between their actual and expected earnings where their actual earnings also incorporate the potential flow of interest on a coupon bond over the severance payment received at the time of displacement and where the expected earnings are defined in the same way as before (DWEL II). To estimate our parameters of interest, consider the data-generating process of the earnings of a typical displaced worker. In period 0, they are given by: yi0 c0 0 xi0 i i0 (2.5) while in period h they are given by: yih ch h xih i ih (2.6) Thus, if we knew the parameter vector {ct, } t t 0,h, a consistent estimate of would be given by the following before-and-after estimator: ^ it, where it yit ct t xit. Thus, ^ it it. However, we do not know {ct, } t t 0,h. To circumvent this lack of knowledge, we estimate an earning equation in a sample representative of the population in peri- ods 0 and h. Thus, we estimate the parameter vector {ct, } t t 0,husing a control group. Note that our estimator of is the simplest version of the conditional difference-in-difference matching estimator (see Heckman, Ichimura, and Todd 1997). In our sample, t 1991 and 2001. We estimate the parameter vec- tor {ct, } t t 1991,2001 by estimating earnings equations using household survey data from the Greater Buenos Aires agglomerate for the waves of October 1991 and May 2001. We sample males between 18 and 59 years old who have a job at the time these household surveys were con- ducted. The dependent variable is the logarithm of the monthly earnings of the workers in their main occupations. We exclude unpaid workers from the sample. The conditioning variables are a set of schooling dum- mies and the age and the age squared of the sampled individuals. The schooling dummy variables measure the maximum level of the educa- tional system attended by an individual and whether or not it was com- pleted. The base category in the regression function is the complete pri- mary school. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 101 Table 2.12 The Effect of Privatization on Displaced Workers' Earnings Flows DWEL I DWEL II ( ) ( ) 0.73*** 0.54*** (0.055) (0.052) Number of displaced workers' Number of displaced workers' observations: 150 observations: 139 *** Significant at 1 percent level. Note: DWEL refers to the change in the excess workers earn over their estimated market wages. Market wages are estimated from an earnings regression on the household survey. Only displaced workers currently working in 2001 are considered. DWEL I considers only current earnings. DWEL II adds the potential interest income on the severance payment (estimated at a monthly rate of 0.5 percent). The logarithm of earnings is used for the estimation. The estimate of DWEL II does not change at all if we do not impute the potential monthly flow of interests because of investing the severance payment in a secure coupon bond to a few observations that report they actually obtain monetary profits from the investment they did with the sever- ance payment they received after displacement. Standard errors, in parentheses, are computed by assuming that our estimate of {ct, } t t 1991,2001coincides with the true parameter values. Source: Authors' calculations. Since the household survey obtains data only on earnings in the month previous to the survey, we consider only the earnings of the displaced workers that were occupied at the time they were surveyed in 2001. To es- timate DWEL II, we assume a monthly interest rate of 0.5 percent on the severance payment. In table 2.12 we present our best estimates of DWEL I and DWEL II. The results are unambiguous: displaced workers face long-term substan- tial earnings losses. Our estimate of our statistic DWEL I is 51.8 percent of the earnings before privatization.30 This estimate is substantially higher than the one obtained by applying the before-and-after estimator to the displaced workers' data set, which is 39.4 percent. However, there are two reasons why this latter measure does not capture the full effect of displacement on workers' earnings. First, it does not control for macroeconomic factors that cause changes in workers' earnings regardless of whether they are displaced. Second, this measure does not account for the earnings growth that would have occurred in the absence of job loss; in the long term, workers' earnings may return to their levels before displacement, but not to the levels expected before their job losses (see Jacobson, LaLonde, and Sullivan 1993). As expected, our statistic DWEL II is somewhat lower than DWEL I. It is 41.7 percent; still, we estimate quite a substantial earnings loss after displacement, well above the earnings losses because of displacement 102 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER estimated in the United States. Thus, it appears that there is a huge re- distribution cost associated with the privatizations of SOEs: displaced workers incur substantial earnings losses. What is more, since unem- ployment is higher among displaced workers than among a comparison group in the population, the earnings losses because of displacement are higher than the ones estimated by our statistic DWEL II. Indeed, we es- timate that after taking into account unemployment, the earnings losses because of privatization are approximately 50 percent of the real earnings of the workers before privatization. Finally, it is worth considering where the earnings losses of the displaced workers come from. Figure 2.3 shows the correlation of our estimates of Figure 2.3 Displaced Workers' Earnings Rents: 1991 and 2001 1991 2 1 0 -.5 -1.8 -1 0 1 2 2001 Note: Figure shows a correlation between displaced workers' rents in 1991 and 2001 (before and after their displacement). Rents are measured as the difference between their earnings and the expected earnings from an earnings regression for all workers in the household survey. Most workers enjoyed positive rents before their displacement, but roughly half of them did in 2001. (Earnings of 2001 do not include interest flow.) Source: Authors' calculations. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 103 1991 and 2001. Notice that tmeasures the rent of a worker (that is, the difference between what an employed worker gets from his/her employ- ment relationship and his/her outside option). As can be observed, almost all workers extracted (positive) rents in 1991, while only half of them were still obtaining (positive) rents in 2001. Conclusions In this chapter, we evaluated both the efficiency and some significant dis- tributional impacts of the Argentine privatization program. This was done by considering privatization as a policy instrument and by exploiting the fact that exposure to privatization of a group of economic units (SOEs, public banks, households, and workers) varied both by unit and by year. We then used a similar statistical identification strategy to document some of the benefits and costs of privatization. Although we were not able to identify all the efficiency and distributional impacts of privatization by applying this treatment and control group approach, we contributed to the literature by documenting a wide set of causal effects of privatization on measures of efficiency and distribution. Following La Porta and López-de-Silanes (1999), we studied the effects of privatization on profitability, operating efficiency, productiv- ity, output, investment, employment, wages, and prices for both financial and nonfinancial privatized firms in Argentina during the 1990s. We also studied two direct measures of the welfare impact of privatizations. First, the Argentine program involved the privatization of local water and sew- erage firms. Changes in the health of the population associated with these privatizations would provide a measure of the impact of privatization that goes beyond transfers of consumer surplus. We evaluated how the privatization of local water and sewerage firms affected both access to service and child mortality. Second, the Argentine program involved massive layoffs. Profitability gains in privatized firms may have been obtained at the expense of workers. We measured the effect of privatiza- tions on workers' earnings by comparing the before-and-after wages of a random sample of laid-off workers from YPF with a matched counter- factual group build up using data gathered from the ongoing household survey. Before discussing our results, a caveat is in order. We perform here a partial equilibrium analysis. This chapter analyzes the effects of privatiza- tion on several measures of firm performance and on the welfare of con- sumers and employees in the markets affected by the privatizations. We do not evaluate the potential macroeconomic effects of a privatization pro- gram as massive as the one implemented in Argentina. We found that profitability of nonfinancial firms increased after privati- zation. Both operating income to sales and net income to sales increased sig- nificantly as a result of privatization. Large increases in operating efficiency 104 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER underpin these gains in profitability. The impact of privatization on (con- ditional) median unit costs shows a reduction in median costs of 10 per- cent. The median sales-to-employment ratio also increased 10 percent because of the privatization of the SOEs. Finally, the impact of privatiza- tion on the median level of productivity shows an increase of 46 percent. Thus, we find a huge overall increase in the operating efficiency of priva- tized firms in Argentina. However, employment cuts are a big part of the story. Employment decreased approximately 40 percent as a result of pri- vatization. Labor productivity increased not only because employment decreased but also because privatized firms increased production. The median level of production increased 25 percent because of privatization. Regarding the impact of privatization on investment, all the measures analyzed are positively and significantly affected by privatization. Invest- ment itself increased at least 350 percent as a result of privatization. Finally, we do not find any statistically significant effect of privatization on prices. Nevertheless, after the privatization period, prices did not de- crease, when the efficiency gains we documented indicate that they should have fallen if quality improvements were not large enough. This suggests that there is a pending important regulatory mission to accom- plish in Argentina. Contrary to the case of nonfinancial firms, we do not find large in- creases in operating efficiency after the privatization of public banks. The impact of privatization on (conditional) mean average costs is null. In addition, the impact of privatization on mean administrative expenses is positive and statistically significant. They have increased 36 percent as a result of privatization. However, other indicators of effi- ciency performed better because of privatization. Output per employee increased 20 percent, while the average number of employees per branch decreased 37 percent as a result of privatization. As in the case of the nonfinancial firms, employment cuts are a big part of the story. Thus, on several indicators, privatized banks seem to be more efficient after the privatization than before. The privatization of public banks has also implied an increase in credit supply. Finally, the average capitalization ratio (net worth to assets) increased 5 percent because of privatization. The higher capitalization rate of the privatized banks means a more solvent system, which is quite important in countries as vulnerable to external shocks as Argentina. On the direct measures of welfare analyzed, we find that the privati- zation of water services had a negative and statistically significant effect on child mortality. The estimated coefficient implies a decrease of approximately 5 percent in child mortality rates induced by the privati- zation of water provision. Turning to our estimate of displaced workers' earnings losses, it appears that there is a huge redistributive cost associ- ated with the privatization of SOEs: displaced workers incur substantial earnings losses. The earnings losses due to displacement, after taking into THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 105 account unemployment, are approximately 50 percent of the real earn- ings of workers before privatization. We have identified extraordinary increases in the efficiency of the pri- vatized firms. We have also found that privatization has succeeded in satisfying other important objectives such as restoring investment and enhancing the solvency of the financial system. Finally, we considered some direct impacts of privatization on welfare and found mixed results. Thus, although we found important benefits of privatization, we also found direct costs associated with them. Overall, however, our results are much in favor of privatization. Appendix 2A: Sector Structure and Data Sources Gas Gas del Estado, a vertically integrated monopoly, was privatized in December 1992 and vertically divided into several production companies, two transport companies, and eight distribution companies. The trans- port and distribution companies operate as local monopolies. ENARGAS (Ente Nacional Regulador del Gas) is the national regulatory authority. Competition is allowed only in the market for large users, who can buy gas directly from producers. Financial statements of Gas del Estado were obtained from ENARGAS; financial statements for the private companies were obtained from the Buenos Aires Stock Exchange (if the firms were publicly traded) and from Inspección General de Justicia. Telecommunications ENTEL (Empresa Nacional de Telecomunicaciones), which was trans- ferred to private hands in November 1990, was divided into two new companies, Telecom and Telefónica de Argentina, to provide telecommu- nications services in the northern and southern parts of the country, re- spectively. The companies operated as regional monopolies until 1999, when entry into the long-distance market was deregulated. Entry in local markets was deregulated in 2000. The regulatory authority is the Comisión Nacional de Comunicaciones (CNC). The sources of information were the financial statements of ENTEL obtained from SIGEN (Sindicatura General de la Nacion), the financial statements of Telecom and Telefónica de Argentina, Heymann and Kosacoff (2000), and statistical information from the International Telecommunications Union for 1991 and 2001. No official financial statements were produced for ENTEL for the years 1989 and 1990. The price structure changed several times during the 1990s, making price 106 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER comparisons difficult. We use the telecommunications index of the CPI as our price variable. Electricity In the electricity sector, the largest SOEs were SEGBA (Servicios Eléctri- cos del Gran Buenos Aires), Agua y Energía Eléctrica, and Hidronor (Hidroeléctrica Norpatagónica). SEGBA was the distributor in the Buenos Aires metropolitan area, but it also generated part of the energy it distributed. Agua y Energía Eléctrica and Hidronor were basically generators of electricity. With the privatization, the electrical sector was vertically divided into generation, transport, and distribution. SEGBA was divided into three distributors (Edenor, Edesur, and Edelap) and five generators (Central Puerto, Central Costanera, Central Dock Sud, Central Dique, and Central Pedro de Mendoza). Hidronor and Agua y Energía Eléctrica were divided into 6 transport companies and 22 generators. Competition occurs, mainly, in the generation activities. The sector is now subject to regulation by the secretary of energy, ENRE (Ente Na- cional Regulador de la Electricidad), and CAMMESA (Compañía Administradora del Mercado Mayorista Eléctrico SA). The secretary of energy is responsible for the norms in the sector, ENRE is responsible for the application of these norms, and CAMMESA is responsible for the co- ordination among the different participants in the market (generators, transporters, and distributors). We obtained SEGBA financial statements for the period 1986­91 from SIGEN and Ministerio de Economía, and the financial statements of Edenor, Edesur, Edelap, Central Puerto and Central Costanera (the firms that resulted from SEGBA divestiture) for the period 1992­94 from ENRE. For Hidronor, we obtained two audit statements for 1987 and 1988 and the final financial statement for 1991. We also obtained data for 99.92 percent (in terms of privatization revenues) of the generator companies that emerged from the privatization of Hidronor. We were able to find information for three of the six transportation companies; the three account for 91.7 percent of the privatization revenues of the electricity transportation companies. Our sources of information were Ministerio de Economía, SIGEN, ENRE, Buenos Aires Stock Exchange, and the companies that were willing to collaborate. Water and Sewerage Obras Sanitarias de la Nación was the provider of water and sewerage services in the Buenos Aires metropolitan area and was transferred to private hands in 1992. Aguas Argentinas is the private company that provides these services under a 35-year concession. The regulatory au- thority for the Buenos Aires metropolitan area is Ente Tripartito de THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 107 Obras y Servicios Sanitarios. In other parts of the country water and sewerage services are provided by a large and heterogeneous group of companies (cooperatives, and municipal and provincial companies), sev- eral of which were also privatized. We focus on Obras Sanitarias de la Nación and Aguas Argentinas. The data for these companies were obtained from official financial statements from Ministerio de Economía and SIGEN. Most of the local providers that are or were (before privatization) cooperatives or local companies do not have financial statements. Airlines Aerolíneas Argentinas was privatized in 1990. It operates as an unregu- lated oligopoly in the domestic market and competes in the international market. Information on Aerolíneas Argentinas was obtained from Minis- terio de Economía (before privatization) and the official financial state- ments of the company (after privatization). Railroads Ferrocarriles Argentinos was the SOE that managed the railroad system for the entire country. In a first stage toward privatization, the company was divided into three segments: cargo, urban passengers, and long-distance passengers. The cargo segment was finally divided into five private compa- nies, which obtained 30-year concessions for the payment of an annual canon and a preset investment schedule. These companies are Buenos Aires al Pacífico San Martín SA, Nuevo Central Argentino SA, Ferrocarril Mesopotámico, Ferrosur Roca, and Ferroexpreso Pampeano. In the urban passenger sector, a new company, Ferrocarriles Metro- politanos SA, was created in April 1991 and then divided into seven lines according to the old Ferrocarriles Argentinos lines. One company, Trenes de Buenos Aires SA, operates two lines (Mitre and Sarmiento). Thus, in this segment six companies operate seven railway lines. The private companies have an investment schedule, and they receive an annual subsidy, because it was thought that they would not be able to make profits. The long-distance passenger segment was unprofitable and deemed un- likely to be attractive to the private sector. The federal government instead offered the operation to the provinces interested in maintaining the serv- ice. Only seven provinces accepted this offer, and the rest of the services were discontinued. The regulatory authority is the Comisión Nacional Reguladora del Transporte. The data for the railways companies were obtained from SIGEN (be- fore privatization) and from the official financial statements of the com- panies and Ministerio de Economía (after privatization). 108 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Postal Services The SOE was Empresa Nacional de Correos y Telégrafos, which was trans- formed into Empresa Nacional de Correos y Telégrafos SA in December 1992 and privatized in September 1997. Since December 1997 it has been controlled and operated privately as Correo Argentino SA. This private group has a 30-year concession in return for an annual fee. The regulatory authority is the CNC. The data for postal services were obtained from SIGEN (before priva- tization) and from the official financial statements of the company (after privatization). Oil YPF (Yacimientos Petrolíferos Fiscales) was the SOE sold to the private sector in 1992. Many of YPF's assets, such as tankers, fleet, two refiner- ies, and most of the primary and secondary drilling areas, were sold sepa- rately. The company has operated in an unregulated market since 1991. Repsol of Spain acquired it in 1999. Because so many of YPF's assets were sold separately, any comprehen- sive comparison between the performance of YPF as a public entity and YPF as a private entity is impossible. Bearing this in mind, we were able to provide some comparisons of the public YPF to what remained of YPF in private hands. The data for both the public and private YPF were obtained from official financial statements from Ministerio de Economía and the firm's Web site. Other Two SOEs that were privatized were under the control of Ministerio de Defensa. SOMISA (Sociedad Mixta Siderurgica Argentina) was the main steel manufacturer in Argentina. Between 1991 and 1995, a "tran- sition company," Aceros Paraná, was formed, which was privatized as SIDERAR in 1995. Tandanor (Talleres Navales Dársena Norte), a ship- yard, was transferred to private ownership in December 1991. Both SOMISA and Tandanor operate in unregulated markets. The data found for SOMISA are incomplete, since the only source of information is an audit report. No data were available for the transition company Aceros Paraná. The data for Tandanor were obtained from SIGEN (before privatization) and from Inspección General de Justicia (after privatization). Banks Even though the major economic reforms in Argentina took place during the first half of the 1990s, the banking sector went through an important THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 109 Appendix Table 2B.1 Description of the Variables Used to Evaluate the Impact of Privatization on the Performance of Nonfinancial Firms Variable Description Fixed assets Value of the company's fixed assets adjusted for (property, plant, inflation. PPE is measured by noncurrent assets. and equipment, or PPE) Sales Total value of products and services sold, minus sales returns and discounts. Operating income Sales minus operating expenses, cost of sales, and depreciation. Operating Ratio of operating income to sales. Operating income/sales income is equal to sales minus operating expenses, cost of sales, and depreciation. Sales equal the total value of products and services sold, minus sales returns and discounts. Net income Operating income plus other normal income minus other normal expenses. Note that extraordinary results and income taxes are excluded. Net income/sales Ratio of net income to sales. Net income equals operating income plus other normal income minus other normal expenses. (Note that extraordinary results and income taxes are excluded.) Sales equal the total value of products and services sold, minus sales returns and discounts. Unit costs Ratio of total cost of sales to sales. Cost of sales equals the direct expense involved in the production of a good (or provision of a service). Sales equal the total value of products and services sold, minus sales returns and discounts. Employment Total number of employees. Employees are taken to be all workers who depend directly on the company. Wages per worker Ratio of the total wage schedule paid by the firm to the total number of workers who depend directly on the company. Sales/employment Ratio of sales to total employment. Sales equal the total value of products and services sold, minus sales returns and discounts. Employment takes into account all workers who depend directly on the company. Operating income/ Ratio of operating income to total employment. employment Operating income is equal to sales minus (Table continues on the following page.) 110 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Appendix Table 2B.1 (continued) Variable Description operating expenses, minus cost of sales, and minus depreciation. Employment takes into account all workers who depend directly on the company. Prices In most cases, ratio of sales to physical output. For multiproduct firms or firms where prices are two- or three-part tariffs, the variable equals the price index of the product constructed by INDEC (Instituto Nacional de Estadísticas y Censos). This latter definition applies only to Obras Sanitarias de la Nación, Entel, Encotel, SOMISA, and YPF. Production Total output of the firm. For some multiproduct firms, this variable is the ratio of total sales to prices, where prices are constructed by INDEC (Instituto Nacional de Estadísticas y Censos). This latter definition applies only to Obras Sanitarias de la Nación, Entel, Encotel, SOMISA, and YPF. Production/ Ratio of production to total employment. employment Production equals total output of the firm. For some multiproduct firms, this variable is the ratio of total sales to prices, where prices are constructed by INDEC (Instituto Nacional de Estadísticas y Censos). This latter definition applies only to Obras Sanitarias de la Nación, Entel, Encotel, SOMISA, and YPF. Employment takes into account all workers who depend directly on the company. Investment Value of expenditure to acquire property, equipment, and other capital assets that produce revenue (gross investment). Investment/sales Ratio of investment to sales. Investment equals the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. Sales equal the total value of products and services sold, minus sales returns and discounts. Investment/ Ratio of investment to total employment. employment Investment equals the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. Employment takes into account all workers who depend directly on the company. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 111 Appendix Table 2B.2 Description of the Variables Used to Evaluate the Impact of Privatization on the Performance of Financial Firms Variable Description ROA Return on assets. Ratio of net income to assets. Net income equals operating income plus other normal income minus other normal expenses. (Note that extraordinary results and income taxes are excluded.) Assets equal the sum of cash holdings ($ US$), public titles, loans, participation in other societies, fixed assets, other assets, intangible assets, and foreign subsidiaries. ROE Return on equity. Ratio of net income to net worth. Net income equals operating income plus other normal income minus other normal expenses. (Note that extraordinary results and income taxes are excluded.) Net worth equals total assets minus total liabilities. Profit margin Ratio of net income to total revenue. Net income equals operating income plus other normal income minus other normal expenses. (Note that extraordinary results and income taxes are excluded.) Total revenue is equal to financial income plus service income plus irrecoverable charges. Operating Ratio of financial income plus irrecoverable charges plus margin services income plus financial expenditures plus services expenditures plus administrative expenses to financial income plus irrecoverable charges plus other income. Interest Ratio of financial incomes plus financial expenditures margin plus irrecoverable charges to loans plus public titles. Branches Number of branches per institution. Employees Number of employees per institution. Output Cash holdings ($ US$) plus public titles plus loans plus deposits. Average cost Ratio of administrative expenses to output. Administrative expenses include wages, tax payments, and asset depreciation. Output equals the sum of cash holdings ($ US$), public titles, loans, and deposits. Capitalization Ratio of capital to assets. Assets equal the sum of cash holdings ($ US$), public titles, loans, participation in other societies, fixed assets, other assets, intangible assets, and foreign subsidiaries. Operating Operating income equals the sum of financial incomes, income irrecoverable charges, and service income. Loan growth Growth in total loans, calculated as the logarithm of loans in year t minus the logarithm of loans in year t 1. 112 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER transformation after the Tequila crisis of December 1994. From the 35 public banks that started the decade, 19 were privatized between 1992 and 1999, 2 were merged with privatized banks, and 14 banks continued to be owned by the public sector. We analyze the privatization of 17 banks. The data set we used was provided by BCRA (the central bank) and contains monthly financial information of all the entities that operated in the Argentine Financial System for the period June 1993 through September 1999. It includes the basic balance sheet accounts, income structure, and some physical data such as the number of employees and branches. Although the data set covers the period in which almost all the bank pri- vatizations took place, not all the information is available for every mo- ment. Particularly, more disaggregated and better quality data are available only for more recent periods. Notes The authors are grateful for the comments of Alberto Chong, Florencio López-de-Silanes, and Máximo Torero. Matias Cattaneo, María Eugenia Gari- botti, Hernán Moscosco, Mariano Tappata, and German Sturzenegger provided excellent research assistance. 1. Although several studies describe the privatization process in Argentina, none of them attempts to identify the causal effects of privatization on broad meas- ures of performance (see, for example, Gerchunoff 1992; FIEL 1999; and Galiani and Petrecolla 2000). 2. The effects of the numerous accusations of corruption associated with the privatization of the Argentine public firms are also excluded from our analysis. 3. The only exception was Ferrocarriles Argentinos, the railway public enter- prise. The company was divided into 11 units (operating lines or corridors) given in concession, and it was possible to find data by business unit for the preprivati- zation period. 4. Income from concessions is not considered in these calculations. Aguas Argentinas and the railway companies were privatized under this legal form. Information on total revenues from concessions is not available. 5. Possibly including a set of control regressors that vary across both units and time. 6. La Porta and López-de-Silanes (1999) use a difference-in-difference estima- tor to isolate the contribution of privatization to firm performance. As a control group, they form industry control groups (three-digit Standard Industrial Classifi- cation, or SIC, code level) using all private firms trading in the Mexican Stock Mar- ket (they used economywide aggregates where no matching firms were found). In our case, since most of the privatizations are the only firm in the sector, we have to rely on the nonprivatized firms to assess the counterfactual scenario of the priva- tized ones. Thus, our estimates of equations 2.1 and 2.2 are the closest we can get to their mean and median sector-adjusted estimates. 7. As explained before, some small privatized firms are not included in our data set. However, even if we had information on those small privatizations, the appropriateness of pooling small and large firms in our econometric analysis would be arguable. 8. Naturally, the number of observations included in the control group each year decreases every time a firm is privatized. Nevertheless, it is worth noting that THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 113 if the statistical model specified to identify the impact of privatization on the random variable y is correct, this is not a nuisance. 9. For several variables, however, the data are not that reliable because of the severe difficulties of producing consistent balance sheet accounts in periods of extreme price instability. We have detected some outliers in 1989. In addition, other outliers are dispersed across the data set. We do not exclude them from the regressions. Instead, as we explain later in the chapter, we present the results from a median regression, which are robust to outliers in the dependent variable. 10. We do not attempt to measure directly the impact of privatization on the stock of fixed assets because of the severe measurement errors in this variable already discussed. 11. In this context, robust connotes a certain flexibility of the statistical proce- dures to deviations from the distributional assumptions of the hypothesized mod- els (see Koenker and Bassett 1978). 12. The percentage change of any variable y with respect to privatization is given by 100 [Exp( ) 1], where is the estimated coefficient in the regression equation 2.1 or 2.2 when the dependent variable is the natural logarithm of y. 13. For example, in the case of YPF, a random sample of laid-off workers shows that in 1991, just before the privatization of the firm in 1993, the mean (me- dian) age of these workers was 43 (43) years while the mean (median) age of the employees in the manufacturing sector was 34 (33) years (household survey, all ur- ban agglomerates). Only 10 percent of the laid-off workers from YPF were younger than 30 years, while 43 percent of the employees in the manufacturing sector were younger than 30 years in 1991. 14. At least 60 percent of the estimated (conditional) median impact of the pri- vatizations on real wages may be explained by the change in the average age of the workers if the data from YPF are representative of all privatizations. We estimated an earnings function using wage data from the random sample of displaced work- ers from YPF for the year 1991 and computed the implied decrease in average real wages as the result of the estimated change in the average age of the workers of YPF after privatization. 15. It is still statistically different from 0 at the 1 percent level of significance. 16. This statistic may overestimate the contribution of layoffs to profits be- cause it assumes that the laid-off workers were completely unproductive. 17. For example, our results on the effect of privatization on child mortality, discussed later in the chapter, may come from a mix of better access and improved water quality. 18. FIEL (1999) also finds that most of the real prices of the goods and services provided by the former SOEs did not increase during the 1990s even though most of them were increased at the beginning of the 1990s. Nevertheless, the prices of the goods and services of most privatized firms are indexed by the U.S. consumer price index. This implies that the prices of the privatized firms could have increased 18.5 percent since 1995 with respect to the domestic CPI. However, these changes in prices are not identified as a result of the privatization even though they are caused by the regulatory framework. Clearly, this regulatory pricing policy is in- consistent with a fixed exchange rate policy like the one adopted by Argentina dur- ing the 1990s. 19. The changes in the interest margins are mainly caused by the reduction in the share of nonperforming loans after the privatization. When we perform a test of mean differences on the interest margin without netting the nonperforming loans, we reject the hypothesis of equal means only at the 10 percent significance level. This improvement in the loan performance after privatization could result be- cause the private banks managed credit decisions better or because the "bad loans" were placed in the residual entity. In the latter case, again, the bank fixed-effect as- sumption in the before-and-after estimator breaks down. 114 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER 20. The first potable water service of Argentina was provided by Obras Sani- tarias de la Nación in 1870. It initially served 30,000 people, and coverage contin- ued to expand until 1960. In the 1980s coverage as a share of population actually contracted. OSN's jurisdiction was nationwide until 1980. At that point, it was re- stricted to Capital Federal and 13 localities of Greater Buenos Aires. Responsibil- ity for service in the rest of the country was transferred to provincial governments (Artana, Navajas, and Urbiztondo 1999). 21. According to our estimates, the served population increased 4.8 percent a year, while population itself increased approximately 1.7 percent a year. 22. Most privatizations also included the provision of sanitation services. 23. Data on household income and inequality (the ratio of the top 10 percent to the bottom 10 percent of the distribution of per capita family income) are ob- tained from the Permanent Household Survey. 24. By rent, we mean the difference between what an employed worker can get from his/her employment relationship and his/her outside option. 25. Individuals were asked to consider the overall impact of displacement from YPF on their welfare, taking into account that they received a severance payment at the time of displacement. Approximately 80 percent of the sample reported that the major benefit lost was the health insurance package. Of the displaced workers who were employed at the time of the survey, half considered that they were in a permanent (stable) job, while 37 percent thought their job was transitory. The remaining 13 percent were unsure how to characterize the stability of their current employment. 26. The female labor force participation rate is 70 percent, although the sam- ple included only 33 females younger than 60 years. 27. Alternatively, we could compare the statistics obtained from the survey conducted to the displaced workers from YPF with the same statistics estimated from the May 2001 wave of the household survey using data from the urban ag- glomerates of Chubut, Greater Buenos Aires, La Plata, Mendoza, Neuquen, and Salta. These regions match geographically the places where our sample of displaced workers resides. 28. The data gathered at the beginning of May 2001 and at the beginning of August 2001 are perfectly comparable. It was only after October 2001 that the level of economic activity in Argentina imploded. 29. These differences are even greater for skilled workers (those with at least some college or a tertiary degree). The unemployment rate among male unskilled (at most high school) workers displaced from YPF is 28 percent, while the male skilled unemployment rate in the same sample is 20 percent. The same statistics in the control group are, respectively, 15.8 and 6.4 percent. 30. Remember that the percentage change of y with respect to privatization is given by 100 [Exp( ) 1]. References Artana, D., F. Navajas, and S. Urbiztondo. 1999. "Governance and Regulation in Argentina." In W. Savedoff and P. Spiller, eds., Spilled Water. Washington, D.C.: Inter-American Development Bank. Burdisso, T., L. D'Amato, and A. Molinari. 1998. "Privatización de Bancos en la Argentina: ¿Un Camino hacia una Banca más Eficiente?" Documento de Tra- bajo 4. Banco Central de la República Argentina, Buenos Aires. Caves, Richard. 1990. "Lessons from Privatization in Britain: State Enterprise Behavior, Public Choice, and Corporate Governance." Journal of Economic Behavior and Organization 13: 145­69. THE BENEFITS AND COSTS OF PRIVATIZATION IN ARGENTINA 115 CEP (Centro de Estudios Públicos). 1998. "Privatizaciones: Un Balance Cuantita- tivo." Ministerio de Economía, Buenos Aires (www.minproduccion.gov.ar). Chamberlain, G. 1984. "Panel Data." In Zvi Griliches and M. Intriligator, eds., Handbook of Econometrics. Amsterdam: North-Holland. Esrey, S. A., J. B. Potash, L. Roberts, and C. Shiff. 1991. "Effects of Improved Wa- ter Supply and Sanitation on Ascariasis, Diarrhoea, Dracunculiasis, Hookworm Infection, Schistosomiasis and Trachoma." Bulletin of the World Health Orga- nization 69: 609­21. FIEL (Fundación de Investigaciones Económicas Latinoamericanas). 1999. La Reg- ulación de la Competencia y de los Servicios Públicos. Buenos Aires. Galiani, Sebastián, and S. Nickell. 1999. "Unemployment in Argentina in the 1990s." Working Paper DTE 219. Instituto Torcuato Di Tella, Buenos Aires. Galiani, Sebastián, and D. Petrecolla. 1996. "The Changing Role of the Public Sec- tor: An Ex-post View of the Privatization Process in Argentina." Quarterly Re- view of Economics and Finance 36: 131­52. ------. 2000. "The Argentine Privatization Process and Its Aftermath: Some Pre- liminary Conclusions." In Melissa Birch and Jerry Haar, eds., The Impact of Privatization in the Americas. Boulder, Colo.: North-South Center Press. Gerchunoff P., ed. 1992. Las Privatizaciones en Argentina--Primera etapa. Buenos Aires: Instituto Torcuato Di Tella. Hall, B. 1995. "Lost Jobs." Brookings Papers on Economic Activity, pp. 221­73. Washington, D.C.: Brookings Institution. Hamermesh, D. 1989. "What Do We Know about Worker Displacement in the United States." Industrial Relations 28: 51­59. Heckman, J., and R. Robb. 1985. "Alternative Methods for Evaluating the Impact of Interventions: An Overview." Journal of Econometrics 30: 239­67. Heckman, J., H. Ichimura, and P. Todd. 1997. "Matching as an Econometric Evaluation Estimator: Evidence from Evaluating a Job Training Programme." Review of Economic Studies 64: 605­54. Heymann D., and B. Kosacoff. 2000. La Argentina de los Noventa­Tomo II­ Desempeño Económico en un Entorno de Reformas. Buenos Aires: Editorial Universitaria de Buenos Aires. Jacobson, L., R. LaLonde, and D. Sullivan. 1993. "Earnings Losses of Displaced Workers." American Economic Review 83: 685­709. Koenker, R., and G. Basset. 1978. "Regression Quantiles." Econometrica 46: 33­50. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." Quarterly Journal of Economics 114: 1193­242. Lustig, Nora. 1992. Mexico: The Remaking of an Economy. Washington, D.C.: Brookings Institution. Megginson, William, Robert Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An Interna- tional Empirical Analysis." Journal of Finance 49: 403­52. Ministerio de Economía, Government of Argentina. 2000. "El Proceso de Privatiza- ciones en la Argentina desde una Perspectiva del Balance de Pagos." Buenos Aires. Ruhm, C. 1991. "Are Workers Permanently Scared by Job Displacement?" Amer- ican Economic Review 81: 319­24. Shirley, M. 2000. "Reforming Urban Water Systems: A Tale of Four Cities." In L. Manzetti, ed., Regulatory Policy in Latin America: Post-Privatization Reali- ties. Miami: North-South Center Press. 116 GALIANI, GERTLER, SCHARGRODSKY, AND STURZENEGGER Shleifer, Andrei, and Lawrence Summers. 1988. "Breach of Trust in Hostile Takeovers." In Alan Auerbach, ed., Corporate Takeovers: Causes and Conse- quences. Chicago: University of Chicago Press. Vining, Aidan R., and Anthony Boardman. 1992. "Ownership vs. Competition: Efficiency in Public Enterprises." Public Choice 73: 205­39. WHO (World Health Organization). 2000. Global Water Supply and Sanitation Assessment 2000 Report. Geneva. 3 Privatization and Firm Performance in Bolivia Katherina Capra, Alberto Chong, Mauricio Garrón, Florencio López-de-Silanes, and Carlos Machicado FOLLOWING THE EXAMPLE SET BY THE Thatcher government in Britain in the early 1980s, the transfer of state-owned enterprises (SOEs) to the private sector has become widespread. By the end of the 1990s, more than 100 countries had adopted some kind of privatization program (Megginson and Netter 2001). The wide acceptance of this policy in Latin America re- sulted from a recognition that the state failed to deliver on its promises more often than not and that more efficient alternatives were available. In particular, the acceptance of market mechanisms meant the abandonment of the import substitution paradigm that had been the region's defining policy during the previous decades. In recent years, however, privatization has come under severe criticism, as the benefits of the program have often gone unnoticed, while its failures have been prominently publicized. Critics argue that the benefits of priva- tization are exaggerated; that when they do arise, they result from ex- ploitation of workers and consumers by the private sector; and that it is always the poor or the government that foots the bill for misconceived pri- vatization schemes. Given the rising backlash against privatization in Bolivia, as well as most of the rest of Latin America, a systematic analysis of the evidence is in order to ensure that the right conclusions are reached and that appropriate decisions are made regarding whether privatization should be rolled back, as the critics propose, or instead be expanded and deepened. Bolivia, like many other Latin American countries, undertook privati- zation with enthusiasm. Measured by both the number of firms sold and their average size, Bolivia's privatization program is small by international 117 118 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO standards but important relative to its economy. During the 1990s Bolivia received privatization revenues equivalent to approximately 12 percent of its 1999 gross domestic product (GDP)--a greater share than Argentina, Brazil, or Mexico (see chapter 1). Our sample of Bolivian firms reflects the mostly small and medium state-owned enterprises that were sold to the private sector and that may be thought to react differently to privatization than the typically large firms privatized in other countries. In addition to shedding light on the Bolivian privatization program, our study may serve as a guide for other countries seeking to privatize small- or medium-size firms. The remainder of this chapter is structured as follows. The next sec- tion presents a review of the privatization conceptual framework, as well as the different arguments for and against privatization. We also review some of the recent literature and survey the existing studies for privatiza- tion in Bolivia. We then describe the Bolivian privatization program and classify it into three waves, which coincide with three different adminis- trations under which the programs were undertaken. Next, we describe our data sample and the efforts conducted to collect information. We dis- cuss the firms considered in the study, as well as the reasons for the exclusion of the remaining observations. We then present our main results regarding the raw and industry-adjusted performance effects of privatization. Finally, we conclude with the policy implications that arise from our main findings. Privatization: Conceptual Framework This section provides a brief overview of the most common goals pursued through privatization, a description of the theoretical framework that underpins most of the recent research, and an overview of the existing studies analyzing the effects of privatization with a special emphasis on Bolivia. The Objectives Pursued with Privatization Privatization can be undertaken for many reasons. As shown in table 3.1, it can result from economic considerations (with a micro- or macroeconomic emphasis), political reasons, firm-related considerations, or a desire to improve consumer welfare.1 Although many of the goals listed in table 3.1 are interrelated, in some cases tradeoffs have to be made. In a first attempt to evaluate the Bolivian privatization program, we chose to follow the footsteps of other recent studies and assume that the main objective behind the program was to improve the efficient use of resources by privatized firms (see, among many others, Wiltshire 1987, Moore 1990, and Nellis 1991). PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 119 Table 3.1 Common Objectives for Privatizations Economic Goals Achieve higher productivity and efficiency Strengthen the role of the private sector in the economy Improve the public sector's financial health Provide autonomy to satisfy financing requirements Political Goals Release the enterprise from political interference Free resources for allocation in other priority areas Make it possible for employees to participate as shareholders Firm-Oriented Goals Improve performance of specific firms Consumer-Oriented Goals Improve services and goods Lower prices and improve quality Source: Authors. The Arguments for and against Privatization In recent years developments in the field of contract property theory have brought researchers' attention to the debate regarding public versus pri- vate provision of goods and services (Shleifer and Vishny 1996; Shleifer 1998). According to contract theory, the government could draw up an agreement, which includes all possible contingencies, so that the contrac- tor would always deliver the expected results. Under these assumptions, it is completely irrelevant whether a good or service is provided by a public or a private entity. However, once we consider the costs in both time and money involved in drawing up a contract that tries to foresee all possible contingencies and once we acknowledge that enforcement of contracts is costly, the applicability of the predictions of this model seems unlikely. In response to the incompleteness of contracts, property theory argues that asset ownership is the foundation for control and thus determines the incentive structure that can explain the performance of firms and man- agers.2 Within this framework, two lines of argument, the social view and the agency view, emerge to explain why a change in ownership can lead to shifts in the performance of privatized firms. The social view argues that SOEs play an important role in counteracting market failures by imple- menting pricing policies that take social marginal costs into account. Thus, the increased profitability of privatized firms may reflect not only effi- ciency gains but also consumer exploitation through the use of market power (Shapiro and Willig 1990). The agency view, in contrast, argues that the benefits of privatization stem from an ownership structure that provides adequate incentives. It 120 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO begins by acknowledging that firms under state control pursue political, social, and economic objectives simultaneously and are thus almost guar- anteed to perform at a lower level than firms in the private sector (Sheshinski and López-Calva 1999). Within this second view, there are two complementary explanations about the inadequate incentive structure for SOEs, depending on who the critical agency conflict lies with--the manager or the politician. The managerial view argues that managers of SOEs may lack high-powered incentives or the necessary monitoring authority to ensure that the firm is run effectively (Vickers and Yarrow 1988). The political perspective stresses that when politicians have control over a firm, they are likely to pursue a multitude of objectives and relegate efficiency considerations to a secondary role (Shleifer and Vishny 1994; López-de-Silanes, Shleifer, and Vishny 1997). The inclusion of political and social goals is likely to diminish the pursuit of economic ones. Also, the frequent changing of the relative importance of these priorities is likely to affect negatively the long-term performance of SOEs. In this sense, privatiza- tion can promote efficiency by allowing the objectives of the enterprise to be more focused and thus reducing the ambiguity attached to these objectives, and by facilitating long-term planning (Commander and Killick 1988). Empirical Studies on Privatization Most empirical studies can be classified into three broad groups: case stud- ies, intercountry studies, and cross-section studies. The first group in- cludes studies such as Ramamurti (1997) and D`Souza (1998), who find that privatization increased productivity and reduced the work force in specific sectors (railroads and telecommunications, respectively); Eckel, Eckel, and Singal (1997), who document significant price decreases in the airline sector following the privatization of British Airways; and New- berry and Pollitt (1997), who find that the producers and shareholders benefit the most from privatization, while consumers and the government benefit only marginally. Among the most prominent cross-section studies are those by Galal and others (1994), Megginson, Nash, and van Randenborgh (1994), Frydman and others (1998), and Boubakri and Cosset (1998). All these studies find that the performance of privatized companies significantly improved. However, another study by Pohl, Claessens, and Djankov (1997), using a sample of 6,300 firms, shows that privatization's positive effects on cor- porate performance are not uniform among different types of firms and performance measures. Finally, among the most prominent examples of intercountry studies are Martín and Parker (1995); Barberis and others (1996); Claessens and Djankov (1998); and La Porta and López-de-Silanes (1999). The major PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 121 findings of these studies are that privatized firms improved their perfor- mance, that the change in management contributed to an increase in the value of the firms, and that the greater business profitability was largely explained by improvements in productivity and not by higher sale prices or reductions in labor costs. Relatively little has been written on Bolivia's privatization program, especially regarding the first privatization phase from 1992 to 1994. Some recent studies provide valuable information for understanding the privatization program in Bolivia; among these are Montero (1994), Re- ordenamiento de las Empresas Públicas (1994), and Requena (1996). Their contributions, however, are limited to a descriptive analysis of the privatization process or a useful, albeit partial, analysis of the financial performance of former SOEs. Another important study is that of Barja and Urquiola (2001), who analyze the impact of capitalizations and the regulatory reforms implemented in conjunction with privatization in or- der to assess their impact on access and availability of services to low- income households. Deeper studies of Bolivia's privatization usually stall because of the scarcity of information, mainly financial data, regarding firms' situation before and after the process. This is precisely the gap this chapter tries to fill by collecting comparable information from a wide sample of privatized firms during the period 1992­2000. The Bolivian Privatization Program The privatization program was very successful in attracting foreign in- vestment, which reached levels heretofore unseen in Bolivia. The increase in foreign investment came primarily from capital investments in gas, elec- tricity, and telecommunications. These increases, together with reforms in the financial sector, caused second-round increases in private domestic in- vestment and consumption. This boom in investment drove economic growth throughout the mid-1990s to the late 1990s. The increased social investment can be understood only in the context of the state reform in which it was implemented. The method of invest- ment delivery changed from an overwhelmingly centralized provision of services to a provision based on municipalities. This reform was imple- mented by a wide-ranging decentralization reform, enacted as part of the Ley de Participación Popular and the Ley de Decentralizacion. The motivation behind this reform was to make expenditures in social in- vestment more efficient by better matching expenditures to local needs. The underlying hypothesis was that communities are better able than the central government to assess their needs and select how best to make investments that address those needs. From its inception the decentral- ization was supported by the Inter-American Development Bank and the 122 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO World Bank, both through credit and lending activities and, more gen- erally, in the context of the Heavily Indebted Poor Countries Initiative and the corresponding Bolivian Poverty Reduction Strategy Paper. The Inter-American Development Bank in particular focused on loans to strengthen municipalities and their ability to plan and execute invest- ment projects. Before 1992 the government sought, through the use of shared risk con- tracts, an increased private involvement in certain sectors (such as mining and hydrocarbons) that had traditionally been reserved for the public sec- tor. These contracts typically transferred assets of SOEs to private admin- istrators, who would then have to undertake the necessary investments to modernize the business or complete the project. Within this framework, the state received a flat fee for the leased assets, and the private firm could claim any residual profits. The first contracts of this nature were signed during the 1980s and are still used today. Although the size of many of these agreements makes them an important object of study, they reflect a qualitatively different phenomenon than the privatizations that would later take place. In 1994 the Bolivian government decided to privatize its state-owned enterprises, which controlled seven industries: oil and gas, petroleum refining, tin mines, railways, electrical power, telephones, and airlines. In some instances, the government used traditional privatization; under this method, the government transfers majority ownership of an SOE to the private sector and has freedom over how to spend the proceeds. More often, however, the Bolivian government relied on capitalization. Under capitalization, the state transfers 50 percent of a company's shares to the investor with the winning bid. It transfers an additional 45 percent of the shares to a private pension fund for the accrued benefit of the public in general, with the remaining 5 percent going to the company employees. The investor takes over the management of the firm and commits to invest the amount it offered to acquire its 50 percent share, in the development of the firm. It must carry out this investment within a specific time period (typically 6­8 years). In addition, the investor agrees to fulfill obligations that encompass expansion and quality goals, to operate under tariff regu- lation, and to fulfill other clauses specified in a long-term contract (typi- cally 40 years). This method of privatization was selected by the center-right political faction led by Gonzalo Sanchez de Lozada because previous privatizations had been strongly criticized. This plan attempted to elicit popular support by distributing shares to the electorate; however, not everyone was happy with the plan. Some said it was unfair to those younger than 21 years of age, who were left out of the distribution. Others objected that their shares went into a pension fund that they would benefit from only in the future or maybe not at all if they died before retirement. Others objected to the sale of strategic industries to foreigners. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 123 The period analyzed here starts in1992 and ends in the early 2000s. It can be broken into three privatization waves, which are associated with different governments and reflect the major pushes given to privatization in Bolivia.3 The first wave started in 1992 with the enactment of the Pri- vatization Law.4 The second wave began in 1994 and introduced the cap- italization schemes. And the third wave started in 1998 and constituted a return to the straight sales seen during the first wave. First Wave With the enactment of the Privatization Law in April 1992, Bolivia began its privatization process. The process was characterized by the complete transfer of companies, which operated in competitive markets and were considered "small" businesses. Under this wave, 34 SOEs were privatized. All were considered to be nonperforming before their sale, they had ample excess capacity and bloated work forces, and most received subsidies from the government (Ministerio de Capitalización 1994). A list of these 34 SOEs can be found in appendix table 3A.1 and includes, for the most part, hotels, bus terminals, milk-processing plants, flour-collection businesses, and similar enterprises. Enterprises transferred to the private sector in this first wave were sold through public auctions and public bidding processes. In a few cases workers were offered the option of purchasing shares in the busi- ness, but this was the exception rather than the rule. Outside consultants were used to help assess the value of the largest enterprises; however, several firms were so small that no proper valuation was carried out. The objectives were to reduce the public deficit by cutting subsidies; increase the efficient use of assets by transferring them to the private sector; and increase social investment in health, education, and basic infrastructure by directing the proceeds of the privatization to these ends. Most firms were sold without any prior restructuring and with the government paying, in advance, for any social benefits owed to existing workers.5 Second Wave The second privatization wave began in 1994, when capitalization was adopted as the method of privatization.6 Capitalization contracts differed from the way in which SOEs had been privatized previously in two im- portant ways. First, the private investor acquired complete managerial control but only 50 percent of the equity. Second, in contrast to a tradi- tional privatization, which involves the sale of the enterprise, no sale was involved and, hence, the national treasury did not receive any proceeds from the transfer. What capitalization called for was an injection of capi- tal to the enterprises under consideration. Thus, through this process, the 124 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO equity of the enterprise doubled, the private investor held 50 percent of the equity in the capitalized enterprise, and the state transferred the other 50 percent to company employees and as nontransferable shares to two newly established private pension funds. Capitalizations were used for the largest SOEs, which had traditionally been considered strategic, and often resulted in the breakdown of state monopolies. For example, the state hydrocarbon enterprise, Yacimientos Petrolíferos Fiscales Bolivianos (YPFB), was capitalized as several differ- ent units. Similarly, the generating units of the national power company Empresa Nacional de Electricidad (ENDE) and two units (Andina and Oriental) of the railroad enterprise Empresa Nacional de Ferrocarriles (ENFE), were capitalized independently. At the same time, other state mo- nopolies were capitalized as a single entity, such as the telecommunica- tions enterprise, Empresa Nacional de Telecomunicaciones (ENTEL), and the national airline, Lloyd Aéreo Boliviano (LAB). Appendix table 3A.1 lists the 10 capitalizations carried out during the second wave, as well as 35 traditional privatizations. The idea behind the capitalizations was to encourage private firms to invest in and run what had been key SOEs. The government's main objec- tives were to attract private investment on a large scale; accelerate job cre- ation; improve managerial and technological efficiency by transferring control to the private sector; and create a long-term saving mechanism to directly redistribute the gains from the privatization into pensions instead of using the gains for public spending. To make the enterprises more at- tractive for bidders, the commercial debts of capitalized firms were often transferred to the national treasury. As evident from the list above, capitalizations were used mainly for firms in noncompetitive markets and especially for those involved in the provision of public services. These firms also employed the largest num- ber of workers and consequently were home to the most active labor unions. According to a study undertaken by the Ministerio de Capital- ización (1994), before the capitalization of SOEs, all of these enterprises had productivity and efficiency indicators below international bench- marks. Third Wave The third wave saw an end to the capitalization program and a return to the classical approach to privatization. From 1998 to 2003, the third wave included the sale of 14 SOEs including a petroleum refinery, sev- eral mining firms, a cement factory, and other assorted enterprises (see appendix table 3A.1). Because of the nature of the firms sold, the third wave also included significant regulation and deregulation of relevant sectors to ensure that consumers would not be exploited by firms with market power. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 125 Reforms in Specific Sectors Capitalization was introduced relatively late in Bolivia's road to liberal- ization and was seen not as a means to cover deficits but rather as an option for attracting foreign investment and improving management in key areas of the economy. Indeed, this process raised significant amounts of capital. Total commitments were about $2 billion, roughly equivalent to 30 percent of GDP, thus contributing to a significant increase in in- vestment.7 This process was complemented with changes to some sec- tors' industrial organization and with the implementation of a regulatory framework seeking to promote competition and efficiency. The main tool in this regard was the Sistema de Regulación Sectorial (SIRESE) Law in 1994, which created a regulatory system for the entire infrastructure sector. In essence, this legislation defined the regulatory institutional structure, including the role of five regulatory agencies (superintenden- cias) for the electricity, telecommunications, hydrocarbons, potable water, and transportation sectors. It also established an oversight agency responsible for systemwide coordination and evaluation, and introduced market competition as one of the foundations of the infrastructure sec- tor. Last, the law formulated the procedures for appeals, hearings, and conflict resolution. This framework is supplemented by four specific laws--covering elec- tricity (1994), telecommunications (1995), hydrocarbons (1996), and potable water (2000)--that introduced changes in the organization of each sector. These laws govern issues related to tariff regulation, entry, service quality, and sanctions and are administered by the sectoral regula- tory agencies of SIRESE. Electricity. The reform of Bolivia's electricity sector is considered one of the most successful to date. Bolivia opted for both vertical and horizontal sep- arations. ENDE, the largest power company, was divided into three gener- ation units and a transmission grid. The generation units were capitalized, and the transmission grid was privatized as a common carrier. In distribu- tion, the Empresa de Luz y Fuerza Electrica de Cochabamba (ELFEC, the Cochabamba electricity distribution company) was also privatized, while the Compania Boliviana de Energia Electrica (COBEE, the second largest generator and distribution company in La Paz) was already private. Among the other major cities, the Cooperativa Rural de Electrificacion (CRE, the Santa Cruz distribution company) remains a cooperative and is considered to be well run along commercial lines. Many of the smaller electric systems, which are cooperatively owned, have yet to be corporatized (that is, to be- come sociedades anonimas). In reforming its electricity industry, Bolivia broke away from some of the accepted models for reform prevailing at that time and became the first country to restructure and privatize the power sector using the capitaliza- tion method. This decision was prompted not by a desire to flout current 126 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO fashionable practice, but rather because it suited the country that way and because the resulting reform program constituted an appropriate and timely response to the situation in Bolivia at that time. Instead of copying other countries' models, Bolivia wisely adopted some of the principles un- derlying current reform trends all over the world and applied them in the degree and form that best suited its own position. By adopting the capitalization approach, Bolivia not only privatized electricity supply but also transferred one-half of the industry's ownership to Bolivian citizens. The shares allocated to Bolivians were entrusted to special pension funds, which were responsible for managing the shares within certain trading parameters. This process developed both the pen- sion funds and the local capital markets. It also ensured that capitalization took place through timely investments that were equal in size to the con- tributions made by private investors. This approach differs from the pre- vious attempts at privatization in the United Kingdom (England and Wales) and Chile. In fact, the Bolivian case might be labeled as a "reverse Chilean" process because the pension funds were used to manage the shares rather than to buy them. Hydrocarbons. Before the reforms, the hydrocarbons industry (oil and natural gas) was dominated by Yacimientos Petrolíferos Fiscales Boli- vianos, a vertically integrated monopoly involved in all activities of the industry. Limited private participation was possible through joint ventures with YPFB. Since the reform, the governmental priority has been to re- move YPFB from production activities and to promote foreign investment to foster a natural gas export industry directed mainly to southern Brazil. The state intends this industry to become the engine of development for other sectors of the national economy. With this goal in mind, reforms and foreign investment were targeted to exploration and the creation of new infrastructure. The opening of a pipeline to Brazil in 1999 made this vision a reality. A general policy promoting private control of all phases of hydrocarbons, including retail commercialization, was adopted for the do- mestic market. To implement these objectives, the Hydrocarbons Law places few restrictions on the export and import of petroleum products and stipu- lates that exploration, production, and commercialization be executed through joint ventures with YPFB. The administration of gas and oil pipelines was transferred, without exclusive rights, to the capitalized Transredes. The administration of other pipelines was entrusted to the private Oil Tanking. Most of YPFB's refining units were transferred to the private Empresa Boliviana de Refinación, while YPFB continues to run the wholesale operations for petroleum products. Telecommunications. In the preprivatization era, the telecommunications industry was divided among the state monopoly, Empresa Nacional de Tele- comunicaciones, which provided national and international long-distance PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 127 communication services, 15 cooperatives with monopolies in fixed local telephone services, and Telecel, a private monopoly in the cellular market. The Telecommunications Law maintained this division until the end of 2001, when entry was liberalized. Until then, ENTEL and the cooperatives had exclusive rights, but the cellular market was opened to competition with the entry of ENTELMovil (a division of capitalized ENTEL). The Telecom- munications Law also encompassed incentives for the exploitation of economies of scope by the most efficient firms. This objective is pursued through two mechanisms: cooperatives failing to achieve improvement goals lose a percentage of their market to ENTEL; and authorization for mergers, acquisitions, and stock swaps. The entry of Nuevatel, a joint venture between COMTECO (the Cochabamba cooperative) and Western Wireless International, has been the only modification to this industrial structure. Nuevatel was created for the purpose of acquiring a PCS (personal communications services) license and began operations in December 2000, intensifying competition in the mobile phone market. Water. While other sectors underwent capitalization and the introduc- tion of regulation, the water industry experienced limited changes and encountered several difficulties. The intention of the reform in this sec- tor was the creation of several concessions (as opposed to actual privati- zation) for the administration of state assets. In practice, however, only one municipal firm, SAMAPA (La Paz-El Alto), was transferred to the private sector in 1997 for administration by Aguas del Illimani, a private firm. It was expected that within a prudent period of time, the necessary legislation would be in place to conform the remaining public water firms to a similar model. However, the long delay in formulating the Potable Water and Sewerage Law (finally approved in 2000), together with a significant failure to create a second concession through the trans- fer of a municipal firm to a consortium, Aguas del Tunari, in Cochabamba, has deferred reforms and somewhat redirected change in this sector. Nevertheless, during 1998 and 1999, the water regulatory agency was able to incorporate the new regulatory regime and sign con- cessions with the existing municipal water firms in Santa Cruz, Oruro, Sucre, and other smaller cities. Under the new model, the concession approach seeks to improve inter- nal efficiency and achieve expansion and quality goals. The new Potable Water and Sewerage Law has four important elements: · Responsibility for the provision of these services is assigned to the municipal governments but can be transferred to water and sewerage providers (WSPs) that are private, municipal, or mixed firms, coopera- tives, or other civil associations recognized by law. · The territory is divided into concession and nonconcession areas: the concession areas are financially sustainable, and services are provided only 128 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO by WSPs, while nonconcession areas are not financially sustainable, and the service can be provided by a local government. · Regulation of WSPs includes tariff regulation using the rate-of-return criteria, investment and efficiency targets, and a five-year regulatory lag. · Universal access in nonconcession areas is to be supported by public investment. The contract signed with Aguas del Illimani reflects these aims. Objec- tives for the 1997­2001 period include 100 percent access to potable wa- ter or sewerage (public fountains excluded) in the areas of Achachicala and Pampahasi, which cover the city of La Paz; 82 percent access to potable water in the city of El Alto by 2001, of which 50 percent was to consist of expansion connections, and 41 percent access to sewerage; and compliance with the goal of expanding, over time, the percentage of households with potable water coverage and sanitation services. Quality control includes norms related to the source of water; its qual- ity, abundance, and pressure; continuity of service; infrastructure effi- ciency; consumer service; and emergencies. Regulated prices are calculated to assist the company to meet its contractual obligations and expansion goals. Although the five-year lag promotes internal efficiency, no produc- tivity factors were incorporated. Furthermore, tariffs were set in dollar terms, payable in bolivianos. Data Sample The task of collecting information for this chapter required the collab- oration of various public and private institutions. Several sources of in- formation were used. From the National Statistical Institute and the National Tax Service, we obtained financial information for the period in which the companies were part of the public sector. The statistical agency provided information on approximately 30 percent of the com- panies in the study. The National Service of Commerce Registry and the General Superintendency helped with postprivatization information. The former provided information on owners of the businesses, while the latter provided annual reports of the capitalized companies. We also carried out a survey to complete the financial information needed for this study. Other information came from the Reorganization Unit, the Vice Ministry of Budget and Accounting, and the National Chamber of Hotels. Data were not always strictly compatible across sources, and an extra effort to confirm sources had to be made. When differences were encoun- tered, data and sources were selected based on the series that matched existing trends most closely.8 The sample's small size also implies that few results are statistically significant. Nonetheless, the data set we constructed PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 129 constitutes, to our knowledge, the most serious and comprehensive data on the transfer of SOEs to the private sector in Bolivia. Because of the small size of many privatized firms (more than 90 per- cent were small or medium-size enterprises), financial and performance information was not always collected or kept up-to-date either before or after privatization. Unlike other countries, very few privatized firms were listed on the stock exchange either before or after their transfer to private hands. This explains the lack of a single source of comprehensive data and the multiplicity of sources needed to construct our database. According to the Reorganization Unit, 94 public enterprises were trans- ferred to the private sector between 1992 and 2003.9 Of these, 83 were pri- vatized under the traditional method, 10 through capitalizations, and 1 as a concession.10 While the number of privatized firms seems large, only 31 firms had sufficient information to include in the final sample. Table 3.2 summarizes the number of firms in the sample and the reason for exclud- ing those that were not analyzed in the study. Four firms were excluded because they were too small and did not have proper accounting records. We were unable to confirm if these firms had effectively been privatized, much less to capture their finan- cial and employment information. Another 9 firms, all from the first wave, were sold as stripped assets, making a comparison impossible. Twenty-nine firms had no financial statements from the period in which they were in public hands, and we were unable to obtain the informa- tion from our surveys. Eleven firms were created from disintegrated monopolies from which we were unable to construct separate financial books to measure performance of the relevant units before privatiza- tion. We were therefore forced to exclude them from the sample. Finally, 10 firms from the third wave were left out of the sample because the sale was too recent for any meaningful postprivatization compari- son to be made. Table 3.2 Reasons for Excluding Firms from the Sample Reason Number of firms Firms too small 4 Firms sold as stripped assets 9 No financial statements from the period when firms 29 were public Firms created from disintegrated monopolies 11 Sale too recent 10 Enterprises not included in the study 63 Enterprises included in the study 31 Total 94 Source: Authors' data. 130 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Results Following the methodology of La Porta and López-de-Silanes (1999), this section seeks to measure the change in performance of the privatized firms by comparing performance indicators before and after privatization. We divide our indicators into the following five groups: profitability, operat- ing efficiency, labor, assets and investment, and output. To calculate estimates that take into account the sometimes large dif- ferences among industries, a difference-in-difference analysis was carried out using information from firms that had been private all along. As a con- trol group, we used data from the Manufacturing Industry Survey (CIIU), using the corresponding industrial classification for each firm; for firms that could not be classified within a corresponding CIIU code, we used an aggregate measure of the whole manufacturing industry as an economy- wide control. Before privatization refers to the average for the four years before privatization, or for the years with available data, while the term after privatization corresponds to the average value for 1999 and 2000, or for the last two years for which information is available.11 Raw Data Table 3.3 presents the results from our unadjusted ratios. Several interest- ing results can be appreciated. The first is that both mean and median profitability increased by 15 percentage points. Privatized firms' ratio of operating income to sales increased from a mean (median) of slightly over 20.1 (12.8) percentage points before privatization to 35.1 (27.9) percent- age points after privatization. Moreover, this increase is statistically sig- nificant at the 10 percent level. The next set of indicators suggests that the increase in profitability was primarily driven by improvements in efficiency. Our cost-per-unit indica- tor suggests that privatized firms underwent significant restructuring and streamlining following privatization. The mean (median) firm reduced its costs by 18.3 (20.8) percentage points, with the difference being signifi- cant at the 5 percent level. Moreover, the magnitude of this increase in efficiency suggests that efficiency gains by themselves can account for the increased profitability of privatized firms. Other indicators of operating efficiency provide additional circumstantial evidence to suggest that transfers from workers and market power are not a source of improved profitability of privatized firms. The mean (median) firm increased its ratio of sales to property, plant, and equipment (PPE) by 36.5 (8.1) per- cent and its ratio of sales to employees by 48.2 (30.4) percent. The most remarkable results, however, are those for the subsample of firms for which we can calculate the ratio of operating income to employees. In this case, the interaction of greater labor efficiency and reduced costs allowed Table 3.3 Changes in Performance of the Sample of Privatized Firms on Bolivia Before privatization After privatization Mean Mean Variable N Median Median t-statistic Z-statistic Profitability Operating income/sales 28 0.2010 0.3511 1.71* 1.64 0.1281 0.2793 Operating income/PPE 20 0.2063 0.2858 0.42 0.37 0.1048 0.1074 Operating efficiency Cost per unit 28 0.8217 0.6386 2.40** 2.26** 0.9110 0.7032 Log(sales/PPE) 23 0.4628 0.0981 1.26 1.09 0.4735 0.3925 Log(sales/employees) 23 11.6759 12.1581 1.59 1.52 11.8141 12.1181 Operating income/employees (thousands) 19 12.6049 73.6712 1.89* 1.76* 10.1750 39.7259 Labor Log(employees) 24 4.6709 4.5811 0.19 0.46 4.7405 4.6094 Log(blue-collar workers) 20 4.1466 4.0935 0.12 0.28 4.2853 4.2377 Log(white-collar workers) 20 1.4807 1.3572 0.25 0.61 1.1705 0.8959 131 Average wage per worker (thousands) 16 10.8515 19.9102 2.49** 2.26** 9.0361 18.8589 (Table continues on the following page.) Table 3.3 (continued) 132 Before privatization After privatization Mean Mean Variable N Median Median t-statistic Z-statistic Average wage per blue-collar worker (thousands) 16 9.5054 16.0569 2.24** 2.04** 7.7061 15.6106 Average wage per white-collar worker (thousands) 16 26.1511 59.9427 2.38** 2.33** 22.4439 44.5518 Assets and investment Log(PPE) 23 17.1634 16.8574 0.43 0.25 16.6764 16.9574 Investment/sales 18 0.0499 0.0560 0.18 1.27 0.0177 0.0110 Investment/employees (thousands) 15 10.8727 10.1345 0.09 1.763* 2.5556 1.0845 Investment/PPE 17 0.0288 0.0232 0.34 0.95 0.0137 0.0073 Log(PPE/employees) 18 12.4286 12.2558 0.47 0.35 12.1583 12.1274 Output Log(sales) 31 15.9962 16.1039 0.16 0.12 15.5094 15.8996 * Significant at the 1 percent level. ** Significant at the 5 percent level. *** Significant at the 10 percent level. Note: N number and PPE property, plant, and equipment. This table presents raw results for the sample of 31 privatized firms. For each empirical proxy, the table presents the number of usable observations and the mean and the median values before and after privatization. Before privatization refers to the average value for the four years before privatization; after privatization refers to the average value for 1999 and 2000, or for the last two years for which information is available. We report t-statistics and Z-statistics (Wilcoxon rank sum) as the test for significance for the change in mean and median values, respectively. Definitions for each variable can be found in appendix 3A.2. Source: Authors' calculations. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 133 the median firm to increase its operating income per worker by a stag- gering 290 percent. Privatization usually leads to employment cuts across the board, which encourages critics to argue that the gains in profitability are merely transfers from workers to the new owners. In the case of Bolivia, the mean (median) firm reduced its work force by 9.0 (13.1) percent, sig- nificantly less than the labor reduction experienced in other countries. For example, La Porta and López-de-Silanes (1999) found that the mean (median) firm in Mexico fired 64.9 (56.8) percent of its work force, while Torero (chapter 8) documents employment cuts of 51.0 (56.1) per- cent for Peru. Moreover, once we consider that the average wage per worker increased by a mean (median) of 83.5 (108.7) percent, it is clear that firms did not increase profits by exploiting workers. Even though the fall in employment is not statistically significant, the increase in wages is significant at the 5 percent level. An additional criticism of privatization is that even if firms do restruc- ture and become more efficient, it is the poorest workers who bear the brunt of the restructuring burden. Privatization is thus opposed not be- cause of concerns regarding its aggregate effect but because it is thought to have a negative redistributive effect. By breaking up employment cuts and wage increases into white-collar and blue-collar workers, we can shed some light on this issue. Regarding employment, the evidence conclusively shows that blue-collar workers fare better than their white-collar col- leagues. While the mean (median) firm fired 12.3 (27.5) percent of its white-collar workers, it fired only 5.3 (4.8) percent of its blue-collar ones. This evidence suggests that it is false to assume, at least in the case of Bolivia, that privatization hurt unskilled workers disproportionately. Re- garding wages, the evidence is mixed but still does not support the view that unskilled workers were exploited. Average wages of white-collar workers increased 129.2 percent, significantly more than the 68.9 percent increase observed for blue-collar workers. However, the median change in wages, arguably a better gauge as it is less susceptible to outlier observa- tions, shows virtually identical increases of about 100 percent for both types of workers. Moreover, even if wages for white-collar workers increased more than those of blue-collar ones, an average real increase of 70 to 100 percent in wages and greater job stability can hardly be consid- ered a disproportionate burden for unskilled workers. Regarding investment, there seems to be no clear-cut trend--the mean and median values of our investment ratios move in opposite directions. It is difficult to know a priori what to expect from investment following privati- zation. On the one hand, firms usually have ample spare capacity and thus room to increase production by using assets more efficiently. On the other hand, production processes may be outdated and require significant invest- ments to bring them up-to-date. Bolivia presents a good example of this am- bivalence, as many of its privatized firms clearly suffered from overcapacity. 134 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO However, some of the larger firms were privatized under capitalization schemes, the precise purpose of which was to increase investment. In any case, the only statistically significant result is the median for the ratio of in- vestment to employees, which shows a decrease of almost 60 percent. Finally, our results corroborate other studies of privatization showing that notwithstanding a reduced work force and no general increases in in- vestment, the mean (median) firm was able to increase output--in the case of Bolivia, by 10.8 (39.0) percent.12 This growth in real sales is the final piece in the puzzle, demonstrating that the main effect of privatization is a substantial improvement in efficiency, which permeates through as higher profits for firms, higher wages for workers, and increased output for consumers. Industry-Adjusted Data Toward the end of the 1990s, the Bolivian economy experienced impor- tant fluctuations in its growth rate and underwent great sectoral trans- formations. For example, its average growth rate ranged from a high of 4.8 percent in 1995­98 to just 0.44 percent in 1999.13 One may, there- fore, wonder whether the strong increases in output, profitability, and wages observed in our sample can be explained by general macroeco- nomic trends or sector-specific events. To isolate the role of privatiza- tion, we present industry-adjusted ratios in table 3.4. In short, we find that the performance improvements observed in the previous section are not explained by industrywide or macroeconomic changes and that in fact some of our industry-adjusted indicators show a more favorable pic- ture of performance by privatized firms than that suggested by the raw indicators described in the previous section. The first result is that industry-adjusted indicators show that output at the median privatized firm increased by 26.7 percent. This would imply that about 70 percent of the raw increase in output results from privatiza- tion, while only 30 percent can be explained by industry or macroeco- nomic changes. Focusing on the mean firm yields an even more favorable interpretation, with industry-adjusted output in privatized firms increas- ing 32.4 percent. Regarding profitability, the results show that privatized firms first con- verged to the levels of their industry peers and then surpassed them. The mean (median) ratio of operating income to sales shows that SOEs under- performed their industry groups by 14.4 (18.3) percentage points and that after privatization they overperformed the same control group by 13.7 (5.9) percentage points. As expected, these gains in profitability stem from larger cost savings and efficiency gains than those experienced by other firms. Industry-adjusted costs per unit dropped by a mean (median) of 6.4 (5.0) percentage points, while the ratio of sales to employees increased 51 (9.6) percent. Table 3.4 Industry-Adjusted Changes in Performance for the Sample of Privatized Firms Before privatization After privatization Mean Mean Variable N Median Median t-statistic Z-statistic Profitability Operating income/sales 28 0.1442 0.1373 3.52*** 3.15*** 0.1830 0.0594 Operating efficiency Cost per unit 28 0.2216 0.1573 0.72 0.23 0.2399 0.1905 Log(sales/employees) 22 7.8357 8.3456 0.74 0.96 8.5944 8.6909 Labor Index of total employees 24 100.00 70.33 3.01*** 5.29*** 100.00 58.98 Index of industry-adjusted real 16 100.00 238.02 1.76* 0.65 wages per worker 100.00 131.86 Output Log(sales) 31 5.6182 5.2942 0.53 0.67 6.3877 6.1206 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number. This table presents industry-adjusted results for the sample of 31 privatized firms. See text for an explanation of the way these results were computed. For each empirical proxy, the table presents the number of usable observations and the mean and median values before and 135 after privatization. Before privatization refers to the average value for the four years before privatization; after privatization refers to the average value for 1999 and 2000, or for the last two years for which information is available. We report t-statistics and Z-statistics (Wilcoxon rank sum) as the test for significance for the change in mean and median values, respectively. Definitions for each variable can be found in appendix table 3A.2. Source: Authors' calculations. 136 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Finally, our industry-adjusted indicators show that employment cuts after privatization are more significant than the raw data suggested, but the wage increases observed are robust. The mean (median) index of total employees dropped 29.7 (41.0) percent following privatization, signifi- cantly above the 9.0 (13.1) percent decrease observed in the raw data. This implies that firms in the control group hired a substantial number of work- ers over the period analyzed. Meanwhile, the industry-adjusted real wages per worker show a robust increase for the mean (median) firm of 138.0 (31.9) percent following privatization. All in all, our industry-adjusted indicators show that macroeconomic conditions and industry-specific changes cannot explain the large per- formance gains documented for privatized firms. Although in some cases a part of the gains can be explained by industry changes, the bulk of the improvement in firm performance is due to restructuring and efficiency gains. In fact, the raw indicators may be underestimating the true impact of privatization as industry-adjusted improvements are larger than the ones shown by the raw indicators. Conclusions This study assembled a comprehensive database for firm characteristics before and after privatization. Despite severe information limitations, we were able to collect sufficient data for one-third of the firms priva- tized between 1992 and 2003. Our sample contains firms of all sizes, with an emphasis on medium and small ones; this sample reflects the structure of privatization in Bolivia. Our results broadly mirror those found for other countries: privatiza- tion leads to a significant increase in efficiency and a corresponding rise in firm profitability. Although privatization results in a decrease in employ- ment, the Bolivian experience shows only mild changes relative to those found for other countries. Notably, real wages increase by almost 100 per- cent, suggesting that SOEs employed too many workers and used them in- effectually but did not overpay them. In this sense it is likely that these jobs were sought after not because they paid well but because they did not re- quire much effort. Investment shows no clear trend; the mean and median values move in opposite directions. This is probably a reflection of Bolivia's esoteric two-tiered structure of privatization in which most firms sold were suffering from overcapacity, but a few of the largest were sold under capitalization schemes, which by definition were aimed at increas- ing investment. Finally, we also find that despite a reduced work force and the ambivalent behavior of investment, firms were able to increase output substantially. These results carry through when we look at industry- adjusted indicators, and in fact, in some cases, shed a more favorable light on the performance of privatized firms. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 137 This study provides a valuable contribution for two reasons. First, it represents the first systematic study of the change in performance of Bolivian SOEs following privatization. Although previous authors have analyzed specific privatizations and others have written descriptive analyses of the program, this chapter represents, to our knowledge, the most comprehensive effort to collect employment and financial informa- tion for privatized firms in Bolivia. Second, the sample of firms priva- tized in Bolivia, unlike most other countries, is composed mostly of medium and small firms. In this sense, the results obtained from this analysis shed some light as to what other countries with similar charac- teristics may expect from privatization. Our results imply that there seem to be no significant differences in the benefits of privatization to small and medium-size firms than those observed for larger SOEs in other countries. Appendix Table 3A.1 List of Privatized Firms, 1992­2001 First wave 1 Fondo Ganadero del Beni y 18 Fábrica de Cerámica Roja de Pando Cobija 2 Fábrica de Objetos de Peltre 19 Hotel Prefectural Liriuni 3 Empresa Forestal Pecuaria 20 Centro de Acopio Yamparaez Tariquia 21 Centro de Acopio Redención 4 Planta Industrializadora de Pampa Quinua 22 Centro de Acopio Tomina 5 Planta de Alimentos 23 Hacienda Ganadera Santa Balanceados Portachuelo Martha 6 Empresa Nacional de La Casta~na 24 Hotel Prefectural de Caranavi 7 Industrias Metálicas (Inmetal) 25 Caba~na Lechera Todos Santos 8 Línea Aérea Imperial Paz 9 Pait - Pl Procesadora de 26 Hotel Prefectural de Pando Caranavi 27 Centro de Acopio Totora 10 Cadenas Andinas sam 28 Centro de Acopio Lourdes 11 Criadero de Truchas Piuisilla 29 Centro de Acopio Betanzos 12 Pollos Bb 30 Fabrica de Aceites Comestibles 13 Taller de Cerámica Artesanal Villamontes Chuquisaca 31 Fábrica Boliviana de Cerámica 14 Fábrica Nal. de Fósforos 32 Fábrica de Cerámica Roja de 15 Hotel La Paz (Ex-Sheraton) Oruro 16 Hotel Crillon 33 Hotel Prefectural de Tarija 17 Fábrica de Cerámica Roja de 34 Ingenio Azucarero Guabira Trinidad (Table continues on the following page.) 138 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Appendix Table 3A.1 (continued) Second wave (sale) 35 Empresa Transportadora de 55 Planta de Alimentos Electricidad (TDE) Balanceados Tarija 36 Empresa de generación eléctrica 56 Planta Industrializadora de Luz y Fuerza de Cochabamba Leche ­ Scz (ELFEC) 57 Planta Industrializadora de 37 Hilandería Santa Cruz Sam Leche Pil ­ La Paz 38 Planta Laminadora de Goma 58 Planta Industrializadora de Sam Leche Pil ­ Cbba 39 Bien Inmueble 59 Planta Industrializadora de 40 Planta de Hilandería Viacha Leche Pil ­ Tarija (Phv) 60 Planta Industrializadora de 41 Hotel Balneario Asahi Leche Pil ­ Sucre 42 Hotel Prefectural Caranavi 61 Ingenio Azucarero Guabira 43 Hotel Prefectural de 62 Fabrica de Aceites Comestibles Copacabana Villamontes 44 Hotel Prefectural de Coroico 63 Planta Elaboradora de Quesos 45 Hotel Prefectural de Sorata San Javier 46 Hotel Prefectural Viscachani 64 Planta de Alimentos 47 Hotel Prefectural de Urmiri Balanceados Tarija 48 Hotel Prefectural de Chulumani 65 Producción de Harinas 49 Hotel Terminal de Oruro Compuestas ­ Tarhui 50 Terminal de Buses Oruro 66 Caba~na de Porcinos "El 51 Terminal de Buses Sucre Zapallar" 52 Terminal de Buses Cochabamba 67 Hilandería Pulacayo 53 Producción de Alimentos de 68 Multipropósito Gran Chaco Maiz Mairana 69 Fábrica Boliviana de Cerámica 54 Planta de Alimentos 70 Fábrica de Cerámica Roja de Balanceados de Chuquisaca Oruro Second wave (capitalization) 71 Empresa Generadora de 76 Lloyd Aéreo Boliviano Electricidad (CORANI) 77 Empresa Petrolera Andina 72 Empresa Nacional de 78 Empresa Petrolera Chaco Telecomunicaciones (ENTEL) 79 Transportadora de 73 Empresa Ferroviaria Andina hidrocarburos (TRANSREDES) 74 Empresa Ferroviaria Oriental 80 Empresa Generadora de 75 Empresa Generadora de Electricidad (Valle Hermoso) Electricidad (GUARACACHI) Third wave 81 Refinería de Petróleo (EBR) 83 Estaciones de servicio de 82 Planta de almacenaje de aeropuertos carburantes y Poliductos 84 Gasolineras PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 139 Appendix Table 3A.1 (continued) Third wave 85 Empresa Minera Vinto ­ 90 Empresa Metalúrgica Vinto ­ Antimonio Esta~no 86 Financiera de Desarrollo S.A. 91 Empresa Minera Huanuni 87 Campo Geotérmico Laguna 92 Planta Industrial Oruro Colorada 93 Empresa Minera Colquiri 88 Fábrica Nacional de Explosivos 94 Planta de Productos Lácteos Sam Milka 89 Fábrica Nacional de Cemento Source: Authors. Appendix Table 3A.2 Description of the Variables Used in Tables 3.3 and 3.4 Variable Description Operating The ratio of operating income to sales. Operating income/sales income is equal to sales minus operating costs and minus depreciation. Sales are equal to the total value of products and services sold, nationally and interna- tionally. Operating costs is equal to the direct expense involved in the production of a good (or provision of a service). This includes the day-to-day expenses incurred in running a business, such as sales and administration. It is also called operating expenses. Operating The ratio of operating income to property, plant, and income/PPE equipment. Operating income is equal to sales minus operating costs and minus depreciation. Operating costs is equal to the direct expense involved in the production of a good (or provision of a service). Prop- erty, plant, and equipment is equal to the value of a company's fixed assets adjusted for inflation. Cost per unit The ratio of operating costs to sales. Operating costs is equal to the direct expense involved in the production of a good (or provision of a service). This includes the day-to-day expenses incurred in running a business, such as sales and administration. Sales are equal to the total value of products and services sold, nation- ally and internationally. Log(sales/PPE) Natural logarithm of the ratio of sales to property, plant, and equipment. Sales are equal to the total value of products and services sold, nationally and (Table continues on the following page.) 140 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Appendix Table 3A.2 (continued) Variable Description internationally. Property, plant, and equipment is equal to the value of a company's fixed assets adjusted for inflation. Log(sales/ Natural logarithm of the ratio of sales to total number of employees) employees. Sales are equal to the total value of prod- ucts and services sold, nationally and internationally. Employees corresponds to the total number of workers who depend directly on the company. Operating The ratio of operating income to total number of income/ employees. Operating income is equal to sales minus employees operating costs and minus depreciation. Employees corresponds to the total number of workers who depend directly on the company. Log(employees) Natural logarithm of the total number of employees. Employees corresponds to the total number of full time workers who depend directly on the company. It does not include individuals who are retired or working on commission. Log(blue-collar Natural logarithm of the total number of blue-collar workers) workers. Blue-collar workers perform un- or semi- skilled labor for modest to low wages. They perform tasks directly related to the (mass) production process or menial services. Typically, they are factory-line or maintenance workers. Log(white- Natural logarithm of the total number of white-collar collar workers. White-collar workers perform skilled labor workers) and administrative tasks for modest to high salaries. They are individuals involved in sales, administration, and management. Average wage The average wage paid per worker in each firm. The per worker consumer price index was used as a deflator to calculate real wages. A similar procedure is used for the calculation of real wages per blue-collar and white-collar workers. Log(PPE) Natural logarithm of property, plant, and equipment. Property, plant, and equipment is equal to the value of a company's fixed assets adjusted for inflation. Investment/sales The ratio of investment to sales. Investment is equal to the value of expenditure to acquire property, equip- ment, and other capital assets that produce revenue. Sales are equal to the total value of products and services sold, nationally and internationally. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 141 Appendix Table 3A.2 (continued) Variable Description Investment/ The ratio of investment to employees. Investment is employees equal to the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. Employees corresponds to the total number of workers (paid) who depend directly on the company. Investment/PPE The ratio of investment to property, plant, and equip- ment. Investment is equal to the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. Property, plant, and equipment is equal to the value of a company's fixed assets adjusted for inflation. Log(PPE/ Natural logarithm of the ratio of property, plant, and employees) equipment to total number of employees. Property, plant, and equipment is equal to the value of a com- pany's fixed assets adjusted for inflation. Employees corresponds to the total number of workers who depend directly on the company. Log(sales) Natural logarithm of sales. Sales are equal to the total value of products and services sold, nationally and internationally. Index of total For each firm, the index takes value of 100 for the employees preprivatization period. The average of the last two years after privatization value is computed by aug- menting the preprivatization value by the difference between the cumulative growth rate of employment of the firm and the cumulative growth rate of employ- ment of the control group in the postprivatization period relative to the average employment in the four years that preceded privatization. Industry control groups are given by an index of economywide total employment. Index of For each firm, the index takes value of 100 for the industry- preprivatization period. The average of the last two adjusted real years after privatization value is computed by aug- wages per menting the preprivatization value by the difference worker between the cumulative growth rate of real wages per worker of the firm and the cumulative growth rate of real wages per worker of the control group in the postprivatization period relative to the average real wage per worker in the four years that preceded pri- vatization. Industry control groups are given by an index of economywide real wages per worker. Source: Authors. 142 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Notes The authors are indebted to Patricio Amador, Jose Caballero, Cecilia Calderon, Gabriela Enrique, Paola Espinoza, Jesús Mogrovejo, and Monica Yanez for pro- viding skillful research assistance. 1. For a detailed discussion of privatization effects on these objectives as well as their interrelation, see Sheshinski and López-Calva (1999). 2. Some extensions to these two initial arguments analyze more deeply the relationship between property, efficiency, and contracts. These include agency theory, public choice theory, property rights theory, and transactions costs analy- sis; see Commander and Killick (1988) and Hodge (1999), among others. 3. From 1989 to 1993, the government of Jaime Paz Zamora; from 1993 to 1997, the government of Gonzalo Sanchez de Lozada; and from 1997 to 2001, the government of Hugo Banzer Suarez. 4. Privatization Law 1330, April 24, 1992. 5. See López-de-Silanes (1997) for a discussion of prior restructuring meas- ures and their effect on the price paid for privatized SOEs. 6. The Capitalization Law that made this type of privatization possible was enacted on March 21, 1994. 7. All dollar values are U.S. dollars unless otherwise indicated. 8. Many of the differences resulted from the application of distinct ac- counting systems, as many of the balance sheets and income statements were not audited. 9. This number includes one concession. The Reorganization Unit is a unit within the Ministry of Investment and Privatization and serves as the government's agency responsible for privatization. As such, it maintains records of the privatiza- tions undertaken in Bolivia. 10. The number of capitalized enterprises can be counted in two different ways: as the number of enterprises before capitalization (5), or as the number of enter- prises resulting from the capitalization process (10). For this chapter we have adopted the second measure. 11. For the purpose of calculating the average of the last two years, we consider the year of privatization as a postprivatization year. 12. For example, the median increase in output of privatized firms was 17.4 percent in Brazil, 69.0 percent increase in Colombia, 68.2 percent in Mexico, and 40.8 percent in Peru; see chapters 4, 6, and 8, and La Porta and López-de-Silanes 1999. 13. During the 1990s increased openness brought a greater influx of foreign capital. This, in turn, favored the growth of the economy, raised employment lev- els, and increased mean labor productivity (Jiménez, Pereira, and Hernany 2002). References Barberis, Nicholas, Maxim Boycko, Andrei Shleifer, and Natalia Tsukanova. 1996. "How Does Privatization Work? Evidence from the Russian Shops." Journal of Political Economy 104: 764­90. Barja, Gover, and Miguel Urquiola. 2001. "Capitalization, Regulation, and the Poor: Access to Basic Services in Bolivia." Working Paper, funded by the United Nations University/World Institute for Development Economics Research. Helsinki. PRIVATIZATION AND FIRM PERFORMANCE IN BOLIVIA 143 Boubakri, Narjess, and Jean-Claude Cosset. 1998. "The Financial and Operating Performance of Newly Privatized Firms: Evidence from Developing Countries." Journal of Finance 53: 1081­110. Claessens, Stijn, and Simeon Djankov. 1998. "Politicians and Firms in Seven Cen- tral and Eastern European Countries." Policy Research Working Paper 1954. World Bank, Washington, D.C. Claessens, Stijn, Simeon Djankov, and Gerhard Pohl. 1997. "Ownership and Cor- porate Governance: Evidence from the Czech Republic." Policy Research Paper 1737. World Bank, Washington, D.C. Commander, Simon, and Tony Killick. 1988. "Privatisation in Developing Coutries: a Survey of the Issues." In Paul Cook and Colin Kirkpatrick, eds., Pri- vatisation in Less Developed Countries. New York: St. Martin's Press. D'Souza, Juliet. 1998. "Privatization of Telecommunication Companies: An Em- pirical Analysis." Working Paper, Mercer University, Macon, Ga. Eckel, Catherine, Doug Eckel, and Vijay Singal. 1997. "Privatization and Effi- ciency: Industry Effects of the Sale of British Airways." Journal of Financial Economics 43: 275­98. Frydman, Roman, Cherryl W. Gray, Marek Hessel, and Andrzej Rapaczynski. 1998. "Private Ownership and Corporate Performance: Some Lessons from Transition Economies." Working Paper. New York University, C.V. Starr Cen- ter for Applied Economics, New York. Galal, Ahmed, Leroy Jones, Pankay Tandon, and Ongo Vogelsang. 1994. Welfare Consequences of Selling Public Enterprises. Oxford, U.K.: Oxford University Press. Hodge, Graeme A. 1999. Privatization: An International Review of Performance. Boulder, Colo.: Westview Press. Jiménez Wilson, Rodney Pereira, and Werner Hernany. 2002. "Bolivia: Efecto de la Liberalización Sobre el Crecimiento, Empleo, Distribución y Pobreza." In Liberalizacion, Desigualdad y Pobreza: America Latina y el Caribe en los 90, Capitulo 5. Programa de las Naciones Unidas para el Desarrollo (PNUD), Comision Economica para America Latina y el Caribe (CEPAL). La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." Quarterly Journal of Economics 114 (4): 1193­242. López-de-Silanes, Florencio. 1997. "Determinants of Privatization Prices." Quar- terly Journal of Economics 107 (4): 965­1025. López-de-Silanes, Florencio, Andrei Shleifer, and Robert Vishny. 1997. "Privatiza- tion in the United States." Rand Journal of Economics 28 (3): 447­71. Martín, Stephen, and David Parker. 1995. "Privatization and Economic Perfor- mance Throughout the UK Business Cycle." Managerial and Decision Eco- nomics 16: 225­37. Megginson, William L., and Jeffry M. Netter. 2001. "From State to Market: A Sur- vey of Empirical Studies on Privatization." Journal of Economic Literature 39: 321­89. Megginson, William L., Robert C. Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An In- ternational Empirical Analysis." Journal of Finance 49 (2): 403­52. Ministerio de Capitalización. 1994. "Monitor de la Capitalización, números espe- ciales." La Paz, Bolivia. 144 CAPRA, CHONG, GARRÓN, LÓPEZ-DE-SILANES, AND MACHICADO Montero, Marcelo. 1994. "Análisis Financiero Nacional, Regional e Individual de las Empresas Privatizadas." Ley N° 1178 SAFCO (Sistema de Administración Financiera y Control Gubernamental). La Paz. Moore, Jacqueline. 1990. "Privatization: The Liberals View to Improving Effi- ciency and Performance of Industry." I. I. Pty, Ltd., Sydney. Nellis, John. 1991. "Privatization in Reforming Socialist Economies." In J. Bohn and V. Kreacic, eds., Privatization in Eastern Europe: Current Implementation Issues. Yugoslavia: International Centre for Public Enterprises in Developing Countries. Newberry, David, and Michael G. Pollitt. 1997. "The Restructuring and Privati- zation of Britain´s CEGN ­ Was It Worth It?" Journal of Industrial Economics 45: 269­303. Pohl, Gerhard, Stijn Claessens, and Simeon Djankov. 1997. "Privatization and Re- structuring in Central and Eastern Europe: Evidence and Policy Options." Technical Paper 368. World Bank, Washington, D.C. Ramamurti, Ravi. 1997. "Testing the Limits of Privatization: Argentine Rail- roads." World Development 25: 1973­93. Reordenamiento de las empresas públicas. 1994. Government of Bolivia, La Paz. Requena, Mario. 1996. "La Experiencia de Privatización y Capitalización en Bolivia." Serie de Reformas de Política Pública 38, Comisión Económica para América Latina y El Caribe, Santiago, Chile. Shapiro, Carl, and Robert Willig. 1990. `'Economic Rationales for the Scope of Pri- vatization." In B. N. Suleiman and J. Waterbury, eds., The Political Economy of Public Sector Reform and Privatization, 55­87. London: Westview Press. Sheshinski, Eytan, and Luis F. Lopez-Calva. 1999. "Privatization and Its Bene- fits: Theory and Evidence." HDII Discussion Paper 698. Harvard University, Cambridge, Mass. Shleifer, Andrei. 1998. "State Versus Private Ownership." Journal of Economic Perspectives 12(4): 133­50. Shleifer, Andrei, and Robert Vishny. 1994. "Politicians and Firms." Quarterly Journal of Economics 109 (4): 995­1025. ------. 1996. "A Theory of Privatization." Economic Journal 106: 309­19. SIRESE (Sistema de Regulación Sectorial). 2000. "La Regulación Sectorial en Bolivia." Annual Report, Superintendencia General, La Paz. Vickers, John, and George Yarrow. 1988. "Privatization: An Economic Analysis." Cambridge, Mass.: MIT Press. Wiltshire, Kenneth. 1987. "Privatization: The British Experience: An Australian Perspective." Committee for Economic Development of Australia and Longman Cheshire, Melbourne. 4 Costs and Benefits of Privatization: Evidence from Brazil Francisco Anuatti-Neto, Milton Barossi-Filho, Antonio Gledson de Carvalho, and Roberto Macedo THE BRAZILIAN PRIVATIZATION PROGRAM has been a major undertaking by international standards. Between 1991 and July 2001, the state transferred its control of 119 firms and minority stakes in a number of companies to the private sector. The auctions produced $67.9 billion in revenues, plus the transfer of $18.1 billion in debt. The government also sold $6 billion in shares of firms that remained under state control, obtained $10 billion from new concessions of public services to the private sector, and sold $1.1 billion in scattered, noncontrolling stakes in various private companies owned by BNDES, the National Social and Economic Development Bank. The magnitude of the Brazilian privatization program is among the largest in the world, making it worthy of closer analysis. The Brazilian program has also been large in relative terms. Lora and Panizza (2002) compared the cumulative value of the privatization efforts between 1988 and 1999 as a proportion of gross domestic product (GDP) in 10 South and Central American countries (Argentina, Bolivia, Brazil, Costa Rica, Ecuador, El Salvador, Honduras, Paraguay, Peru, and Uruguay). Brazil came in third with a rate of 5 percent of GDP, above the average of 2.7 percent, and was surpassed only by Peru (6 percent) and Bolivia (9 percent).1 The value of privatizations in 5 of the other countries did not exceed 1 percent of GDP. Despite its magnitude, the Brazilian program has been largely ignored in the international literature. For instance, a survey by Megginson and Net- ter (2001, p. 326) recognized the Brazilian program as "likely to remain very influential," because of its scale and the size of the country. Their sur- vey did not include any specific analysis of the Brazilian program, however, 145 146 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO largely because of the paucity of studies and because most of the literature that does exist has been published only in Brazil and in Portuguese. Fur- thermore, the existing studies have their shortcomings, as their review in this chapter makes clear. Therefore, there is room for adding to both the Brazilian and the international literature. It is also important to disseminate findings among the Brazilian public at large. The economy's performance was very disappointing in the 1990s. Some groups, among them politicians and journalists, have often ex- pressed their frustration with privatization and other reform and adjust- ment policies, blaming them for the sluggish growth of the economy. In part because of this, the program all but stalled after 1998. Thus, it is cru- cial to show the results of the privatization program to shed light on a dis- cussion largely based on unwarranted conclusions. With regard to theoretical aspects, privatization is the subject of a wider and continuing debate on the role of the government in the economy. The analysis in this chapter is primarily focused on the relative effectiveness of private versus public ownership of companies that underwent privatization in Brazil. As a working hypothesis, the chapter tests the proposition that private ownership is more effective, but it also looks at the ways in which privatization results in increased profits, such as higher prices and reduced employment. Moreover, it discusses the management of the privatization process in terms of its macroeconomic implications and its objective of de- mocratizing capital ownership, among other issues. In this fashion, the chapter provides empirical evidence important to understanding the role of public ownership in the country, as well as the process by which the state has been stepping back from an entrepreneurial role. The next section describes the Brazilian privatization program and sur- veys the literature on it. The chapter then presents the variables and the data set used in the empirical analysis. The methodology and the empiri- cal results are summarized and other benefits and costs are discussed. The chapter then examines public opinion on the privatization program in Brazil and compares these views with those in other countries in the same region. It also evaluates the perspectives for new privatization efforts in the country, before summarizing the major conclusions. The Brazilian Privatization Program and the Literature The Brazilian privatization program has three components: the federal National Program of "Destatization" (NPD), which started in 1991; sim- ilar programs at the state level, which began in 1996; and the privatiza- tion program of the telecommunications industry.2 This last component was launched in 1997 as a program at the federal level, separate from the NPD but running parallel to it. We refer to it here as the telecom program. Its auctions, mostly taking place in 1997 and 1998, produced a total of COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 147 $28.8 billion in revenues plus $2.1 billion in debt transfers. The NPD yielded a total of $28.2 billion in revenues plus $9.2 billion in debt trans- fers, while the state-level program produced a total of $27.9 billion in revenues plus $6.8 billion in debt transfers.3 Altogether, electricity accounted for 31 percent of the total value of the auctions; telecommunications, 31 percent; steel, 8 percent; mining, 8 per- cent; oil and gas, 7 percent; petrochemicals, 7 percent; financial, 6 percent; and others, 2 percent. Largely because of the telecom program, the value of privatizations reached a peak in 1997­98, a period that accounted for 69 percent of the total value as of July 2001. This statistic has important implications for our later analysis of the effect the privatization program had on Brazil's fiscal crises and external imbalances.4 In any discussion of privatization, it is important to know what enter- prises the government owned before the privatization, which enterprises were privatized, and which remained under government control. We have little information on the initial situation in the various Brazilian state gov- ernments and on what remains to be privatized there. We therefore focus on the federal level only, the most important part of the program, where our information covers the whole program, except for the concession of public services. An Overview of Privatization at the Federal Level In 1980 the federal government undertook a survey of all its "entities," in- cluding companies, foundations, port authorities, research institutes, and councils in charge of professional registration. There were 560 such insti- tutions, of which 250 were organized as firms (mainly in the form of cor- porations). In the 1980s some minor privatizations occurred, and a few firms were also closed. At the start of the program in 1991, other entities had also ceased to exist. As a result, the program was launched with 186 firms still under government control. At the end of 2000, mainly because of the privatization program, this number was reduced to 102. Appendix table 4A.1 lists the companies privatized by the federal gov- ernment since 1990. Appendix table 4A.2 lists the firms privatized on be- half of some states by BNDES, some minority controlling stakes formerly held by the federal government, and firms privatized by the state of São Paulo. In both tables, we list the firms included in our sample and the rev- enue obtained from their privatization. Note is also made of the compa- nies that were listed on the São Paulo stock exchange before privatization. Appendix table 4A.3 lists the enterprises owned by the federal govern- ment that were not privatized. The group includes, among others, hospi- tals, port authorities, the postal service, an agricultural research firm, and BNDES. Among the industries where companies remain to be privatized are the electricity industry, where privatization of several major companies has been postponed; the oil industry; and the financial sector, in which a 148 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO few federal banks and most state banks have already been privatized (the latter group having been federalized for this purpose). The table also lists a group of entities organized as corporations over which the government ex- ercises 100 percent control. Some of these entities are government agencies disguised as corporations. These firms are directly linked to the federal budget, from which they receive practically all the resources they use. The privatization program has made little progress since 1998. Among other reasons, privatization and other liberalization measures coincided with sluggish growth, which weakened support for the program. More- over, some accusations that the government had used excessive methods to bring interested groups to the telecom auctions caused a furor in the press and led the minister of telecommunications to resign in 1998. Fur- thermore, if continued, the program would extend into politically sensitive areas such as electricity, where the states are very strong; oil, where the gi- gantic Petrobrás still arouses strong nationalistic feelings; and the almost 200-year-old Banco do Brasil, which plays an important role in financing farmers and therefore enjoys strong political support. The Brazilian Literature on Privatization In reviewing this literature, we concentrate on the studies that have ad- dressed the status of the state-owned enterprises (SOEs) before and after privatization, as that is the major focus of this chapter. At a later point we refer to the literature on other issues as well. Three studies are worth reviewing. Pinheiro and Giambiagi (1997) of BNDES presented an overall evaluation of the preprivatization perfor- mance of federal SOEs in the 1981­94 period. They showed disappointing figures for the SOEs, in terms of both profitability and dividends received by the national treasury. Over that entire period, the ratio of profits to net assets was 2.5 percent, on average. Moreover, from 1988 to 1994, years for which data on dividends were available, the ratio of profits to assets accounted for only 0.4 percent of the equity capital owned by the federal government in the SOEs. One of the causes of this disappointing performance was the SOEs' wage policies. Macedo (1985) undertook a comprehensive analysis of wage differentials between private firms and SOEs. His data consisted of wages and other characteristics of individual workers, obtained from forms filled out by firms every year, as required by the Ministry of Labor.5 He com- pared the wages of the workers in private firms and SOEs of approximately the same size in 10 industries. After controlling for differences in education, age, gender, and experience, he found sizable wage differentials in favor of the workers at the SOEs.6 The third study is Pinheiro (1996). He analyzed the performance of 50 former SOEs before and after privatization, using data up to 1994. His data covered one to four years before and after privatization for each company COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 149 and came from data sets similar to those used in this study but that were complemented by questionnaires filled out by the firms and delivered to BNDES for that purpose. Unfortunately, the bank's policy prevents the use of the data by outsiders. The study covered eight variables: net sales, net profit, net assets, investment, fixed investment, number of employees, debt, and an index of liquidity. From these variables, another six were derived to measure efficiency: sales and profit by employee, the rate of return meas- ured by the ratios of profit to sales and profit to net assets, and the propen- sity to invest, with respect to both sales and assets. No comparison was made with the performance of the private sector as a control group, nor was a distinction made between companies listed on the stock exchange and those that were unlisted. Pinheiro (1996) concluded that "in general, the obtained results confirm that privatization brings a significant improvement . . . [in] the performance of the firms. Thus, for most of the variables, the null hypothesis of no change in behavior is rejected in favor of the alternative hypotheses that privatization increases the production, the efficiency, the profitability and the propensity to invest, reduces employment and improves the financial indicators of the firms."7 In the current study, we add to this literature in various respects. The study was carried out by an independent team, whereas most of the previ- ous major studies were produced by staff members of BNDES. The study covers a larger number of firms, looks at privatization up to the year 2000, and uses data that can be disclosed. We took explicit care to avoid a se- lection bias by including both large and small privatized firms, SOEs, and firms in which the state held minority control, as well as firms listed on the stock exchange and firms that were unlisted. In addition to tests of means, the empirical work also employs panel data analysis. Moreover, the analy- sis of performance before and after privatization is also compared with the indicators observed in the private sector during the same periods. The importance of this last feature must be underscored, as the Brazil- ian economy underwent various cycles in the pre- and postprivatization periods. In summary, after strong growth in 1994 and 1995, when a mod- est number of companies were privatized, the economy's performance was sluggish until a strong recovery took hold in 2000, after the privatization program had passed its peak. Thus, economic cycles might have affected the performance of former SOEs. The absence of control for this effect could have blurred the results of the impact of privatization. The Data Set and the Variables Our data set is based on the annual financial statements (balance sheets, in- come statements, and cash flows) of the former SOEs and of a number of private companies used as a control group. Brazilian accounting standards 150 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO and procedures, as established by law and regulatory agencies, have re- mained the same for the whole period, thus facilitating our analysis.8 The data range from 1987 to 2000. The financial statements were obtained from two consulting firms (Economática and Austin Assis) and a non- governmental organization (the Getúlio Vargas Foundation). All three col- lect financial statements from several sources, including newspapers. We excluded from our analysis the privatizations in the financial sector, as that sector has a unique structure, involves specific issues, and would have re- quired specialized analysis. We also excluded the cases in which BNDES sold minor, noncontrolling participations in scattered companies as part of its portfolio as a development bank. Thus, we focused only on sales of con- trol packages, both of a majority and of a minority nature. Table 4.1 ex- plains the coverage of the sample. To proceed, it is necessary to distinguish privatization contracts (or auctions) from privatized enterprises. A number of former SOEs were sold as a block, and the successful bidder for an operational holding company was also given access to the control of its subsidiaries. In the case of the telecommunications sector, for instance, five amalgamated blocks of pri- vatization auctions covered the entire local, cellular, long-distance, and in- ternational restructured segments. As a result, the data set of the sampled companies covers 66 privatization contracts, corresponding to 102 firms. From the figures given in table 4.1, Table 4.1 Description of the Privatization Program and Coverage of the Sample, 1991­2000 Number of Number of Auction results Type of privatization contracts companies (US$ millions)a Privatization program Financial sector 9 9 5,112.30 Minority sales in SOEs 6 6 6,164.10 BNDES participations -- -- 1,146.00 Control package sales 103 147 76,878.20 Total 118 162 89,439.20 Companies included in sample State minority control 16 16 1,299.20 State majority control 50 86 70,709.80 Total 66 102 72,009.00 Note: SOEs state-owned companies and BNDES National Social and Economic Development Bank. a. These values include transferred debt ($17.8 billion) and offers to employees in the telecommunications industry ($0.3 billion) but exclude concessions of new services ($7.7 billion). Source: Authors' calculations. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 151 it can be inferred that the sample covers 64 percent of the control pack- ages, 69 percent of the firms they include, and 94 percent of the total value of the auctions. The smaller number of companies in the mean and median tests is explained by the methodology adopted and described in the next section. No information was available from the sources listed above for the 37 contracts not included, which correspond to 45 companies and yielded proceeds of $4.9 billion, as listed in appendix table 4A.1. Attempts to gather information from BNDES White Books were frustrated by nondisclosure and confidentiality rules governing data held by the bank.9 Table 4.2 sum- marizes the number of Brazilian companies privatized according to industry classification, along with their value at the auctions and the percentage they represented of the total value of each industry. All but three of the electricity companies were included in the sample. The sample includes one of the three gas distribution companies that were privatized at the state level. More petrochemical, fertilizer, and chemical plants are included in the sample than are excluded. The excluded group includes various limited liability companies, which are not required to make their balance sheets and income statements public. Although they are not important in economic size, the release of informa- tion on the privatized railways and ports could yield interesting case studies, as privatization accompanied the restructuring of these industries even as the government retained an active role in them. The railways were operated un- der regional branches of the federal railway network and were split into re- gional companies for privatization purposes only. The regional port facilities had been separate companies, operating under a federal holding company. In this case, privatization led to the creation of specialized terminals to be leased to private operators, with part of the infrastructure facilities remaining in the hands of SOEs. Thus, if data were available, one could compare the per- formance of both private firms and SOEs working side by side. The companies under the heading "others" include miscellaneous ac- tivities such as bus terminals, data processing, and ferryboats or small firms that are not organized as corporations and are also not required to make their balance sheets and income statements public. Thus, what was left outside of our sample represents only a minor part of the program, but not an uninteresting group for industry- and firm- specific studies. Their absence, due to insurmountable difficulties, does not jeopardize the relevance of our sample as representative of the com- panies that underwent privatization in Brazil. When the information was available, as it was for most of the companies and the most important ones, it was included in the sample. Our data set involves essentially the same variables used by La Porta and López-de-Silanes (1999) in their study of the Mexican case. Fifteen financial indicators, according to seven criteria, make up this set of vari- ables, as described in table 4.3. 152 Table 4.2 Privatized Brazilian Companies by Industry Classification Companies included Auction value Percent Companies excluded Auction value Percent Industry classification in sample (US$ millions) included from sample (US$ millions) excluded Aviation 1 455.0 100.0 0 0.0 0.0 Bank and financial 0 0.0 0.0 8 5,107.0 100.0 industrya Electricity and power 23 29,959.1 96.35 3 1,134.0 3.65 plants Fertilizers 2 452.0 91.50 3 42.0 8.50 Mining 2 6,864.0 100.0 0 0.0 0.0 Oil and gas 3 5,538.0 97.44 1 146.0 2.56 Others 0 0.0 0.0 7 809.0 100.0 Petrochemicals 17 2,156.0 75.86 7 686.0 24.14 Ports and container 0 0.0 0.0 9 461.0 100.0 terminals Railways 2 1,076.0 62.50 6 646.0 37.50 Steel 6 8,187.0 100.0 0 0.0 0.0 Telecommunications 44 23,858.0 100.0 0 0.0 0.0 Water and sewage 2 592.0 100.0 0 0.0 0.0 Total 102 79,137.1 89.75 44 9,031.0 10.25 a. All information about this industry is excluded from the sample as in La Porta and López-de-Silanes (1999). Source: Authors' calculations. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 153 Table 4.3 Summary of Results Appendix tables Indicator 4C.1 4C.2 4C.3 4C.4 Profitability Operating income/sales Operating income/PPE Net income/sales Return on assets Return on equity Operating efficiency Log(sales/PPE) Operating cost/sales Assets Log(PPE) Investment/sales Investment/PPE Output Log(sales) Shareholders Payout Finance Current Long-term debt/equity Net taxes Net taxes/sales Note: PPE property, plant, and equipment. See appendix 4B for definitions of the variables. See appendix 4C for full results of the mean tests. Appendix tables 4C.1 and 4C.2 refer to the results obtained upon the calculation of mean and median values for two years before and two years after privatization (Method 1). The adjustments on these calculated values are made for comparisons with the private sector and produce the results in appendix table 4C.2. The results in appendix tables 4C.3 and 4C.4 are based on the same procedure, but the mean and median values are calculated for all years before and after privatization (Method 2). The shading means the indicator mean difference is significant at least at the 10 percent level. The sample size of companies taken for the mean and median tests is 73. Auction results amount to $68.1 billion. Empirical Analysis Two different approaches were adopted to examine changes in per- formance after privatization: mean and median tests, and panel data analysis. 154 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Mean and Median Tests For the mean and median tests, two different methods were used. Under the first method, for each indicator a comparison is made between the mean and median values of the two years following privatization and their values in the two years before privatization.10 The second method fully utilizes the information in the data set by comparing the mean and medi- ans of all years after privatization with their values in the years before. The Brazilian economy experienced cycles over the course of the pe- riod during which privatization took place. Thus, changes in perfor- mance could reflect cyclical movements of the economy, rather than the impact of privatization. To circumvent this problem, in each method we also used, as an alternative procedure, a control group of private compa- nies. The performance of the privatized companies was adjusted by tak- ing the difference between the indicator for the privatized enterprise and the average of the indicator for the control group. Thus, we followed a procedure close to the one used by La Porta and López-de-Silanes (1999, p. 1211), who adopted, in their words, "industry-adjusted changes in performance for the sample of privatized firms."11 Table 4.3 summarizes the results in terms of their signs and statistical significance. The complete results are presented in appendix 4C. Profitability. In general, the results indicate an improvement in the prof- itability of the privatized companies. Considering the ratio of operating income to property, plant, and equipment (PPE), return on assets (ROA), and return on equity (ROE), performance after privatization improves re- gardless of the comparison method we used. The increase of operating in- come to PPE is evident once the change in the mean or median is always positive and significant, at least at the 10 percent level. The statistics for ROE and ROA are also always positive. In the case of ROE, three of the four statistics are significant, while for ROA only two reveal significance. A slightly different picture appears when we consider the ratio of oper- ating income to sales. Looking at the sign of the change, this indicator im- proves after privatization under the second method but not under the first and not in comparison with the private sector, when the change becomes negative and significant at the 10 percent level (appendix table 4C.2). Lit- tle can be said about the ratio of net income to sales. The sign of the coef- ficients varies across methods and fails to present statistical significance. At the firm level, various reasons could account for results of this kind. At this point, the method's weakness in investigating in detail the sources of variance becomes apparent, underscoring the importance of using a dif- ferent approach to test explanatory variables other than privatization, as is done shortly by using panel data analysis. Operating Efficiency. The results strongly suggest an improvement in effi- ciency. In all tables we observe an increase in the ratio of sales to PPE and COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 155 a reduction in the ratio of operating costs to sales. In the case of sales to PPE, all the statistics are positive and significant, strongly suggesting that privatized firms became more efficient in the use of their assets. Regarding operating costs to sales, all the statistics present a negative sign, while only one of them lacks significance at the 10 percent level. As illustrated in ap- pendix table 4C.1, operating costs dropped by one-third in the two years after privatization, compared with the two years before privatization, thus providing evidence of improved efficiency at the operational level. Assets and Output. Apparently, privatization had a negative impact on in- vestment. In all the tables, the Log (PPE) and investment-to-sales statistics present a negative sign. These results seem consistent with the increase in efficiency we found. When considering the ratio of investment to PPE, which reflects the rate of investment, there is no clear picture: the sign changes across the two methods and the private sector adjustment, but none of the statistics is significant. The effect of privatization on sales is a small but significant increase, observed in all tables. The statistics that test for difference in average are significant at the 1 percent level (appendix tables 4C.1 and 4C.3). There is a small increase in sales even after the adjustment to the performance of the private sector is made (appendix tables 4C.2 and 4C.4). Finance and Shareholders. With respect to the payout ratio--that is, the ratio of total dividends paid to net operating income--no conclusive evi- dence was obtained. The sign of the coefficient is consistently negative al- though never significant. A lack of information could be responsible for this finding, since this variable could be calculated only for a reduced num- ber of firms (45).12 A clearer picture emerges with the financial management indicators. We observe an increase in the ratio of current assets to current liabilities, both in absolute terms and in comparison with the private firms in our control group. The statistics for the difference in average are consistently positive and significant. Moreover, the adjusted mean and median are negative (ap- pendix tables 4C.2 and 4C.4), meaning that former SOEs, when compared with the control group, continued to present lower short-term solvency. The overall improvement indicates that SOEs, having government backing, are less concerned with achieving sound financial performance. With respect to the ratio of long-term debt to equity, we observe that when privatized firms are seen in isolation, privatization has a positive impact, as the coefficients are significant and show an increase (appendix tables 4C.1 and 4C.3). When compared with the performance of the pri- vate firms, however, the change in coefficients becomes negative, and a different picture emerges (appendix tables 4C.2 and 4C.4). In any case, in the same tables the mean values after privatization (0.108 and 0.002, respectively) indicate that the leverage of former SOEs converged to val- ues observed in the private sector. 156 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO The results regarding financial structure are similar to those reported by La Porta and López-de-Silanes (1999) and can be explained by the al- most null probability of insolvency of state-owned enterprises, since their credit status is guaranteed by the government. Having lost government backing, these firms were forced to adjust by decreasing their long-term- debt-to-equity ratio and increasing their current-assets-to current-liabilities ratios. Net Taxes. Our results indicate a clear decrease in the ratio of net taxes to sales. All the coefficients are negative and significant at the 1 percent level. There are two reasons for finding a clear and significant decrease in net taxes after privatization in Brazil. This variable is defined as the difference between calculated taxes and allowed deductions. Because the allowed de- ductions do not come in the form of explicit subsidies, it is worthwhile to describe them in detail to interpret the results more accurately. Three general categories of deductions apply: fiscal incentives, com- pensation for previous losses, and tax credits. Losses incurred in one par- ticular year may be deducted from income tax over several years. This, in particular, affected companies that were highly dollar-indebted when the devaluation of the Brazilian real occurred in early 1999. In fact, losses of this sort were also responsible for a decrease in net taxes even for the con- trol group in 2000. With respect to tax credits, an important dimension is the legal treat- ment of the premium paid on asset value in mergers and acquisitions. Brazilian corporate law recognizes the premium, and it was regulated in the mid-1990s. The acquiring company is allowed to set up a reserve ac- count for the premium and amortize it over a period of 5 to 10 years. When the reason for the premium paid over assets is based on expected fu- ture profits, the rebate is allowed for a period of up to 5 years. This bene- fit applies to mergers and acquisitions in general. Thus, both the overall private sector under restructuring and the privatized companies have been beneficiaries of these rebates. The existence of an explicit provision setting out the premium paid for concessions as expected future profits facilitates the use of this sort of tax credit in privatization. Therefore, reasonable explanations exist for our finding that net tax payments have decreased after privatization. In particular, evidence provided by individual data in the iron and steel industry supports this argument. On average, most of the firms in that in- dustry were subjected to Brazilian government rules concerning income tax before the privatization, which established the amount of tax that was due. After privatization, the pattern of this indicator changes a great deal. This result is readily observable since the individual data for this industry stretch over more years than any other in our sample, which allows us to state that Companhia Siderúrgica de Tubarão (CST) and Usiminas started paying more income tax after their privatization in 1993 and 1992, COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 157 respectively. Clearly, four out of six firms in this industry did the same as CST and Usiminas: Acesita, Aço Minas Gerais, COSIPA, and CSN, sug- gesting that the tax credit for premiums paid on assets would explain this unexpected decrease in taxes. Taken as a whole, these results support the view that privatization brought improvements in the performance of the firms. However, as pointed out, the mean and median tests leave room for a more compre- hensive analysis that fully utilizes the variance of the data set and allows for examining other aspects of the privatization process. This is the focus of the next section. Panel Data Analysis We start with a brief description of the technique used in this subsection. It is a dynamic version of panel data analysis and focuses on individual heterogeneities over time, in particular the discontinuous effect of pri- vatization. This approach is an alternative to generalizations of con- stant-intercept-and-slope models for panel data, which introduce dummy variables to account for effects of variables that are specific to individual cross-sectional units, but stay constant over time, together with the effects that are specific to each time period, but the same for all cross-sectional units. The analysis is dynamic because the lagged value of the independent variable is included in the model, and the panel is un- balanced as the data set is missing some observations for some firms. Many economic relationships are dynamic in nature, and another advan- tage of the panel data approach is that it allows for a better understand- ing of the dynamics of adjustment of a particular variable. However, the inclusion of a lagged dependent variable in the model causes problems, which are well known in the literature.13 Following it, we opted to apply the Arellano and Bond (1991) GMM-IV (generalized method of moments instrumental variables) method to estimate the parameters of the panel data model used in the empirical analysis.14 The Model and Variables. To assess the effect of privatization on each performance indicator listed in appendix table 4B.1, we relied on the fol- lowing econometric model: Iit i Iit1 Pit Xit Mpt eit, where: Iit represents the performance indicator for firm i in year t; Iit 1 represents the performance indicator for firm i in year t 1; Pit, referred to as PRIVATIZATION, is a dummy variable that as- sumes a value of 1 if company i had already been privatized in year t, and 0 otherwise; 158 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Xit is a set of control variables that are also firm specific (see appendix 4D for definitions of control variables); and Mpt, referred to as PRIVATE MEAN, is the mean value of the per- formance indicator for the group of privatized firms listed in appendix 4A, defined only over time, that is, assuming for every year the same value across the cross-sections of privatized firms. Empirical Results. The panel results are shown in table 4.4. Before dis- cussing them, it should be noted that it is possible to decompose the Xit vector of the model into two groups. The first group comprises the dummy variables that are associated with the environment confronted by the firms and that are effective for all companies in the sample (TRADABLE, REG- ULATED, and LISTED). The other contains firm-specific variables that affect only the companies that are the focus of the study (PRIVATIZA- TION, SPLIT/MERGER, and MINORITY CONTROL). In the case of firms operating in TRADABLE industries, one observes the predominance of an inferior performance compared with those in nontrad- able industries. This is found in the indicators of profitability, operational efficiency, output, sales, and indebtedness. Firms also seem to pay higher net taxes to sales. The reason for an underperforming tradable sector is that, for most of that period, the country was promoting trade liberalization, a process aggravated from 1995 to 1998 by an exchange rate overvaluation. Because the effects of REGULATION are likely to be different in the various industries, a more detailed analysis would be required to inves- tigate them. In any case, the comparative analysis of regulated industries versus unregulated ones shows a slightly better performance among reg- ulated industries in terms of profitability, investment, sales, and indebt- edness. A possible explanation for this is that regulation encouraged companies to improve their performance, in particular, by setting more realistic prices for the regulated activities. For the privatized companies that underwent restructuring in the form of SPLIT/MERGERS, no discernible effect was found on profitability indicators. Moreover, the same companies present evidence of inferior results in terms of operational efficiency, assets and outputs, indebtedness, and net taxes. Notice that this dummy variable is in effect only for the period in which the intervention occurred and it is associated with the pri- vatization intervention. So the inferior results would have to be interpreted with respect to the performance of firms that were privatized without restructuring. There is an open debate on the virtue of government-led adjustments in debt, labor force, and firm activities before privatization (Megginson and Netter 2001). The MINORITY CONTROL dummy shows two significant coeffi- cients for profitability performance indicators, but it also shows mixed re- sults for operational efficiency, with a relatively large ratio of sales to PPE and a somewhat high ratio of operational costs to sales. Firms privatized Table 4.4 Changes in Performance: GMM-IV Panel Data Analysis Performance indicators Log Log Log LTD/ Variable OI/S OI/PPE NI/S ROA ROE (S/PPE) OC/S (PPE) I/S I/PPE (S) Current E NT/S Privatization 0.056*** 0.033 0.003 0.016*** 0.062*** 0.070*** 0.015*** 0.012*** 0.032*** 0.057*** 0.008 0.140*** 0.029* 0.006*** 0.008 0.055 0.005 0.003 0.005 0.009 0.003 0.004 0.009 0.010 0.004 0.015 0.020 0.001 Tradable 0.001 0.030*** 0.006*** 0.007*** 0.030*** 0.026*** 0.005*** 0.005** 0.003 0.034*** 0.010*** 0.019 0.028*** 0.006*** 0.003 0.007 0.002 0.002 0.004 0.005 0.002 0.002 0.005 0.007 0.003 0.017 0.010 0.0005 Regulation 0.032*** 0.013 0.003 0.0006 0.011*** 0.030*** 0.035*** 0.027*** 0.003 0.015 0.024*** 0.056 0.130*** 0.004*** 0.005 0.016 0.003 0.003 0.005 0.007 0.002 0.003 0.007 0.009 0.003 0.019 0.017 0.001 Split/mergers 0.028 0.070 0.010 0.0007 0.005 0.041*** 0.032*** 0.018*** 0.055*** 0.066*** 0.004 0.102*** 0.235*** 0.004** 0.039 0.105 0.010 0.004 0.004 0.010 0.006 0.005 0.020 0.012 0.004 0.022 0.022 0.001 Minority 0.016 0.066 0.040*** 0.007 0.026** 0.065*** 0.008* 0.011 0.021 0.065*** 0.016*** 0.348*** 0.137*** 0.023*** control 0.018 0.097 0.011 0.007 0.011 0.012 0.005 0.009 0.022 0.022 0.005 0.062 0.041 0.005 Listed 0.057* 0.063 0.067*** 0.018*** 0.029*** 0.035*** 0.016*** 0.008* 0.122*** 0.115*** 0.007** 0.016 0.085*** 0.004*** 0.034 0.065 0.014 0.003 0.005 0.013 0.006 0.004 0.020 0.011 0.003 0.016 0.024 0.001 Private 1.287*** 1.070*** 0.425*** 0.751*** 0.710*** 0.520*** 0.770*** 0.101*** 0.954*** 0.970*** 0.115*** 0.635*** 0.887*** 1.070*** mean 0.053 0.041 0.040 0.040 0.037 0.037 0.061 0.008 0.032 0.024 0.008 0.055 0.030 0.067 Lagged 0.195*** 0.080*** 0.556*** 0.035*** 0.172*** 0.831*** 0.555*** 0.912*** 0.304*** 0.075*** 0.926*** 0.458*** 0.186*** 0.040*** variable 0.019 0.020 0.015 0.005 0.016 0.007 0.017 0.004 0.012 0.013 0.006 0.013 0.041 0.011 Exchange 0.020*** 0.012*** ratea 0.003 0.004 Constant 0.137*** 0.015 0.062*** 0.022*** 0.070*** 0.210*** 0.022*** 0.029 0.120*** 0.164*** 0.220*** 0.299*** 0.122 0.004 0.034 0.052 0.015 0.004 0.008 0.019 0.014 0.043 0.024 0.0187 0.050 0.080 0.032 0.003 N 1798 2158 1960 2257 1903 2044 1580 2561 1702 2185 2073 2120 2256 1598 Pseudo R2 0.332 0.352 0.441 0.397 0.468 0.554 0.561 0.373 0.491 0.538 0.725 0.584 0.610 0.447 159 (Table continues on the following page.) 160 Table 4.4 (continued) Performance indicators Log Log Log LTD/ Variable OI/S OI/PPE NI/S ROA ROE (S/PPE) OC/S (PPE) I/S I/PPE (S) Current E NT/S Sargan Testb 0.000 0.000 0.001 0.002 0.000 0.000 0.005 0.000 0.010 0.004 0.000 0.000 0.000 0.012 (Prob 2 ) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: GMM-IV generalized method of moments instrumental variables. For definitions of the performance indicators, see appendix 4B. The table shows the estimated coefficients for unbalanced panel data regressions, corresponding to each performance indicator. See text for explanation. Error terms are shown in parentheses. a. Dummy variable introduced in order to overcome a devaluation bias in those variables that are defined in absolute terms after 1999. Though this effect has caused impacts in all of the performance indicators, they can be overcome for those which are defined as a ratio of two variables. b. Sargan Test for Over-Identifying Restrictions. All the null hypotheses were rejected, validating the use of the instruments chosen. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 161 with these characteristics pay the highest net taxes relative to sales of all privatized firms and have lower indebtedness, indicating that they did not benefit as much from the corporate tax credits that came with privatiza- tion and that they had inferior access to credit compared with other pri- vatized companies. The dummy LISTED reveals a clearer positive effect for all criteria of performance. In particular, coefficients for four of the five profitability indicators presented positive and significant signs. Listed companies also showed better indicators for operational efficiency, output, and assets, as well as the largest indicator for long-term debt to equity. The net-tax-to- sales indicator has a negative sign again, indicating corporate tax benefits for listed companies. Thus, the results presented by this dummy caution against the bias in selecting only firms listed on the stock exchange for privatization studies. It is possible now to assess the net effects of PRIVATIZATION as a change in the intercept of each indicator. In general, its impact comes as hypothesized and stronger than the one revealed by the mean and median tests. As it is the key variable under investigation, the results for the vari- ous indicators of performance are discussed in more detail. Profitability. Three out of the five indicators of profitability presented in table 4.4 clearly reveal the improvement that comes with privatization, as its estimated parameters are positive and significant at the 1 percent level. Privatization coefficients for the ratio of operating income to sales (OI/S), return on assets (ROA), and return on equity (ROE) show an increase of 5.6 percent, 1.6 percent, and 6.2 percent, respectively. Operational Efficiency. Significant at the 1 percent level, the coefficients of the privatization dummy show the expected sign, that is, an increase of 7 percent in the ratio of sales to PPE, or Log(S/PPE), and a reduction of 1.5 percent in the ratio of operating costs to sales (OC/S). Assets and Output. Significant PPE and sales are the only indicators measured by their absolute values, which are measured in dollars. The Brazilian currency suffered a major devaluation early in 1999 that was not reversed in 2000. To capture the negative effect on these indicators, we introduced a dummy variable taking the value of 1 in those years, and 0 otherwise. For both Log(PPE) and Log(S) the estimated coefficients were negative and significant at 1 percent. In the Log(S) equation ad- justed in this fashion, the impact of privatization on sales is smaller (0.8 percent) than in the Log(PPE) equation. Even after taking devaluation into account, the privatization coeffi- cient for the property, plant, and equipment indicator, or Log(PPE), is negative, indicating a reduction of 1.2 percent in the productive assets of the firms. This result is consistent with the coefficients for privatization with respect to other asset indicators. An increase in the ratio of sales to 162 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO PPE, or Log(S/PPE), intensifies the use of productive assets, so a reduc- tion of 3.2 percent in the investment-to-sales (I/S) ratio is likely. As the ratio of investment to PPE (I/PPE) shows a positive coefficient of 5.7 per- cent for privatization, the indication is that investments after privatiza- tion are moving into working capital. The indication of an increase in investments in the form of working cap- ital is confirmed by a strong impact of privatization on the ratio of current assets to current liability: a 14 percent increase, significant at the 1 percent level. With respect to the ratio of long-term debt to equity (LTD/E), a likely outcome is that the privatized companies will seek to reduce the cost of cap- ital, combining equity and debt in an efficient way. Conversely, state-owned enterprises are likely to increase debt, saving the national treasury from in- vesting in their equity as their credit status, guaranteed by the government, has a small probability of default. This situation may lead to large LTD/E ratios. After privatization and the loss of government backing, privatized firms are forced to adjust by decreasing this ratio and increasing the ratio of current assets to current liabilities. Accordingly, the privatization coefficient for LTD/E shows a reduction in indebtedness of 2.9 percent, significant at the 10 percent level. It is interesting to observe the LTD/E coefficients esti- mated for SPLIT/MERGERS ( 23.5 percent) and MINORITY CONTROL ( 13.7 percent), which magnify the impact of losing government backing. With respect to net taxes, the coefficient of privatization is negative and significant at the 1 percent level. The reasons are those presented in the mean and median tests, now confirmed by a panel data analysis. Other Variables in the Model. The coefficient of the private mean is pos- itive and significant for all indicators, reflecting the impact of overall business and macroeconomic conditions. It also cautions against another distortion of some studies on privatization in which the impact of priva- tization from the changes in these conditions over time is not isolated. The coefficients of the lagged variable, instrumented by its two-period version, are all positive and significant, revealing that the past behavior of firms' indicators has a strong influence on their current performance. On top of this effect, other variables exert influence, such as those encountered above. Finally, we emphasize once more that the obtained evidence on de- creasing tax revenue, mainly after privatization, is a special phenomenon due to tax credits. It therefore does not reflect a permanent reduction in income for the Brazilian government. Other Benefits and Costs of the Program The improvement in the performance of privatized firms shown in the pre- vious section can be viewed as a benefit, as it contributes to the efficiency of the economy as a whole. This section addresses other benefits, as well COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 163 as some costs, of the privatization program. It also seeks to identify some sources of the gains made by privatized firms in the form of reductions in employment and increases in prices. Employment One of the weaknesses of the Brazilian data is that no comprehensive, reli- able, and unified record of the number of employees exists for the priva- tized companies either before or after their sale. Financial statements and annual reports, including those of listed firms, are not required to include information on employment, and companies provide it at their own dis- cretion. There are also no uniform requirements for including payroll in- formation in these reports and statements, which bundle wage and salary costs together with other operational costs. Even when employment and payroll data are available, their analysis is handicapped for other reasons. In Brazil, there are strong incentives for the adoption of outsourced services, such as security, cleaning, maintenance, and accounting. Outsourcing has become a widespread means of reducing labor costs, as service providers are usually smaller firms and pay lower wages. In addition, one often finds workers disguised as business owners to avoid heavy taxation of wages and salaries.15 Most workers prefer for- mal contracts with employers; firms and unions also press for this and are more successful with SOEs. It is therefore very likely that privatization has led to an extension of outsourcing. Thus, a reduction of employment in a company would not necessarily mean a reduction in the jobs generated by its activities along its chain of suppliers.16 Given this picture, we first examine the employment effects at the in- dustry level, where there are aggregate data. Then, for a limited number of former SOEs, we focus on employment data from the files of Exame, a business magazine that collects financial statements and reports of Brazilian firms, as well as scattered employment data from the magazine and other sources. In Brazil the most important source of data on formal employment is the Annual Survey of Social Data (RAIS) from the Ministry of Labor and Employment. All firms and the government are required annually to list workers and their characteristics such as year of entrance, white-collar versus blue-collar, unionized, level of education, and so forth. Individual firms cannot be identified in the samples. This source has consistent data for the period 1995 to 1999. Table 4.5 shows data on employment for the industries in which the most important privatizations have occurred. One can see that until 1997 the private sector was responsible for less than one-twentieth of employ- ment in electricity, about one-third in water and sewage, a quarter in telecommunications, and a fifth in piped gas distribution. By 1999, in both the telecommunications and the gas distribution sectors, the larger part of 164 Table 4.5 Employment in Selected Industries, by Public or Private Ownership, 1995­99 1995 1996 1997 1998 1999 Public Private Public Private Public Private Public Private Public Private Industry (%) (%) Total (%) (%) Total (%) (%) Total (%) (%) Total (%) (%) Total Mining 18 82 39,131 18 82 38,060 1 99 31,447 1 99 39,955 1 99 35,763 Petroleum 76 24 14,442 82 18 21,546 72 28 16,963 62 38 13,923 39 61 10,590 Fertilizers 18 82 6,460 9 91 7,145 11 89 8,395 1 99 12,563 1 99 11,907 Petro- 5 95 15,739 2 98 14,947 0 100 19,018 1 99 26,263 1 99 28,935 chemicals Iron and 5 95 376,220 5 95 369,234 2 98 385,064 2 98 429,965 2 98 446,949 steel Electricity 97 3 149,100 97 3 128,545 95 5 99,871 64 36 111,225 55 45 95,870 Gas 92 8 3,257 89 11 2,640 83 17 1,551 60 40 1,763 31 69 1,437 distribution Water and 68 32 135,313 72 28 146,791 66 34 159,588 66 34 145,375 62 38 149,822 sewage Telecom- 80 20 107,689 77 23 113,126 75 25 117,740 19 81 105,284 26 74 109,478 munications Note: Number of employees as of December 31. Source: Ministry of Labor and Employment, 1995 Survey of Social Data. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 165 employment moved to private companies. In the electricity and water and sewage sectors, employment is still largely in SOEs and public enterprises but now with a significant mix of private firms. The table shows a clear reduction in employment following privatiza- tion in the electricity and piped gas distribution industries. In telecommu- nications, the effect of privatization on reducing employment is less clear, in part because the provision of services expanded very rapidly after priva- tization. Worth mentioning is the case of the water and sewage industry: still largely in the hands of the government and not expanding as quickly as telecommunications, its employment ranks high in stability among the industries shown in the table. Note also the recovery of employment in petrochemicals and in iron and steel, showing that after employment ad- justs following privatization, the growth of investment and production leads to new jobs. The employment data from Exame's files cover companies privatized in the period 1995­2000; some data are missing, but enough are available to allow a comparison between the pre- and postprivatization years. Figure 4.1 shows data for 49 companies in the form of a box-plot diagram. The impact of privatization on employment emerges clearly from the plotted data, with 43 companies showing a reduction in employment after priva- tization and only 6 revealing an increase. Tests were performed by taking the average number of employees in at least two years before and after pri- vatization. The Wilcoxon signed rank test, calculated for the difference in means, is equal to ­5.217 and significant at the 1 percent level. A para- metric t-test is also calculated and equals 3.906, which is significant at the 5 percent level. Our conclusion is that a share of the costs of privatization has been borne by some of the workers directly employed by the former SOEs who lost their jobs either in the process of adjustment for the sale or thereafter. This is an inevitable outcome of privatization as new owners seek higher efficiency. Thus, this reduction in employment was one of the sources of gains from privatization. However, as privatized firms invest and expand their activities, at some point employment begins to increase, although the same workers are not necessarily rehired, and some of them might con- tinue to suffer the costs of displacement and reallocation. In the Brazilian case, the widespread use of outsourced services often blurs the picture, as positive impacts are not necessarily captured by the direct employment data of privatized firms, particularly in the telecommunications industry. Prices Following privatization, newly established regulatory agencies moved to en- courage more realistic prices, particularly in the areas of electricity and telecommunications. The government had to announce this policy during the privatization process to guarantee the success of the auctions. Moreover, 166 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Figure 4.1 Formal Employment before and after Privatization, 1995­99 Number of employees (thousands) 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 N= 49 49 Employment before Employment after Note: The figure shows a box plot for the lower and upper bounds of the number of employees in our sample before and after privatization. The shaded region represents the percentile while the black line inside represents the median. The two lines at the extremes are the percentiles, and the stars stand for outlier observations beyond two standard deviations. See the text and table 4.5 for a more detailed explanation. Source: Authors' calculations. the overvaluation of the Brazilian real and the trade liberalization reforms undertaken since the early 1990s meant that the tradable industries were ex- posed to increased competition. To show some of the relevant changes in relative prices, we made a comparison of various price indexes at the industry level with an overall price index, the CPI-A calculated by the Brazilian census bureau. We took COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 167 August 1994 prices as a reference for the other indexes. In the tradable in- dustries, such as iron and steel, nonferrous metals, nonmetallic minerals, fertilizers, and plastics, which are greatly affected by the overvaluation of the real, domestic prices lagged behind CPI-A variations for the years 1994 to 1998. After devaluation, prices in these industries clearly caught up to the CPI-A. This sheds light on one of the results of the previous section, where it was found that firms in these industries had shown a shakier per- formance than firms in other industries, in part because of the restraint im- posed by the overvalued real. For telephone rates, the price effects started when the telecommunica- tions industry was being prepared for privatization, as early as 1996. In particular, the minimum monthly fee for access to a line increased sharply. This has been a source of gains to the telephone companies, but no one in Brazil would dispute that it was followed by a massive expansion of serv- ices to the point of destroying the market that previously existed for trad- ing telephone lines, at prices sometimes reaching $2,000­$3,000, or even more when the dollar was overvalued. In electricity, the rate restructuring began in 1995. Privatization itself started in 1997, and the concessionaries signed an incentive contract with a clause allowing a pass-through of noncontrollable costs. Thus, with the devaluation in 1999, they were allowed to adjust prices for the dollar- denominated contracts they had, for instance, with suppliers from Paraguay. Our conclusion is that prices have been a source of gains to privatized firms in the telecommunications and electricity industries. Regulation combined with privatization made prices follow contracts and other rules, thus reducing the scope for political manipulation that existed when the government played a larger entrepreneurial role in these industries. In the telecommunications industry, this government role practically ceased to exist, but it is still strong in electricity, particularly in power generation. A Social Cost: No Democratization of Capital Ownership Macedo (2000) points out that some groups in Brazilian society were excluded from the privatization auctions and therefore from the oppor- tunities to gain from them. As a rule, the privatization program did not resort to public offers to any significant degree. Moreover, some public sector liabilities could have been exchanged for shares of the SOEs being privatized. Among these liabilities are the unfunded ones of the present and future pensioners of the social security system and the deposits that formal workers hold in their accounts of the Workers' Tenure Guaran- tee Fund. This fund, known as FGTS (Fundo de Garantia de Tempo de Serviço), accumulates on a monthly basis a percentage of wages and salaries, to be used in case of termination or dismissal of the workers. Macedo's conclusion was that because of this discrimination, the privati- zation process failed in one of its stated objectives: democratizing capital 168 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO ownership in Brazil. Only recently were workers allowed to use their FGTS deposits in successful public offers for a block of Petrobrás shares and another block of remaining state-owned shares of the Vale do Rio Doce mining company. Effects on the Development of Capital Markets A goal of the program was to maximize the revenue from sales. Many of the former SOEs were structured as public companies and therefore sub- ject to laws governing the stock market. Before 1996 minority investors in Brazil were protected with features such as "tag-along" (giving mi- nority investors the right to sell their shares at the same price as the man- aging block in case of change in control) and oppressed minority rights (having their shares bought back at book value in cases of restructuring, such as mergers or divestitures). Because some companies had to be re- structured for privatization (for instance, Telebrás, the state holding company for telecommunications, was split into 12 different firms), there was a fear that lawsuits from minority shareholders might hamper the privatization process or reduce the revenues from auctions, or both. This concern led the government to reform the legislation in such a way that the amendments to the corporate law revoked the tag-along and the oppressed minority rights clauses. To mitigate the impact, the legislation entitled nonvoting shares to an additional 10 percent in dividends over those paid to voting shares. In any case, as the postprivatization experi- ence has shown, without the protective clauses, minority shareholders have in several cases been victims of controlling groups' opportunistic behavior. At the end of the 1990s, influential works such as those of La Porta and others (1997), Levine (1997), and Levine and Zervos (1998) helped to confirm the view that the development of capital markets is important to promote economic growth, and that protecting minority investors' rights is the best way to promote capital markets. In 2000 and 2001 the Brazilian Congress discussed a bill to increase minority shareholders' rights, including restoring the tag-along and oppressed minority rights. Unfortunately, the new law that emerged establishes tag-along for only 80 percent of the minority shares, and the new oppressed minority rights have been extensively criticized as inadequate. Thus, the adverse effect of privatization on stock markets is likely to last. A Macroeconomic Cost: No Effective Debt Reduction and Delayed Devaluation Macedo (2000) also claims that privatization had a macroeconomic cost because the generated revenues to the government budget, and to the ex- ternal accounts through foreign direct investment, delayed a genuine fiscal COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 169 adjustment and the necessary devaluation of the real. It is important to un- derstand the details of this argument because it warns of the risks of mis- using privatization resources in conditions of fiscal and external imbal- ances and in the presence of soft budget constraints. Privatization was intended to help the fiscal crisis and the external im- balance, but this intended benefit was lost because in its first term (1995­98), the administration of Fernando Henrique Cardoso increased the fiscal deficit. Moreover, the new currency, the real, had clearly be- come overvalued immediately after its release in 1994. With its political capital linked to price stabilization, the government opted for defending the real, afraid of the impact of devaluation on prices. Very high interest rates were the main policy instrument. These developments had the ef- fect of seriously aggravating the budget deficit and debt, the payment of debt interest, and the external imbalance. Thus, public debt increased from 29.2 percent of gross domestic product (GDP) in 1994 to 52.5 per- cent in 2001; debt interest grew from 5.8 percent of GDP in 1996 to 11.8 percent in 2001; and the current account deficit went from less than 0.5 percent of GDP in 1994 to around 4 percent in 1997 and has remained at that level.17 Thus, the privatization program did not accomplish the objective of reducing public debt. On the contrary, public expenses increased, more than offsetting the inflow of resources from the privatization auctions.18 Macedo (2000) also argues that privatization allowed the financing of higher current account deficits, particularly in 1997 and 1998 when the privatization program peaked and attracted substantial foreign invest- ment. Although this inflow is usually considered a positive consequence of the process, it also contributed to the postponement of a badly needed devaluation of the Brazilian currency.19 Privatization as a Tool for Imposing Fiscal Discipline: The Case of the States The fiscal policies of Brazilian states also contributed to the ballooning fis- cal deficits and debt from 1994 to 1997. Tight public-sector budgets as a whole came only after 1998, when the size of the debt started to cause dis- comfort in the financial markets, and the external imbalance continued to deteriorate. The federal government then started to generate huge primary surpluses and was also increasingly able to impose fiscal discipline on the states. Privatization of the states' assets played an important role in this process. The states had their debt transferred to the federal government, to which they became indebted themselves, but at more favorable interest rates. To obtain this benefit, the states had to make commitments to re- strain further indebtedness on their part and also to privatize. Thus, the federal government was able to impose a tight fiscal constraint on the states that it had not adopted itself. 170 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Public Opinion Privatization has not been popular in Brazil. A 2001 Latinbarómetero public opinion survey conducted in 16 Latin American countries reported that 53 percent of the respondents in Brazil believed that privatization had not been beneficial to the country (Lora and Panizza 2002). Nonetheless, Brazilian public opinion about privatization was found to be more favor- able than that of its neighbors: on average 63 percent of the respondents in all the surveyed countries believed that privatization had not been ben- eficial to their nations. The countries in which the public appeared to be less discontented with privatization were Chile (47 percent expressed dis- satisfaction) and Venezuela (46 percent). For all the other countries, ap- proval ratings were lower than in Brazil.20 Several factors contributed to the unpopularity. In most cases the av- erage citizen is not able to identify fully the benefits of privatization such as those analyzed in this chapter. The creation of nonbanking SOEs in Brazil, such as steel and mining, followed the Second World War, when the main motivation was the belief that the state had to play a major role in "strategic" industries, the products of which tend to be remote from the pressing concerns of the population. Thus, one cannot expect the pub- lic to be concerned with the outcome of privatization in these industries or to be inclined to evaluate its technicalities. The total privatization of the telecommunications industry and the partial privatization of the electricity SOEs produced mixed outcomes for consumers. Both were followed by higher rates, which have blurred the favorable impact of a major expansion of telecommunications serv- ices. In electricity, a further negative impact emerged in 2001, when the country had to face rationing due to the low levels of the reservoirs of the hydroelectric plants, which constitute the basis of power generation in the country. Opponents of privatization were eager to blame it for the crisis. It is also important to highlight that privatization coincided with slug- gish growth, particularly after the program peaked in 1997­98. Therefore, dissatisfaction with lower economic gains or even losses, such as those emerging from the higher rates of unemployment, are likely to have de- veloped into criticisms of government policies in general and privatization in particular. Moreover, as already pointed out, the government failed in its objective of using the program to democratize capital ownership. Only recently has it resorted to successful public offerings in which workers were entitled to participate by using their FGTS deposits. Thus, as a rule the common cit- izen was left out of the process and its benefits in the form of rewards to the controllers and shareholders. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 171 Opposition to the program also resulted from its unfavorable treat- ment in the media, court battles to impede the auctions, attempts to disrupt the auctions themselves, sometimes followed by police inter- vention, and so forth. The news coverage of the privatization of the telecommunications industry was particularly negative, as there were accusations that some government authorities had been involved in arm-twisting to attract and assemble groups to participate in the auc- tions. Recorded tapes of conversations held by government authorities among themselves and with interested parties reached the press. Even though the legal battles were decided in favor of privatization, the up- roar was serious enough to cause the minister of communications to re- sign in November 1998. News of this sort has inevitably aroused suspicions that the process has been tainted by wrongdoings. Lora and Panizza (2002) found that oppo- sition to privatization, again measured by the percentage of those who do not consider it beneficial, was lower in Brazil than in its neighbors. Gen- erally approval ratings are higher in those countries with extensive priva- tization and limited corruption. In this respect, Brazil ranks second only to Chile in an evaluation involving the above-mentioned group of 16 coun- tries. In any case, although privatization is faring better in relative terms, clearly it is not popular in the country, a finding that is not surprising given the reasons discussed here. Although the privatization program has not been popular in general, a different picture emerges from a study by Lamounier and De Souza (2002), which focused only on the opinion of a group called the "Brazil- ian elites," composed of 500 businessmen (including leaders of associations of small and medium firms), union leaders, congressmen, high-ranking members of the executive and judiciary branches of government, journal- ists, religious leaders, directors of nongovernmental organizations, and intellectuals. On average, 62 percent responded that they approved or tended to approve of privatization. The rates ranged between 87 percent for members of the executive branch of government to 13 percent for union leaders, whose rate was the only one below 45 percent. Another question was directed at the performance of the companies after privati- zation. In this case, the approval rate (percentage of good or above) showed large variations by industries.21 Perspectives The overall unpopularity and its causes are among the reasons for the privatization program's having come to a virtual standstill since 1998. According to BNDES, the proceeds from the auctions, including new 172 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO concessions of public services, fell from $26.3 billion in 1997 and $35.7 billion in 1998 to $4.2 billion in 1999, $10.2 billion in 2000 (this figure includes the privatization of a state bank, Banespa, which totaled $3.6 billion and had been in the pipeline for a long time), $2.8 billion in 2001, and $2.2 billion in 2002.22 Other factors also explain the current status of the program. First, mov- ing ahead would mean including those SOEs that have stronger political patronage than those privatized thus far. For example, on the list of re- maining SOEs is the almost two-century-old Banco do Brasil, a commer- cial bank of which the federal government is the controlling shareholder. It holds the government's accounts and is the key player in providing agri- cultural credit, which is subsidized by the federal budget. As a result, it has built a major constituency, as private banks have refrained from being more active in agricultural credit. Its staff, traditionally selected by public examinations, is a breeding ground for government officers. Some of them have reached the ministerial level or have become members of Congress and are very influential. Moreover, the bank is not wholly state-owned but also has private shareholders who act as a group to maintain its current status. Another example is the giant oil company Petrobrás. The company was established in 1954, following a strong nationalist stance against foreign oil companies. Petrobrás proved effective in finding oil in Brazil. It moved into offshore drilling in the 1980s and has set worldwide records in deep-water exploitation. Domestic production that currently accounts for 90 percent of the country's needs is seen as a sign of suc- cess. It had a monopoly in prospecting, production, and importing in the upstream market until 1995. Since then, it continues to have a vir- tual monopoly in these activities, as well as in refining. Because oil is as- sociated with national security issues, the military sees continuing gov- ernment control of Petrobrás as crucial. Moreover, the company also has private shareholders who support its current and very profitable status. In the electricity industry, the privatization process occurred mainly in the distribution sector. A few important companies in this sector were kept by state governments unwilling to move in the direction of privatization. With respect to the generation segment, the state of São Paulo privatized a large part of its assets. At the federal level, only one subsidiary of a fed- eral holding company, Eletrobrás, was privatized. The three remaining subsidiaries control around 60 percent of the country's generation. After the 2001 drought, which led to rationing, the process of sector restruc- turing stalled. The rationing stimulated industry and households to adopt energy-saving measures, and in the aftermath demand has not recovered its previous levels. Both rationing and demand reduction brought losses to the industry, exacerbating the dollar indebtedness of some privatized com- panies since 1999. With distribution and generation companies currently COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 173 suffering huge losses, the federal government, which regulates the entire industry, is preparing a new sector arrangement. At the same time, BNDES has to find a way to manage huge debts by companies on the verge of default. Thus, it is an industry in disarray, not attractive to private in- vestors, and in need of reorganization before any discussion of a new round of privatizations. Despite these shortcomings, there are no plans to reverse the privatiza- tion that has occurred in Brazil, either at the moment or in the foreseeable future. The new federal government, inaugurated in 2003, is for the first time led by the Worker's Party, which won the presidential election as an opposition party. It fought privatization in Congress and in the courts in the 1990s, but since taking power it has adopted conservative fiscal and monetary policies and avoided condemning privatization. In this context, there is no room for privatization reversal, nor has the government even been suggesting it in discourse. Apparently, the government is likely to keep the program stalled, that is, no privatization reversal but no further advances. Even in light of these new political developments, the possibility of re- suming the privatization effort should not be ignored: a serious fiscal problem remains in the form of large and difficult-to-manage public deficits and debt. They have been kept under control at the cost of huge increases in the tax burden, which moved from 25.7 percent of GDP in 1993 to a record level of 35.9 percent in 2002, an exceptionally high rate for a developing country.23 Under such conditions, a new start of the pri- vatization program could help to alleviate the fiscal accounts. Moreover, since the new government has been willing to reconsider many of the cher- ished dogmas it subscribed to when in opposition, there is a chance that even its current stance against new privatization efforts might be recon- sidered as well. Thus, to give new life to privatization, it is important to continue monitoring the process and publicizing the results of the program and the inefficiencies of the remaining SOEs. In addition, the objective of democratizing capital ownership by means of public offers should be brought to the front line, both for its own merits and to attract wider po- litical support, in particular by making privatization more appealing to President Lula da Silva's government. Summary and Conclusions This chapter has focused mainly on the changes in the performance of companies that have been privatized in Brazil since 1991. It confirmed pre- vious findings that the firms became more efficient after privatization. It has contributed to the literature, first by bringing to a wider audience stud- ies available only in Brazil and in Portuguese. It is also more up-to-date than previous studies, since it covers data up until 2000. In terms of the 174 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO companies covered, it is the most comprehensive study thus far. In the sample, a selection bias was avoided by including both large and small firms, as well as those listed and unlisted on the stock exchange. All com- panies for which information was available have been included in the analysis. In addition to tests of means and medians, the research also re- sorted to panel data analysis in an attempt to utilize fully the information provided by the data. Moreover, the analysis of performance before and after privatization was also compared with that of the private sector, taken as a control group over time. Finally, this study was undertaken by an in- dependent team, whereas most of the previous ones were done by staff members of BNDES. In addition to the findings of improved efficiency, the chapter has iden- tified some sources of gains made by privatized firms in the form of re- duced direct employment and increased prices. The chapter has also shown costs in the sense that the benefits of privatization could have been higher had the government not used the revenues to sustain its misguided policy of enlarging fiscal deficits and adopting high interest rates to defend the real. Moreover, foreign investment attracted by privatization and the high interest rates also contributed to the postponement of devaluation. In any case, what is to blame are the fiscal, interest rate, and exchange rate policies, not privatization itself. The benefits could also have been greater had the government not neg- lected the opportunity privatization offered for democratizing capital ownership. In the capital markets, privatization also brought costs in the form of reducing the rights of minority shareholders, thereby hampering the development of these markets. The study has also shown that although there is evidence that a ma- jority of the Brazilian elite approve of privatization, the majority of the general population does not view privatization as beneficial, as revealed by public opinion surveys. After pointing out some of the reasons behind this unpopularity and looking at the current status of the program, the chapter concludes that the door to new privatization efforts remains open. One suggestion is to give the program popular appeal in the form of public offers in which workers would be entitled to participate with their own financial assets, including the deposits they hold in the FGTS. With respect to future research, it is particularly necessary to further clarify costs, to look at the impact of privatization at the industry level, and to examine the role of the regulatory agencies that have emerged in the wake of the state's backing away from its role as an entrepreneur. To conclude, we return to Megginson and Netter (2001), quoted at the start of this chapter. The Brazilian privatization program is indeed likely to remain influential because of its scale and the size of the country. We hope that privatization will continue to have an impact because of the suc- cesses and benefits of the program, not for the mistakes that have been made. Appendix 4A Table 4A.1 Federal State Enterprises Privatized, 1991­2000 Listed Date of Auction result Included in before Auction Company name auction (US$ million)a sampleb privatizationb USIMINAS Usinas Siderúrgicas de Minas Gerais 10/24/91 2,310 1 1 (Usiminas) Usiminas Mecânica (Usimec)c 0 0 CELMA Cia. Eletromecânica 11/01/91 96 0 0 MAFERSA Mafersa S.A. 11/11/91 50 0 0 COSINOR Cia. Siderúrgica do Nordeste 11/14/91 15 0 0 (Cosinor) Cosinor Distribuidora (Cosinor Dist.)c 0 0 SNBP Serviço de Navegação da Bacia do 1/14/92 12 0 0 Prata INDAG Indag Fertilizantes 1/23/92 7 0 0 AFP Aços Finos Piratini 2/14/92 109 1 0 PETROFLEX Petroflex Indústria e Comércio S.A. 4/10/92 255 1 1 COPESUL Cia. Petroquímica do Sul 5/15/92 871 1 1 CAN Cia. Nacional de Álcalis 7/15/92 87 0 0 Álcalis Rio Grande do Norte 0 0 (Alcanorte)c CST Cia. Siderúrgica de Tubarão 7/16/92 to 837 1 1 7/23/92 175 NITRIFLEX Nitriflex 8/6/92 35 1 0 FOSFÉRTIL Fertilizantes Fosfatados S.A. 8/12/92 226 1 1 (Table continues on the following page.) 176 Table 4A.1 (continued) Listed Date of Auction result Included in before Auction Company name auction (US$ million)a sampleb privatizationb POLISUL Polisul 9/11/92 188 0 0 PPH PPH 9/29/92 94 0 0 GOIASFÉRTIL Goiás Fertilizantes S.A. 10/8/92 22 0 0 ACESITA Cia. Aços Especiais Itabira 10/23/92 697 1 1 Acesita Energética (Energética)c 0 0 Forjas Acesita (Fasa)c 0 0 CBE Cia Brasileira de Estireno 12/3/92 11 1 0 Poliolefinasc 3/19/93 87 0 0 CSN Cia. Siderúrgica Nacional 4/2/93 2,028 1 1 Fábrica de Estruturas Metálicas S.A.c 0 0 ULTRAFÉRTIL Ultrafértil S.A. Indústria e Comércio 6/24/93 226 1 0 de Fertilizantes COSIPA Cia. Siderúrgica Paulista 8/20/93 1,470 1 1 AÇOMINAS Aço Minas Gerais S.A. 9/10/93 721 1 0 OXITENO Oxiteno 9/15/93 56 1 1 PQU Petroquímica União S.A. 1/25/94 328 1 1 ARAFERTIL Arafértil Fertilizantes ­ ARAFÉRTIL 4/15/94 13 0 0 CARAÍBA Mineração Caraíba LTDA 7/28/94 6 1 0 ACRINOR Acrinor 8/12/94 13 0 0 COPERBO Coperbo 8/16/94 32 0 0 CIQUINE Ciquine 8/17/94 30 1 1 POLIALDEN Polialden 8/17/94 19 1 1 POLITERO Politeno 8/18/94 73 1 1 EMBRAER Empresa Brasileira de Aeronáutica 12/7/94 455 1 1 (Embraer) Embraer Aircraft Corporation (EAC)c 0 0 Embraer Aviation International (EAI)c 0 0 Indústria Aeronáutica Neiva (Neiva)c 0 0 ESCELSA Espírito Santo Centrais Elétricas S.A. 7/11/95 522 1 1 COPENE Cia. Petroquímica do Nordeste 8/15/95 745 1 1 CPC CPC 9/29/95 161 0 0 CQR CQR 10/5/95 2 1 0 SALGEMA SALGEMA 10/5/95 183 0 0 NITROCARBONO Nitrocarbono 12/5/95 37 1 1 PRONOR Pronor 12/5/95 99 1 1 POLIPROPILENO Polipropileno 2/1/96 86 1 1 KOPPOL Koppol 2/1/96 70 0 0 LIGHT Light Serviços de Eletricidade S.A. 5/21/96 3,094 1 1 DETEN Deten 5/22/96 12 1 0 POLIBRASIL Polibrasilc 8/27/96 111 0 0 EDN Estireno do Nordeste- EDN 9/26/96 16 1 1 CVRD Cia. Vale do Rio Doce 5/6/97 6,858 1 1 CODESP Terminal de Contêiners Tecon 1 9/17/97 251 0 0 (Codesp) CDRJ CDRJ ­ Porto de Angra do Reis 11/5/98 8 0 0 CDRJ CDRJ-Terminal de Conteineres 1 ­ 9/3/98 79 0 0 Porto de Sepetiba CDRJ CDRJ-Terminal Roll-on Roll-off do 11/3/98 26 0 0 177 Porto do Rio (Table continues on the following page.) 178 Table 4A.1 (continued) Listed Date of Auction result Included in before Auction Company name auction (US$ million)a sampleb privatizationb CDES Cia. Docas do Espírito Santo- Casi 5/6/98 26 0 0 de Capuaba CDES Cia. Docas do Espírito Santo ­Casi 5/13/98 9 0 0 de Paul CODEBA Cia. Docas da Bahia 12/21/99 21 0 0 RFF Rede Ferroviária Federal S.A. 7/18/97 15 0 0 (Nordeste) RFF Rede Ferroviária Federal S.A. (Oeste) 3/5/96 63 0 0 RFF Rede Ferroviária Federal S.A. (SP) 11/10/98 206 1 0 RFF Rede Ferroviária Federal S.A. (Sudeste) 9/20/96 870 1 0 RFF Rede Ferroviária Federal S.A. (Sul) 12/13/96 209 0 0 RFF Rede Ferroviária Federal S.A. 11/22/96 18 0 0 (Tereza Crisitina) RFF Rede Ferroviária Federal S.A. 6/14/96 316 0 0 (Centro-Leste) MERIDIONAL Banco Meridional do Brasil S.A. 12/4/97 240 0 0 EMBRATEL Embratel 7/29/98 2,276 1 1 TELESP Telesp Operacional, Borda do Campo 7/29/98 4,967 2 1 CENTRO Telepar, Telebrasília, Telegoiás, and 7/29/98 1,778 3 3 SUL other four closed companies: CTMR, Telemat, Teleron, Teleacre NORTE LESTE Telerj, Telebahia, Telemig, Telpe, 7/29/98 2,949 16 8 Telma, Telest, Teleceará, Teleamazon, and other nine closed companies Aggregate transferred debt of these 2,125 companies Telecom offers to employees 293 TELESP CEL. Telesp Celular 7/29/98 3,082 1 0 SUDESTE CEL. 7/29/98 1,168 1 0 TELEMIG CEL. 7/29/98 649 1 0 CELULAR SUL 7/29/98 601 4 0 NORDESTE CEL. 7/29/98 567 7 0 LESTE CEL. 7/29/98 368 2 0 CENT. OESTE CEL. Telegoiás Celular and other five closed 7/29/98 378 3 1 companies TELE NORTE C 7/29/98 161 2 0 GERASUL Centrais Geradoras do Sul do Brasil 9/15/98 1,962 1 0 S.A. GUARARAPES GUARARAPES 12/7/98 0.1 1 1 DATAMEC Datamec S.A. 6/23/99 49 0 0 BANESPA Banco do Estado de São Paulo 11/20/00 3,604 0 0 Petrobrás Petrobrásd 8/9/00 4,032 1 1 Total 56,841.20 75 38 a. Includes transferred debt; values are current U.S. dollars. b. These two columns show the number of companies included in the sample and the number of representative companies with shares traded at BOVESPA, respectively. c. Sold with mother company. 179 d. Minority shares privatization in remaining state-owned enterprise. Source: BNDES. 180 Table 4A.2 Companies Privatized by BNDES on Behalf of Brazilian States, Minority Shares Privatized by Federal Government, and São Paulo State Privatization Program Date of Auction result Included in Listed before Company name auction (US$ million)a sample privatization Banco Banerj S.A. ­ BANERJ 6/26/97 289 0 0 Banco de Crédito de Minas Gerais S.A ­ Credireal 8/7/97 112 0 0 Banco do Estado da Bahia ­ BANEB 7/22/99 147 0 0 Banco do Estado de Minas Gerais ­ BEMGE 9/14/98 494 0 0 Banco do Estado de Pernambuco S.A. 11/17/98 153 0 0 Banco do Estado de Santa Catarina ­ BESC 9/30/97 28 0 0 Centrais Elétricas Cachoeira Dourada 9/5/97 854 1 1 Centrais Elétricas do Pará S.A. ­ CELPA 7/9/98 504 0 0 CELPE 2000 1,135 1 1 Centrais Elétricas Matogrossenses S.A. ­ CEMAT 11/27/97 814 1 1 CESP Paranapanema 7/28/99 1,164 1 1 CESP TIETÊ 11/1/99 1,140 1 1 Cia de Gás de São Paulo ­ COMGÁS 4/14/99 1,076 1 1 Cia União de Seguros Gerais 11/20/97 45 0 0 Cia. Centro Oeste de Dist. de Energia Elétrica ­ (AES-SUL) 10/21/97 1,436 1 1 Cia. De Eletricidade de Minas Gerais ­ CEMIGb 5/28/97 1,053 1 1 Cia. De Eletricidade do Estado da Bahia ­ COELBA 7/31/97 1,965 1 1 Cia. De Eletricidade do Rio de Janeiro ­ CERJ 11/20/96 951 1 1 Cia. De Navegação do Rio de Janeiro ­ CONERJ 2/5/98 29 0 0 Cia. De Saneamento Básico de São Paulo ­ SABESPb 7/31/97 375 1 1 Cia. De Saneamento Básico do Paraná ­ SANEPARb 6/8/98 217 1 1 Cia. Energética de Brasília ­ CEBb 4/30/97 74 1 1 Cia. Energética do Ceará ­ COELCE 4/2/98 1,338 1 1 Cia. Estadual de Gás do Rio de Janeiro ­ CEG 7/14/97 430 1 0 Cia. Fluminense de Trens Urbanos 7/15/98 240 0 0 Cia. Metropolitano do Rio de Janeirob 12/19/97 262 0 0 Cia. N. NE de Dist. de Energia Elétrica ­ CEEE ­ (RGE) 10/21/97 1,635 1 1 Cia. Paranaense de Energia ­ COPELb 9/20/96 413 1 1 Cia. Paulista de Força e Luz ­ CPFL 11/5/97 2,833 1 1 Cia. Riograndense de Telecomunicações ­ CRT 6/19/98 2,496 1 1 COSERN 12/12/97 718 1 0 EBE ­ BANDEIRANTE DE ENERGIA 9/17/98 1,235 1 0 Elektro Eletricidade e Serviços S.A. ­ ELEKTRO 7/16/98 1,917 1 0 Eletricidade de São Paulo S.A. ­ Metropolitana 4/15/98 3,445 1 0 Empresa Energética de Mato Grosso do Sul ­ ENERSUL 11/19/97 783.0 1 0 Empresa Energética de Sergipe ­ ENERGIPE 12/3/97 560 0 0 Estrada de Ferro Paraná Oeste S.A. ­ Ferroeste 12/10/96 25 0 0 Riogás S.A. 7/14/97 146 0 0 Terminal Garagem Menezes Côrtes 10/28/98 67 0 0 Total 32,598 24 18 Note: BNDES National Social and Economic Development Bank. a. Includes transferred debt; values are current U.S. dollars. b. Minority shares in remaining state-owned enterprises. Source: BNDES. 181 182 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4A.3 Remaining State-Owned Enterprises Productive sectors Eletrobrás group (Electricity) · Centrais Elétricas Brasileiras S.A. ­ ELETROBRÁS Boa Vista Energia S.A. ­ BOVESA Centrais Elétricas de Rondônia S.A. ­ CERON Centrais Elétricas do Norte do Brasil S.A. ­ ELETRONORTE Centro de Pesquisas de Energia Elétrica ­ CEPEL Companhia de Eletricidade do Acre ­ ELETROACRE Companhia de Geração Térmica de Energia Elétrica ­ CGTEE Companhia Energética de Alagoas ­ CEAL Companhia Energética do Amazonas ­ CEAM Companhia Energética do Piauí ­ CEPISA Companhia Hidro Elétrica do São Francisco ­ CHESF Eletrobrás Termonuclear S.A. ­ ELETRONUCLEAR Empresa Transmissora de Energia Elétrica do Sul do Brasil S.A. ­ ELETROSUL FURNAS Centrais Elétricas S.A. LIGHTPAR ­ Light Participações S.A. Manaus Energia S.A. ­ MANAUS ENERGIA Petrobrás group (Oil) · Petróleo Brasileiro S.A. ­ PETROBRÁS Braspetro Oil Services Company ­ BRASOIL Petrobrás Distribuidora S.A. ­ BR Petrobrás Gás S.A. ­ GASPETRO Petrobrás Internacional S.A. ­ BRASPETRO Petrobras International Finance Company ­ PIFCO Petrobrás Química S.A. ­ PETROQUISA Transportadora Brasileira Gasoduto Bolívia-Brasil S.A. ­ TBG Indústria Carboquímica Catarinense S.A. ­ ICC (Em Liquidação) Petrobrás Transporte S.A. ­ TRANSPETRO Fronape International Company ­ FIC Ports Companhia Docas do Ceará ­ CDC Companhia Docas do Espírito Santo ­ CODESA Companhia das Docas do Estado da Bahia ­ CODEBA Companhia Docas do Estado de São Paulo ­ CODESP Companhia Docas do Maranhão ­ CODOMAR Companhia Docas do Pará ­ CDP Companhia Docas do Rio de Janeiro ­ CDRJ Companhia Docas do Rio Grande do Norte ­ CODERN Transportation Rede Ferroviária Federal S.A. ­ RFFSA (In process of liquidation) Rede Federal de Armazéns Gerais Ferroviários S.A. ­ AGEF (In process of liquidation) COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 183 Table 4A.3 (continued) Other BB-Administradora de Cartões de Crédito S.A. ­ BB-CAR BB-Corretora de Seguros e Administradora de Bens S.A. ­ BB-COR BB-TUR Viagens e Turismo Ltda. BEM Serviços Gerais Ltda. ­ BEM SG BEM Vigilância e Transporte de Valores S.A. ­ BEM VTV Casa da Moeda do Brasil ­ CMB Centrais de Abastecimento de Minas Gerais S.A. ­ CEASA/MG Companhia de Armazéns e Silos do Estado de Minas Gerais ­ CASEMG Companhia de Entrepostos e Armazéns Gerais de São Paulo ­ CEAGESP COBRA ­ Computadores e Sistemas Brasileiros S.A. Empresa Brasileira de Correios e Telégrafos ­ ECT Empresa Brasileira de Infra-Estrutura Aeroportuária ­ INFRAERO Empresa de Processamento de Dados da Prev. Social ­ DATAPREV Empresa Gerencial de Projetos Navais ­ EMGEPRON Hospital Cristo Redentor S.A. ­ REDENTOR Hospital Fêmina S.A. ­ FÊMINA Hospital Nossa Senhora da Conceição S.A. ­ CONCEIÇÃO Indústria de Material Bélico do Brasil ­ IMBEL Sistema de Processamento de Dados, Planej. e Adm. de Cartões de Crédito Ltda. ­ SISPLAN Telecomunicações Brasileiras S.A. ­ TELEBRÁS Financial sector · Banco do Brasil S.A. ­ BB BB-Banco de Investimento S.A. ­ BB-BI BB-Distribuidora de Títulos e Valores Mobiliários S.A. ­ BB-DTVM BB-Financeira S.A., Crédito, Financ. e Investimento ­ BB-FIN BB-Leasing Company Ltd. ­ BB-LEASING BB-Leasing S.A. Arrendamento Mercantil ­ BB-LAM Brasilian American Merchant Bank ­ BAMB · Banco Nacional de Desenvolvimento Econômico e Social ­ BNDES Agência Especial de Financiamento Industrial ­ FINAME BNDES Participações S.A. ­ BNDESPAR · Banco do Estado de Goiás S.A. ­ BEG BEG Distribuidora de Títulos e Valores Mobiliários S.A. ­ BEG DTVM · Banco do Estado de Santa Catarina S.A. ­ BESC BESC Distribuidora de Títulos e Valores Mobiliários S.A. ­ BESCVAL BESC Financeira S.A. Crédito, Financiamento e Investimento ­ BESCREDI BESC S.A. Arrendamento Mercantil ­ BESC LEASING · Banco do Estado do Ceará S.A. ­ BEC BEC Distribuidora de Títulos e Valores Mobiliários S.A. ­ BEC DTVM (Table continues on the following page.) 184 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4A.3 (continued) · Banco do Estado do Maranhão S.A. ­ BEM BEM Distribuidora de Títulos e Valores Mobiliários Ltda. ­ BEM DTVM Banco da Amazônia S.A. ­ BASA Banco do Estado do Amazonas S.A. ­ BEA Banco do Estado do Piauí S.A. ­ BEP Banco do Nordeste do Brasil S.A. ­ BNB Caixa Econômica Federal ­ CEF IRB-Brasil Resseguros S.A. ­ IRB-BRASIL RE Financiadora de Estudos e Projetos ­ FINEP Enterprises included in the fiscal budget Companhia Brasileira de Trens Urbanos ­ CBTU Companhia de Desenvolvimento de Barcarena ­ CODEBAR Companhia de Desenvolvimento dos Vales do São Francisco e do Parnaíba ­ CODEVASF Companhia de Navegação do São Francisco ­ FRANAVE Companhia de Pesquisa de Recursos Minerais ­ CPRM Companhia Nacional de Abastecimento ­ CONAB Empresa Brasileira de Comunicação S.A. ­ RADIOBRÁS Empresa Brasileira de Pesquisa Agropecuária ­ EMBRAPA Empresa Brasileira de Planejamento de Transportes ­ GEIPOT Empresa de Trens Urbanos de Porto Alegre S.A. ­ TRENSURB Hospital de Clínicas de Porto Alegre ­ HCPA Indústrias Nucleares do Brasil S.A. ­ INB Nuclebrás Equipamentos Pesados S.A. ­ NUCLEP Serviço Federal de Processamento de Dados ­ SERPRO VALEC - Engenharia, Construções e Ferrovias S.A. Other Centrais de Abastecimento do Amazonas S.A. ­ CEASA/AM Petrobrás America Inc. ­ AMERICA Petrobrás U.K. Limited ­ BUK Source: Ministry of Planning, Budget and Administration, Department of Coordination and Control of State Enterprises, Executive Secretary. Appendix 4B Table 4B.1 Description of the Variables Variable Definition Profitability Operating Ratio of operating income to sales. Operating income/sales income is equal to sales minus operating (OI/S) expenses, cost of sales, and depreciation. Sales are equal to total value of products and services sold minus sales returns and discounts. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 185 Table 4B.1 (continued) Variable Definition Operating income/ Ratio of operating income to property, plant, and property, plant equipment, which is the value of a company's and equipment fixed assets adjusted for inflation. Operating (OI/PPE) income is equal to sales minus operating expenses, cost of sales, and depreciation. Net income/sales Ratio of net income to sales. Net income is equal (NI/S) to operating income minus interest expenses and net taxes paid. Sales are equal to total value of products and services sold minus sales returns and discounts. Return on assets Ratio of operating income, which is equal to (ROA) sales operating expenses, cost of sales, and depreciation minus interest expenses and net taxes paid, to total assets. Return on equity Ratio of operating income, which is equal to (ROE) sales minus operating expenses, cost of sales, and depreciation minus interest expenses and net taxes paid, to equity. Operating efficiency Log(sales/PPE) Log of the ratio of total value of products and services sold minus sales returns and discounts to property, plant, and equipment, which is the value of a company's fixed assets adjusted for inflation. Operating Ratio of operating expenses to sales, defined as costs/sales (OC/S) the total value of products and services sold minus sales returns and discounts. Assets Log(PPE) Log of property, plant, and equipment, which is the value of a company's fixed assets adjusted for inflation. Investment/sales Ratio of investment to sales, defined as the total (I/S) value of products and services sold minus sales returns and discounts. Investment is calculated as the difference of property, plant, and equipment along time. Investment/PPE Ratio of investment to property, plant, and (I/PPE) equipment, which is the value of a company's fixed assets adjusted for inflation. Output Log(sales) Log of sales, the total value of products and services sold minus sales returns and discounts. (Table continues on the following page.) 186 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4B.1 (continued) Variable Definition Shareholders Payout ratio Ratio of total dividends to operating income minus interest expenses and net taxes paid. Finance Current Ratio of current assets to current liabilities. Long-term Ratio of long-term debt to equity debt/equity (LTD/E) Net taxes Net taxes/sales Ratio of net taxes to sales. Net taxes are equal to (NT/S) corporate income taxes paid net of direct subsidies or tax credits received during the fiscal year. Source: Authors' calculations. Appendix 4C Table 4C.1 Change in Performance: Tests of Means and Medians (Two Years before Privatization versus Two Years after, without Adjustment) Mean and Mean and median median Variable N before after Z-test Profitability Operating income/sales 66 0.037 0.042 0.536 0.072 0.108 0.523 Operating income/PPE 67 0.092 0.141 3.556*** 0.035 0.107 3.566*** Net income/sales 66 0.000 0.008 0.595 0.034 0.039 0.677 Return on assets 70 0.860 0.008 0.291 0.014 0.011 1.287 Return on equity 70 1.152 0.046 0.662 0.019 0.039 0.862 COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 187 Table 4C.1 (continued) Mean and Mean and median median Variable N before after Z-test Operating efficiency Log(sales/PPE) 63 0.273 0.006 5.520*** 0.201 0.009 5.492*** Operating costs/sales 58 0.375 0.251 2.631*** 0.200 0.196 2.917*** Assets Log(PPE) 67 6.001 5.946 1.981* 5.891 5.813 1.983* Investment/sales 57 0.295 0.032 2.550** 0.158 0.093 2.476** Investment/PPE 57 0.115 0.094 1.202 0.101 0.104 0.202 Output Log(sales) 66 5.644 5.876 4.335** 5.403 5.643 4.301** Shareholders Payout ratio 45 71.40 55.99 0.089 30.78 48.66 0.166 Finance Current 70 0.847 1.009 2.755*** 0.745 0.866 3.089*** Long-term debt/equity 63 0.636 0.701 2.506** 0.181 0.269 2.506** Net taxes Net taxes/sales 65 0.024 0.010 3.834*** 0.017 0.007 3.343*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number; PPE property, plant, and equipment. The table presents, for each empirical proxy, the number of usable observations and the mean and median values before and after privatization. The full sample contains 102 privatized companies. Before privatization refers to the average of the two years before the sale; after privatization refers to the average of the two years following privatization. We report Z-statistics for signed rank tests and Wilcoxon rank sum tests for change in mean and median values, respectively. Definitions for each variable can be found in appendix table 4B.1. Source: Authors' calculations. 188 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4C.2 Change in Performance: Tests of Means and Medians (Two Years before Privatization versus Two Years after, with Adjustment) Mean and Mean and median median Z-test Variable N before after (1) Profitability Operating income/sales 66 0.097 0.430 2.944*** 0.084 0.019 2.944*** Operating income/PPE 67 0.092 0.141 3.556*** 0.005 0.222 5.713*** Net income/sales 66 0.004 0.105 1.476 0.020 0.012 1.534 ROA 70 0.870 0.014 0.824 0.003 0.012 1.369 ROE 70 1.194 0.025 1.768* 0.030 0.021 1.698* Operating efficiency Log(sales/PPE) 63 0.548 0.298 3.980** 0.522 0.218 3.876** Operating costs/sales 58 0.174 0.065 1.837 0.014 0.021 0.809 Assets Log(PPE) 67 1.445 1.002 1.286 0.955 0.871 1.370 Investment/sales 57 0.223 0.058 1.887** 0.117 0.066 1.795* Investment/PPE 57 0.038 0.024 0.774 0.026 0.039 0.264 Output Log(sales) 66 0.774 0.901 3.306*** 0.274 0.457 3.598*** Shareholders Payout ratio 45 0.309 ­0.263 0.229 28.62 ­5.805 0.299 Finance Current 70 0.510 0.250 3.238*** 0.605 0.250 3.768** LTD/equity 63 0.254 0.108 0.210 0.142 0.325 0.021 (Table continues on the following page.) COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 189 Table 4C.2 (continued) Mean and Mean and median median Z-test Variable N before after (1) Net taxes Net taxes/sales 65 0.018 0.014 3.578*** 0.005 0.003 3.575*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number; PPE property, plant, and equipment; ROA return on assets; ROE return on equity; and LTD long-term debt. The table presents, for each empirical proxy, the number of usable observations and the mean and median values before and after privatization, adjusted to take into account unstable macroeconomic policies. See text for full explanation. The full sample contains 102 privatized companies. Before privatization refers to the average of the two years before the sale; after privatization refers to the average of the two years following privatization. We report Z-statistics for signed rank tests and Wilcoxon rank sum tests for change in mean and median values, respectively. Definitions for each variable can be found in appendix table 4B.1. Source: Authors' calculations. Table 4C.3 Change in Performance: Tests of Means and Medians (All Years before and after Privatization, without Adjustment) Mean and Mean and median median Z-test Variable N before after (1) Profitability Operating income/sales 71 0.052 0.050 1.511 0.080 0.096 1.037 Operating income/PPE 70 0.057 0.291 3.042*** 0.045 0.097 3.408*** Net income/sales 71 0.067 0.042 0.815 0.010 0.039 0.889 ROA 73 0.812 0.017 2.967*** 0.003 0.026 2.311** ROE 73 1.109 0.021 2.258** 0.008 0.038 2.150** Operating efficiency Log(sales/PPE) 64 0.247 0.076 5.600*** 0.285 0.012 5.244*** (Table continues on the following page.) 190 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4C.3 (continued) Mean and Mean and median median Z-test Variable N before after (1) Operating costs/sales 64 0.428 0.245 3.138*** 0.255 0.207 2.756*** Assets Log(PPE) 70 6.889 5.994 1.141 5.911 5.809 0.952 Investment/sales 62 0.191 0.038 1.406 0.202 0.113 1.157 Investment/PPE 62 0.017 0.118 1.288 0.085 0.098 0.168 Output Log(sales) 71 5.823 6.004 2.032* 5.800 5.974 1.956* Shareholders Payout ratio 59 34.406 30.860 0.138 38.848 42.268 1.232 Finance Current 73 0.849 1.106 2.662*** 0.843 0.905 2.642*** LTD/equity 66 0.529 0.576 3.192*** 0.167 0.298 3.302*** Net taxes Net taxes/sales 68 0.015 0.009 3.821*** 0.018 0.006 4.296*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number; PPE property, plant, and equipment; ROA return on assets; ROE return on equity; and LTD long-term debt. The table presents, for each empirical proxy, the number of usable observations and the mean and median values before and after privatization. The full sample contains 102 privatized companies. Before privatization refers to the average of all the years before the sale; after privatization refers to the average of all the years following privatization. We report Z-statistics for signed rank tests and Wilcoxon rank sum tests for change in mean and median values, respectively. Definitions for each variable can be found in appendix table 4B.1. Source: Authors' calculations. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 191 Table 4C.4 Change in Performance: Tests of Means and Medians (All Years before and after Privatization, with Adjustment) Mean and Mean and median median Z-test Variable N before after (1) Profitability Operating income/sales 71 0.050 0.005 0.107 0.072 0.036 0.241 Operating income/PPE 70 0.003 0.385 6.112*** 0.010 0.207 6.387*** Net income/sales 71 0.084 0.064 0.693 0.005 0.014 0.262 ROA 73 0.831 0.003 3.130*** 0.017 0.003 2.736*** ROE 73 1.159 0.012 3.236*** 0.440 0.014 3.223*** Operating efficiency Log(sales/PPE) 63 0.677 0.255 5.226*** 0.600 0.264 4.914*** Operating costs/sales 64 0.236 0.066 3.199*** 0.090 0.025 2.819*** Assets Log(PPE) 69 1.380 1.017 2.001* 1.111 0.993 1.885* Investment/sales 62 0.098 0.011 0.394 0.123 0.086 0.730 Investment/PPE 62 0.018 0.055 1.385 0.022 0.029 1.072 Output Log(sales) 71 0.906 1.178 2.333*** 0.855 1.029 2.599*** Shareholders Payout ratio 59 0.082 5.963 0.731 29.35 9.292 1.169 Finance Current 73 0.526 0.232 3.653*** 0.503 0.313 3.937*** LTD/equity 66 0.233 0.002 2.086** 0.107 0.238 2.286** (Table continues on the following page.) 192 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO Table 4C.4 (continued) Mean and Mean and median median Z-test Variable N before after (1) Net taxes Net taxes/sales 68 0.007 0.005 3.173*** 0.007 0.002 3.534*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number; PPE property, plant, and equipment; ROA return on assets; ROE return on equity; and LTD long-term debt. The table presents, for each empirical proxy, the number of usable observations and the mean and median values before and after privatization, adjusted to take into account unstable macroeconomic policies. See text for full explanation. The full sample contains 102 privatized companies. Before privatization refers to the average of all the years before the sale; after privatization refers to the average of all the years following privatization. We report Z-statistics for signed rank tests and Wilcoxon rank sum tests for change in mean and median values, respectively. Definitions for each variable can be found in appendix table 4B.1. Source: Authors' calculations. Appendix 4D Table 4D.1 Definition of the Control Variables Included in the Vector of the Econometric Model Expected Variable Definition impact on iit Split/merger A dummy variable that takes the ( ) or ( ) value 1 if the company had been split or merged as a result of its privatization, and 0 otherwise. Minority control A dummy variable that takes the ( ) value 1 if the government, owned only a minority participation before privatization, and 0 otherwise. Listed A dummy variable that takes the ( ) value 1 if the privatized enterprise was listed on the São Paulo stock exchange before privatization, and 0 otherwise. (Table continues on the following page.) COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 193 Table 4D.1 (continued) Expected Variable Definition impact on iit Tradable A dummy variable that assumes the ( ) value 1 if the privatized firm is in a tradable goods industry, following three criteria: its typical product is included in the international classification of tradable goods; it is free from nontariff restrictions; and its effective protection tariff is not redundant, and 0 otherwise. Regulateda A dummy variable that assumes a ( ) or ( ) value of 1 if the privatized firm is in a regulated industry, and 0 otherwise. Private mean The yearly average of each ( ) or ( ) performance indicator based on private companies only. a. Regulated refers to privatized firms that belong to an industry whose prices have been controlled by the government before and after privatization. Source: Authors' calculations. Notes The authors acknowledge financial support from Fundação Instituto de Pesquisas Econômicas (FIPE) and the Latin American and Caribbean Research Network Pro- gram (LACRNP) of the Inter-American Development Bank; the assistance of Economática; Austin Assis, and the Getúlio Vargas Foundation for providing the data sets; and the research assistance of Renata Domingos and Alan de Genaro Dario. 1. These numbers, as well as others quoted from the same study, have been rounded. 2. This section draws from Macedo (2000) and updates and extends his analysis. 3. These values exclude concessions of public services. All dollar amounts are U.S. dollars. 4. The major source of data on the Brazilian privatization program is BNDES, which was given the task of managing it, including a part developed at the state level. The reports and other documents used as sources are BNDES 1999a, 1999b, and 2001. 5. The same kind of data is used later in the analysis of employment effects. This data base is known as RAIS (Relação Anual das Informações Sociais, or An- nual Survey of Social Data). 194 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO 6. The differential, net of the workers' characteristics, reached a peak of 80 percent when the workers' characteristics were valued according to the private sec- tor criteria, as measured by the regression coefficients of the workers' characteris- tics in the wage equation of that sector. 7. More recently, in a seminar sponsored by BNDES to celebrate the 10th an- niversary of the privatization program, Pinheiro (2000) presented some additional and updated results, again based on data that cannot be disclosed, this time cover- ing 55 firms. The analysis simply compared the performance of the firms before and after privatization, thus not relating their performance to that of the private firms. Pinheiro found sizable increases in net operational revenues, investment, net profit, productivity, and tax collections as well as a reduction in employment, in some cases compensated by an expansion in contracted-out services. 8. High rates of inflation plagued the economy from 1986 to 1994, a period in which indexation following legal rules was widespread. Because the analysis is developed in terms of ratios based on flow variables, such as operating income to sales, the problems of inflation and indexation are circumvented. For a few cases in which the absolute value of the indicator is used, the original values in Brazilian currency were converted into U.S. dollars. 9. In the process of privatization, BNDES franchised to interested bidders the existing information on the firms. The files are kept by BNDES, but they are con- sidered a proprietary right of the winning bidder. BNDES has also occasionally used questionnaires to gather information from firms after privatization. BNDES responded to our request for both types of data saying that it could not make them available to third parties. 10. This procedure differs from that of La Porta and López-de-Silanes (1999) in that they used one fixed year for the period after privatization. In the Mexican case, privatization was heavily concentrated in a few years. In Brazil it has extended over a decade and more. Therefore, a fixed year for comparison would be inade- quate. 11. Our adjustment, however, could not be done by industry, as some priva- tized enterprises do not have a corresponding match in the private sector. This is the case, for instance, with the major mining company (Companhia Vale do Rio Doce), the telecommunications companies, and many companies in the energy sector. 12. This information was available only for listed companies. 13. According to Baltagi (1995), inclusion of a lagged dependent variable ren- ders the OLS (ordinary least squares) estimator biased and inconsistent even if the error terms are not serially correlated. Baltagi (1995) and Hsiao (1986) also demonstrate that the same problem affects GLS (generalized least squares) and FGLS (feasible generalized least squares) estimators. Finally, the instrumental vari- able (IV) estimation method alone leads to consistent, but not necessarily efficient, estimates of the parameters in the model, because it does not make use of all the available moment conditions (Ahn and Schmidt 1995). 14. For the lagged dependent variables, this method resorts to instrumental variables that are obtained in a dynamic panel data model once existing orthogo- nality conditions between these lagged values and the disturbances are taken into account. A set of valid instruments is represented by all the dependent variables lagged more than one period. In this chapter, the parameter estimates were ob- tained by using as an instrument the independent variable lagged two years. 15. The incentives gained strength after new "social rights" were established by the constitution of 1988, as detailed by Fernandes (1998). 16. Pinheiro (2000) tackled both the direct impact and the effect of contracting out on employment, on the basis of questionnaires sent to the privatized firms by BNDES. He found a 33 percent reduction in the total number of formal workers. COSTS AND BENEFITS OF PRIVATIZATION: EVIDENCE FROM BRAZIL 195 For production workers, the reduction was 29.5 percent, evidence that overstaffing was concentrated in white-collar workers. In absolute numbers, he found that, ex- cluding telecommunications, the total reduction was 10,000 workers in the year of privatization and 35,000 in the year before, thus showing adjustment by SOEs be- fore privatization. In the telecommunications sector, he found that 145,000 new jobs were contracted out to expand services. This number might sound high, but notice that in this country of 170 million inhabitants, the number of fixed tele- phone lines increased from 9.6 per 100 people in 1996 to 21.4 in 2000, while the number of cellular phones rose from 1.6 per 100 people to 12.9, an expansion that has required a lot of labor, particularly in the case of fixed lines. 17. Data obtained from the Brazilian Central Bank and Department of Treasury. 18. Macedo (2000) compares the fiscal picture at that time to what Kornai (1979) calls a soft budget constraint, typical of centralized governments whose budgets are only vaguely monitored or controlled by Congress and society, if at all. Under such conditions, the only effective constraint emerges when markets react to the piling up of debt and the interest rate becomes a problem in itself. 19. Interestingly enough, the devaluation in early 1999 came after the telecom- munications privatization auction in 1998, in which the presence of foreign direct investment was stronger. These investments were seen by the market as a sign that the government could hold on to the overvalued real. As the program came to a halt, devaluation came sooner than expected. 20. The other countries are Argentina, Bolivia, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Paraguay, Peru, and Uruguay. 21. The highest ratings were given to the aviation industry (80 percent), in which Embraer, the only former SOE, has been very successful; steel (65 percent); and telecommunications (58 percent). The lowest ratings were received by rail- roads (9 percent), electricity (13 percent), and one airline (11 percent)--a small company that belonged to the state of São Paulo and was individually privatized in the mid-1980s. 22. In 2000, for the first time since the program started, the federal government resorted to a public offer of minority shares in Petrobrás, totaling $4 billion, in which workers were also allowed to participate with their FGTS deposits. The op- eration was very successful, as was another public offer of a remaining state-owned block of minority shares of Companhia Vale do Rio Doce, sold in 2002 for $1.9 billion. 23. The source of the tax burden data is the Secretary of Federal Revenue, Min- istry of Finance, as published by Folha de São Paulo (May 10, 2003). References Ahn, S. C., and P. Schmidt. 1995. "Efficient Estimation of Models for Dynamic Panel Data." Journal of Econometrics 68: 5­27. Arellano, M., and S. Bond. 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations." Review of Economic Studies 58: 277­97. Baltagi, B. H. 1995. Econometric Analysis of Panel Data. New York: John Wiley & Sons. BNDES (Banco Nacional de Desenvolvimento Econômico e Social). 1999a. Priva- tizações no Brasil ­ 1991/99. Rio de Janeiro. 196 ANUATTI-NETO, BAROSSI-FILHO, CARVALHO, AND MACEDO ------. 1999b. Sistema de Informações. Rio de Janeiro. ------. 2001. "Privatização, www.gov.br. Fernandes, R. 1998. "Encargos Sociais e Demanda por Trabalho no Setor Formal da Economia." Economia Aplicada 2(3): 553­78. Hsiao, C. 1986. Analysis of Panel Data. Cambridge, U.K.: Cambridge University Press. Kornai, J. 1979. "Resource vs. Demand-Oriented Systems." Econometrica 47 (4): 801­19. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." The Quarterly Journal of Economics (Novem- ber): 1193­242. La Porta, Rafael, Florencio López-de-Silanes, Andrei Shleifer, and Robert Vishny. 1997. "Legal Determinants of External Finance." Journal of Finance 52 (3): 1131­50. Lamounier, B., and A. De Souza. 2002. "As Elitres Brasileiras e o Desenvolvimento Nacional: Fatores de Consenso e Dissenso." Relatório de Pesquisa. São Paulo: IDESP (Instituto de Estudos Econômicos, Sociais e Políticos). Levine, Ross. 1997. "Financial Development and Economic Growth: Views and Agenda." Journal of Economic Literature (35): 688­726. Levine, Ross, and S. Zervos. 1998. "Stock Markets, Banks, and Economic Growth." American Economic Review (88): 537­58. Lora, Eduardo, and U. Panizza. 2002. "Structural Reforms in Latin America under Scrutiny." Paper prepared for the seminar "Reforming Reforms," held at the annual meeting of the Board of Governors, Inter-American Development Bank, Fortaleza, Brazil. Available at www.iadb.org/res. Macedo, R. 1985. Os Salários nas Empresas Estatais. São Paulo: Nobel. ------. 2000. "Privatization and the Distribution of Assets and Income in Brazil." Working Paper 14. Carnegie Endowment for International Peace, Washington, D.C. Megginson, William L., and Jeffry M. Netter. 2001. "From State to Market: A Survey of Empirical Studies of Privatization." Journal of Economic Literature 39 (June): 321­89. Megginson, William L., Robert C. Nash, and Matthais van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An In- ternational Empirical Analysis." Journal of Finance 49 (2): 403­52. Pinheiro, Armando C. 1996. "Impactos Microeconômicos da Privatização no Brasil." Pesquisa e Planejamento Econômico 26 (3): 357­98. ------. 2000. "Após a Privatização." Rio de Janeiro: BNDES. Presentation in Powerpoint format available at www.bndes.gov.br. Pinheiro, Armando C., and F. Giambiagi. 1997. "Lucratividade, Dividendos e In- vestimentos das Empresas Estatais: Uma Contribuição para o Debate sobre Pri- vatização no Brasil." Revista Brasileira de Economia 51 (January­March): 93­131. 5 The Effects of Privatization on Firms: The Chilean Case Ronald Fischer, Rodrigo Gutiérrez, and Pablo Serra THIS CHAPTER EVALUATES THE EFFECTS OF privatization on the efficiency of firms and institutions in Chile. One of the chief characteristics of the Chilean privatization process is that it has been all-encompassing. In the three decades that followed the fall of the government of the socialist president Salvador Allende (December 1970­December 1973), all the banks and firms that had been acquired or expropriated by the Allende administration were either privatized or liquidated. Farms that had been expropriated after the agrarian reform of 1965 were privatized, as well as a majority of the firms that were state-owned before December 1970. The military government of Augusto Pinochet also privatized the pen- sion system and a part of the health insurance system. It promoted vouch- ers for subsidized private schools and allowed free entry of new institutions into university and other tertiary education. Finally, the private sector im- proved or built and undertook the operation of most large infrastructure projects such as highways, seaports, airports, water reservoirs, and even jails. In addition, in a bid to decentralize government, local governments (municipalities) became responsible for the lowest level of the public health care system as well as for public schooling. The privatization effort has been part of a much wider process of eco- nomic liberalization that Chile initiated in 1974. The process represents a major reversal of the policies the country had followed since the 1940s, when the state played a role not only through the public firms, but also through regulations and other mechanisms (Galetovic 1998). The govern- ment set interest rates and exchange rates, as well as regulating almost 3,000 prices for goods and services. As part of its import substitution 197 198 FISCHER, GUTIÉRREZ, AND SERRA strategy, the state protected those sectors deemed essential. All of these mechanisms started to disappear in 1974 with the country's shift toward a market economy in which the price system was the main determinant of resource allocation and the private sector was the centerpiece of the economy. Moreover, many traditional supervisory activities were sur- rendered to the private sector. It is possible to distinguish three main phases of the privatization process, even though any chronological division is arbitrary. In the first phase, which covers the period 1974­83, 259 firms that were expropri- ated or illegally taken during the Allende administration were restored to their original owners. The government also sold or liquidated an addi- tional 112 nonfinancial firms acquired in the same period (retaining 1 in that category). In addition, 33 of the 65 firms owned by the government before 1970 were also privatized or liquidated. Nevertheless, in 1983 the government still owned 45 firms, some of them because they were consid- ered of strategic importance and some because there were no takers. These included all of the major telecommunications and electric power firms, as well as copper mining companies. In the second phase, from 1984 to 1989, the state privatized the telecommunications and electric power firms as well as most of the firms previously considered strategic: the CAP steel works, the national airline LAN Chile, and other major firms. It also finished selling the last few firms that had been acquired by the socialist government. By 1989 only 19 of the 65 firms dating from the pre-Allende period were still state- owned. The privatization of state-owned enterprises (SOEs) slowed down in the 1990­2001 period. However, the government sold the three main water and sewage companies and completed the privatization of the elec- tricity sector. The distinguishing feature of this period, however, is the privatization--through concession contracts--of infrastructure manage- ment. Since 1993, the main highways and airports have been built, main- tained, and operated by private investors. The main state-owned ports have also been franchised to private firms. Most analysts ascribe the strong growth of the Chilean economy that began in 1985 (after a severe crisis in the first half of the decade) to eco- nomic liberalization. If we accept this premise, we may still question the specific contribution of privatization. So many systemic changes occurred at the same time that it is difficult to evaluate the individual contribution of a particular policy. Nevertheless, Larraín and Vergara (1995) suggest that the rest of the program would not have been credible in the absence of a privatization process. Moreover, privatization was important in help- ing balance the budget and in developing capital markets.1 In this chapter, however, we focus on the direct effect of privatization on the efficiency of privatized sectors and therefore do not look at the global effects on the rest of the economy. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 199 Privatization of SOEs The Chilean privatization of SOEs has been a long-lasting and yet unfinished process. There are still 38 firms--most of them of economic importance-- that remain in public hands. They include ENAP, the monopoly oil refin- ery; Codelco, a copper mining concern that is the largest company in Chile; ENAMI, a copper refinery; Banco del Estado, the fourth largest commercial bank; the post office; the subway; the Chilean mint; the rail lines; 10 ports; Zofri, a free trade zone; and other minor companies, rep- resenting, in all, around 9 percent of total gross domestic product (GDP) in 1998 (Hachette 2000). The perception that Chile has advanced further along the privatization route than most other countries is probably due to the fact that the bulk of traditional infrastructure and social services has been privatized rather than to the extent to which the state has retired from the productive sector. As can be seen from table 5.1, 125 firms were privatized between 1974 and 2001. Most of these firms, however, were owned by the state for only a short time, and only 65 of these stayed in public hands long enough to count as true SOEs. Most of the firms acquired during the socialist administration (1970­73) were privatized by 1978, and by 1983 only 1 of those firms was still state-owned. At least 55 other firms controlled by the government have been liquidated. Several of these were viable only while protected by large tariffs and other nontariff barriers, and thus became nonviable after the opening of the economy (Hachette and Lüders 1994). Note also that in the 1979­89 period many state- owned firms were created by the breaking up of larger firms and were later sold. Table 5.1 Nationalization and Privatization of Firms in Chile, 1970­2001 Number of firms 1970­73 1974­78 1979­83 1984­89 1990­2001 Beginning of period 65 179 82 45 44 Acquired 113 1 0 0 0 Created 1 0 10 29 12 Privatized 0 70 14 27 14 Liquidated 0 28 20 3 4 No information 0 0 13 0 0 End of period 179 82 45 44 38 Note: The table does not include Pehuenche, which was privatized as a project. It does include Corporación del Cobre (Codelco) and 10 seaports originating in the breakup of Emporchi. Source: Authors. 200 FISCHER, GUTIÉRREZ, AND SERRA Compared with later processes (in Mexico, for instance, see La Porta and López-de-Silanes 1999), privatization in Chile was not transparent in its early stages. This can be explained partially by the violent social con- vulsions that affected Chile from 1970 through 1982. This era saw a coup overthrowing the socialist government and bringing in a dictatorship, three large economic crises (1973, 1975, and 1982), and major structural changes in the economy. Furthermore, policymakers explored untried policies in the absence of a free press--there is almost no record of priva- tizations that took place in the 1970s. Moreover, accounting books con- vey little about the value of a firm as inflation rates reached levels of more than 500 percent in some years, and bookkeeping regulations were loose and not upgraded until the 1982 crisis. Therefore, most of the usable data on privatization for Chile correspond to the firms privatized since the early 1980s that remained public corporations (that is, that trade shares on the stock market), because they were required to publish financial informa- tion. More than half of these firms provide public services, and their data are contaminated by the effects of regulation, while the remaining firms are usually dominant in their markets or have sizable market shares. The most important conclusion we draw in this section is that contrary to other documented cases (see La Porta and López-de-Silanes 1999), most SOEs in Chile were fairly efficient before being sold, except perhaps in the sense of overinvestment in the electric power sector. As a matter of fact, employment increased after privatization in most firms. As a result, priva- tization did not substantially improve the behavior of privatized firms, and by many measures, investment was lower than average for their sec- tors after privatization. 2 A second conclusion is that in the case of several variables of interest, the main divide is that between firms operating in a regulated market and those acting in a competitive market. In particular, privatized firms that face competition have had lower profit rates, with profitabilities that are similar to those of their respective industries, while firms in the regulated sector have had significantly higher profit rates than the average for their industries at the 2-digit SIC (Standard Industrial Clas- sification) level. In the case of the regulated sectors, the effects of privati- zation may result from differences in management efficiency or from the introduction of new regulations on the sector, or from the interplay of those two factors. Thus, it is necessary to evaluate the regulations in order to understand the impact of privatization. Privatization of Regulated Sectors This section focuses on the privatization of utilities that occurred in the 1980s and on the private infrastructure franchises of the 1990s. In these sec- tors the government switched from the role of provider to the role of regu- lator. The government was aware that in some cases adequate incentives THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 201 were required in order for privatization to increase welfare. For that rea- son the regulations that were introduced before privatization tended to promote competition whenever it was feasible and to stimulate efficient behavior when competition was impossible. Privatization of Utilities In this section we analyze the postprivatization performance of regulated utilities and relate it to regulatory legislation. We focus on the electricity and telecommunications companies that were privatized in the 1980s.3 The gains in efficiency from privatization derive both from the differential efficiency between the public and private management and from the effect of the rules and regulations that were imposed on the sector. During the 1980s Chile reformed and liberalized its electricity and telecommunications sectors. The process started in the late 1970s, with the establishment of new regulatory bodies and the introduction of new legis- lation in 1982, and culminated in the privatization of the major firms be- tween 1985 and 1989. One trait that infrastructure-based sectors share is that competitive segments coexist with other segments that constitute a natural monopoly. Chile's policy has been to introduce competition wher- ever possible and to regulate noncompetitive segments of industry. The government believed that market discipline played an essential role in eco- nomic policy, so much so that one of the first economic laws it introduced (in October 1973) was a thorough revamping of antitrust legislation. With respect to the Chilean electricity sector, the cornerstone of the re- form process was the introduction of competition in the wholesale con- tract market for energy. The unbundling of transmission services was a prerequisite for wholesale competition to survive. Thus, it was necessary to introduce the principle of open access to the transmission network. The second major change was that investment in power generation was left to market forces. It was assumed that existing firms or potential en- trants would invest in generation capacity whenever a project had a re- turn on capital that was commensurate with the sector's risk. The third major regulatory innovation was the introduction of incentive regulation to calculate the distribution rates. This implies that prices are set so that an efficient distribution company attains a predetermined rate of return (Fischer and Serra, 2000). The legislation regulating the telecommunica- tions sector follows a similar pattern. These laws allow for free market prices in all sectors deemed competitive but regulate rates of basic phone services considered to be local monopolies. As the local network is con- sidered an essential facility for competitors, the 1982 law requires local telephone service operators to provide access to their network to any other operator that requests the service. On average, SOEs increased their profitability and efficiency after pri- vatization, following the trend of the national economy, but the behavior 202 FISCHER, GUTIÉRREZ, AND SERRA of firms providing regulated services stood out. Their labor productivity-- and consequently their profitability--increased more than that of the un- regulated firms. Hence, there is some evidence that the incentive mecha- nisms worked and provided incentives for efficiency. At the same time, the high profit rates of these firms are also evidence of regulatory failure. In fact, the available evidence shows that a large fraction of the efficiency gains was not transferred to consumers as prescribed by the regulatory model. Nonetheless, this situation has changed since the turn of the cen- tury as regulators have become more forceful and competition has made its mark even on sectors previously considered to be natural monopolies. The Privatization of Infrastructure Despite the existence of some early plans to franchise infrastructure dur- ing the Pinochet government, it was the democratically elected Aylwin administration (1990­94) that managed to pass a law allowing private franchises of highways and other infrastructure projects. Because of ini- tial delays and practical hurdles, it was not until the Frei administration (1994­2000) that the groundwork was completed and the franchising of infrastructure went into full swing. For the next six years, during and af- ter the Frei administration, most profitable private projects were fran- chised to national and international firms. Projects worth more than $4 billion are operational or are close to being operational. An additional $2.5 billion in projects was auctioned or was scheduled for auction dur- ing 2002, but construction had not yet started; another $650 million was under consideration at the time of this writing but had not been evaluated in detail. By the mid-1990s, the government discovered that it faced bottlenecks in seaports--a serious problem since most Chilean international cargo is transported by sea. Each port had multiple private cargo transfer and stor- age operators, but there was little investment in equipment, and activities were not well coordinated. The government decided to franchise port ter- minals (frentes de atraque) to private operators.4 Given the scarcity of ports in Chile because of geographical reasons, terminals can be consid- ered essential facilities. To increase efficiency and investment in the ports, the main terminals were auctioned under restrictions on horizontal and vertical integration designed to prevent monopolization of the ports. Overall, the program of infrastructure franchising has been successful. The highway program has encountered few problems, especially com- pared with the experience of Mexico, which eventually cost taxpayers an estimated $8 billion. By now Chile can boast of a substantially upgraded road infrastructure and lower transportation costs. Moreover, since fran- chise auctions were open and competitive, tolls (user prices) should be close to average cost, which is the second-best outcome in the presence of economies of scale. There are, however, potential problems with the THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 203 traffic guarantees the government has included in contracts to facilitate access to loans, since they represent unaccounted contingent liabilities of the government that are pro-cyclical. Finally, there have been noticeable improvements in the efficiency of the privatized ports. The loading and unloading process has become twice as fast in just one year, having a multiplier effect on transportation costs since shorter stays in port mean that more efficient and more capital-intensive ships can afford to operate from Chile. In this chapter we report the effects of privatization on 37 nonfinan- cial firms that were privatized between 1979 and 1999. During this pe- riod 13 additional nonfinancial firms were privatized, but the available data for those entities were insufficient. This is symptomatic of one of the negative features of the Chilean privatization process: the lack of transparency (Hachette and Lüders 1994). There are no public records for privatizations that occurred during the 1970s. Some of these firms subsequently went bankrupt or became private corporations (that is, with no publicly traded shares) and did not publish accounting informa- tion. Therefore, most of the usable data on privatization for Chile cor- respond to the now publicly traded firms that were privatized starting in the early 1980s, which are required to publish financial information. Nineteen of the 37 firms in the sample belong to regulated sectors, so their data are contaminated by the effects of regulation, while the remainder are usually dominant firms in their markets or have sizable market shares.5 In this chapter we report both absolute and normalized (adjusted) changes in various performance ratios before and after privatization. The normalization allows us to compare the behavior of privatized firms to the performance of the sector to which they belong. The rest of the chapter is organized as follows. The next three sections analyze the privatization of state-owned firms and its effect on their per- formance, efficiency, and other parameters. That is followed by a section devoted to a qualitative assessment of privatization of regulated sectors. First, we analyze the privatized utilities and their sectors in more detail. Then we look at the private provision of infrastructure through franchises. The last section concludes. A Brief History of the Privatization of Public Enterprises The Era of State Intervention State participation in the economy has had a long history in Chile, although it has only been truly significant since 1940. After the crisis of the 1930s-- according to Mamalakis (1976), Chile was one of the countries that suf- fered most heavily in the crisis--the country chose an import substitution strategy and more state intervention. Thirty years later, the government 204 FISCHER, GUTIÉRREZ, AND SERRA Table 5.2 State-Owned and State-Seized Firms, 1970­2001 Type of firm 1970 1973 1983 1989 2001 Enterprises 65 179 45 44 38 Banks 1 19 2 1 1 Seized 0 259 0 0 0 Note: Between 1970 and 1973, the government increased from 2 to 68 the number of firms in which it had minority stakes. Source: Corporación de Fomento de la Producción; Hachette and Lüders (1994). Data for 2000 compiled by the authors. owned or had a controlling interest in 65 firms (and 1 bank), 21 of which were created by law and 44 that were controlled by the Corporación de Fomento (CORFO), a government organization created to promote indus- trial production (table 5.2). These firms either operated only in sectors that the state deemed too important to be left to the market or were originally private firms that had gone bankrupt until government intervention saved them.6 CORFO was set up in 1939 to spur economic development through the promotion of investment. It operated through loans and loan guarantees to the private sector, through research and development of projects, and eventually, through their implementation. CORFO established firms that were deemed essential for Chile's development,7 among them Empresa Nacional de Electricidad (ENDESA, 1944), Compañía de Acero del Pací- fico (CAP, 1946), Industria Azucarera Nacional (IANSA, 1953), Empresa Nacional de Telecomunicaciones (ENTEL, 1964), Petroquímica Chilena (Petrox, 1967), Sociedad Química y Minera de Chile (Soquimich, 1968), Celulosa Constitución (Celco, 1969), Celulosa Arauco (1967), and Indus- trias Forestales SA (Inforsa, 1970). There were minority private share- holders in these firms (43 percent in the case of CAP). CORFO was also a minority shareholder in two other firms. Among the firms created by law are Correo y Telégrafos, which has been public since before independence; Ferrocarriles del Estado, founded in 1851; Línea Aérea Nacional (LAN), created in 1931; Em- presa Nacional del Petróleo (ENAP), established in 1950; the Empresa Marítima del Estado, spun off from Ferrocarriles in 1953; the Banco del Estado, established in 1953 by the merger of state-owned financial in- stitutions established in the previous century; Empresa Nacional de Minería, created in 1960; and the Empresa Portuaria de Chile, split off from the customs office in 1960. A change of policy occurred in the late 1960s when the government be- gan a modest process of acquiring private firms. Previously, all SOEs had been created by the state itself, except for those troubled firms unable to repay the CORFO loans. Codelco was established in 1968 to acquire 50 THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 205 Table 5.3 Number of State-Owned Firms, by Year, 1973­2001 Date 1973 1978 1983 1989 2001 State-owned pre-1970 65 46 32 19 15 Acquired 1970­73 113 34 1 0 0 Created 1970­73 1 1 1 1 0 Acquired 1974­78 1 1 0 0 Created 1979­83 10 2 1 Created 1984­89 22 13 Created 1990­2001 10 Total 179 82 45 44 38 Note: The table does not include Pehuenche, which was privatized as a project. It does includes Corporación del Cobre (Codelco) and 10 seaports originating in the breakup of Emporchi. Source: Authors' computations based on Hachette and Lüders (1994) and Corporación de Fomento de la Producción. percent of the shares of the four largest copper mines (copper represented more than 80 percent of all exports). In 1970 Chilectra was acquired by CORFO, which meant that the state owned almost the entire electric power sector. Moreover, in the 1965­70 period, 22 percent of the arable land (4.1 million hectares) was expropriated in a land reform process. Most of the land was not transferred by deed to the peasants (except for 2,600 hectares), but was kept in public hands (Rosende and Reinstein 1986). The pace of state intervention in the economy accelerated in December 1970 when the socialist Allende administration took office with the pro- fessed aim of creating a vast state-owned sector. The target was to acquire all firms whose equity exceeded $500,000 in current dollars, as well as all of the banking sector, the import-export sector, and all utilities. A ma- jority in the Chilean congress opposed this plan, so the executive branch resorted to administrative measures and legal loopholes. First, CORFO offered to buy shares in any bank or publicly traded firm. Given the un- certainty of the times, many investors decided to sell out (table 5.3). In the 1971­73 period CORFO managed to buy a majority share in 113 industrial firms and 14 banks, as well as a minority shareholding in 68 other firms and 5 banks, while creating only 1 new firm (Trans- marchilay). Therefore, in September 1973 the state was a majority con- troller in 179 firms and 15 banks and was a minority shareholder in 70 firms and 4 banks. Another 259 firms were intervened in or nationalized. In this case the administration used preexisting legislation that allowed in- tervention or expropriation of firms when there was a threat of shortages. In the case of strikes, firms stopped operating, so a risk of shortage existed, allowing the intervention of the government. 206 FISCHER, GUTIÉRREZ, AND SERRA To recapitulate, by September 1973 the government controlled 441 firms and 15 banks, and there were few important companies in private hands (the firms under control of the state represented almost 40 percent of GDP). In addition, 66 agro-industrial plants were built or operated, or both, by Socoagro, a subsidiary of CORFO. The state owned 8,979,000 hectares, of which 5,873,000 had been expropriated in 1971­73 (Larroulet 1984), and the share of the economy in the hands of the state was growing apace. The First Round of Privatization After the coup of September 1973, the military government in power be- gan to develop a strategy of economic liberalization. One of its aspects was the return to the original owners of the firms in which the government had intervened. During 1974, 202 firms were returned to their owners; 39 were given back the next year, leaving only 18 firms to be normalized in the next few years. Most of these firms that were returned to their origi- nal owners by 1975 had been in the hands of the government for only a few years and are therefore not representative of SOEs (Meller 1996; Sáez 1996). At the same time, the land that had been expropriated was priva- tized: 28 percent of the land that had been expropriated illegally was re- turned to its original owners, another 52 percent was divided into small landholdings and sold to the peasants at subsidized prices (many of the peasants later resold the land), while the remainder was privatized through public auction or was transferred to the Corporación Nacional Forestal (Hachette and Lüders 1994). Between 1975 and 1977 the government privatized most of the firms that had been acquired in 1971­73. Most of the shareholdings in banks were sold in 1976, leaving a few to be sold the next year. In the 1975­77 period, 70 state-controlled firms were privatized, while 28 other firms were closed and their assets auctioned off (table 5.4). By 1980 the state had control over only 10 of the firms acquired by the socialist government. At the same time, the military dictatorship decided to keep the largest elec- tricity and telecommunications companies. The same strategic reasons made the military government buy a controlling interest in the main tele- phone company in 1974 (thus obtaining control of 100 percent of the telecommunications sector). Of those SOEs that dated to the period before 1970, only 46 were still owned by the state by the end of the 1970s, with the government having either sold or liquidated 30 of them. The state retained the 22 companies that had been created by a special law, but the number of CORFO com- panies shrank from 44 to 11. CORFO sold all of the firms it had acquired through debt capitalization and kept only a portion of the companies it had created. Thus, by the end of the 1970s the state owned the major elec- tric utilities, telecommunications companies, the big mining companies, and a large fraction of the transportation industry (railways, two shipping THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 207 Table 5.4 Privatized State-Owned Enterprises, 1972­2001 Privatized SOEs 1974­78 1979­83 1984­89 1990­2001 Pre-1970 SOEs 10 7 10 3 Acquired 1970­73 60 7 1 0 Created 1970­73 0 0 0 1 Acquired 1974­78 0 0 1 0 Created 1979­83 0 8 1 Created 1984­89 7 7 Created 1990­2001 2 Total 70 14 27 14 Source: Authors' computations based on Hachette and Lüders (1994) and Corporación de Fomento de la Producción. companies, and the national airline) as well as the steel mill. The status of 13 firms is unclear, since they either were liquidated or went bankrupt shortly after privatization. The larger firms were sold at public auctions, although there were postauction negotiations with the auction winners (Hachette and Lüders 1994). The smaller firms were sold directly. Overall, the objective seems to have been to maximize state revenue, which explains why the govern- ment usually offered a controlling interest (generally all of the shares owned by CORFO), rather than selling small lots of shares in the open market. In the case of the banks, the government tried to diversify owner- ship by setting a limit of 1.5 percent on holdings, but this limit was raised after being easily evaded by buyers using shell companies. No serious at- tempts were made to attract foreign investors. The objective of maximizing revenue from sales led to a policy of lend- ing money to the buyers. Thus, only 10­20 percent of the bid was required immediately, and there was a one-year grace period, plus seven years for full repayment, with a low (for those times) real interest rate of 8­12 per- cent per year. The government asked for a loan guarantee of 150 percent of the loan value, but the guarantee could be in the form of shares in the company. In the case of the banks, the minimum payment was 20 percent (on average 23 percent was paid up front), and the loan had to be paid in full within two years at a real interest rate of 8 percent. The government offered such easy conditions because the private sector was still undercap- italized due to the effects of the policies of the early 1970s. The Crisis of the Early 1980s Most financial firms, as well as several banks that had been privatized from 1975 to 1979, were taken over by the state during the economic crisis of 1981­83. Beginning in 1981 several banks became effectively 208 FISCHER, GUTIÉRREZ, AND SERRA insolvent because they could not recover loans from troubled companies, many of them related firms, which were either bankrupt or had suffered severe losses.8 The government took over four banks in November 1981 and two more the following year, all of which were later closed. In Janu- ary 1983 the government had to take over eight additional banks that had failed to repay international loans (three of these banks were later closed down). Ironically, most of the financial institutions that had been priva- tized during 1975 and 1976--representing 55 percent of all financial as- sets--were again being run by the state in the early 1980s (Rosende and Reinstein 1986). By December 1984 the accumulated losses of the financial sector rep- resented more than 200 percent of the sector's equity and reserves and 18 percent of GDP (Valenzuela 1989). To continue to have access to interna- tional credit markets, the government had to guarantee all foreign loans of the banks that it had taken over while rescuing local depositors. The gov- ernment also took over many nonfinancial companies, as well as the pri- vate pension funds (Administradoras de Fondos de Pensiones [AFPs]) that were linked to the troubled banks, either because they had unpaid loans from the banks or because they were owned by the same economic con- glomerates (Rosende and Reinstein 1986). Between 40 and 90 firms were taken over by the state, giving rise to the so-called área rara ("gray sector"). Hence in the 1982 crisis, the state once more became the controller of many previously privatized firms. This new period of state control was fairly short-lived, and firms were not considered to be truly state-owned. The trigger of the crisis may have been international in origin (a large rise in the prime lending rate in 1981 plus a moderate fall in the terms of trade), but the impact was amplified by mistakes in economic policy, some of which were related to the privatization process. The Chilean financial system was sufficiently fragile that the rise in interest rates, cou- pled with the stoppage in capital inflows, weakened the new conglomer- ates, most of which had high debt-to-asset ratios. The mechanisms used for privatization in the 1970s led to concentrated property holdings and gave rise to conglomerates that were highly leveraged (Sanfuentes 1984). In many cases, the buyers of banks used bank deposits to pay the loans incurred in acquiring the banks. When nonfinancial firms were priva- tized in 1976­77, the new owners of banks also used their clients' deposits or loans from other financial institutions to buy the firms. As mentioned, the buyers were required to put up collateral for 150 percent of the loan used to buy state-owned firms, but shares in the firm could be used as collateral. In this way, large and highly indebted conglomer- ates were formed. The lack of regulation in the banking system made it easy for the banks to lend money to related firms, and even when restrictions were imposed on related lending, they were easily eluded. In the case of the two main banks, 21 percent and 50 percent of all loans went to conglomerate members. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 209 Bank regulators did not keep track of the quality of the loan portfolios. Ideology played its part in the lack of regulation, since government econ- omists argued that if the banks were receiving deposits, private investors must have decided that the projects to which the banks were loaning money were profitable, and regulation was unnecessary. However, the regulators failed to realize the effect of implicit deposit insurance on their assumptions. In 1976 depositors in a failing newly privatized bank had been protected from losses, which created the perception among deposi- tors of the existence of implicit state insurance. Moreover, investors in the conglomerates believed that they were too large to fail (Vergara 1996). Regulatory changes to monitor the quality and supervise the concentra- tion of bank loans were put in place only in 1982; a stringent new bank- ing law was not introduced until 1986. In addition to the financial resources from their affiliated banks, the two largest conglomerates managed mutual funds (82 percent), insurance com- panies (53 percent), and pension funds (68 percent) that granted them even more control over the economy (Sanfuentes 1984). These institutions bought shares of firms in the conglomerate, thus raising share prices. The indebtedness of the conglomerates resulted in part from the level of real in- terest rates in the 1975­81 period, which was high because of the excessive demands for credit from the conglomerates to buy even more privatized firms. The high real rates were compensated by capital gains in the stock market. In 1981 the government allowed banks to contract loans abroad, which led to a rapid increase in indebtedness. Firms that had access to in- ternational loans obtained credit at much lower rates than could smaller firms with no access. In less than two years foreign debt doubled, with the two largest groups holding 52 percent of the debt. Starting in 1985 the banks that had been taken over began to be pri- vatized once again. Preferred shares representing 70 percent of equity were sold to new buyers. The banks sold their bad loans to the central bank and were recapitalized. In return, the central bank became a claimant on future profits of the banks.9 When selling the two major banks, the government strove to create a broad-based class of sharehold- ers for two reasons: to provide stability and to make it more difficult to reverse the privatization process. The mechanism was so-called "popular capitalism": buyers were required to put only 5 percent down, while CORFO gave them a 15-year loan for the remainder. There was a 1-year grace period at a zero real interest rate, a 30 percent discount for timely repayment of the loan, and generous tax benefits. The number of shares per buyer was limited (and limits were enforced). Three additional banks were sold to groups of investors. The two main conglomerates had been the owners of the larger AFPs (Provida, Santa María, San Cristóbal, and Alameda), which held 68 per- cent of workers' pension funds. The two largest (Provida and Santa María) were sold under the popular capitalism scheme (without the tax benefits). 210 FISCHER, GUTIÉRREZ, AND SERRA Aetna, which owned 49 percent of AFP Santa María, bought enough shares to get control, while the rest went to small buyers. Banker's Trust bought 40 percent of the shares in Provida, with the remaining shares go- ing to small buyers. The other two AFPs were merged and auctioned un- der the name of AFP Unión. After their recapitalization, the government also auctioned the other firms it had taken over. In most cases, a controlling package was auc- tioned, but in contrast to the procedures of the 1970s, the government required that payment be up-front. Local conglomerates in association with foreign investors bought the major companies. To make the auctions more attractive to foreigners, they were allowed to pay with Chilean bonds that were selling in the market at 60 percent of par value. Unfortu- nately, little information is available about the details of the transactions of that period, as there seem to be no clear records. The Privatization of the Historic SOEs (1985­89) In the period 1985­89, the government privatized 27 firms and closed down 3 other companies, while creating 29 companies by splitting up larger SOEs. Only 3 of the SOEs created were entirely new (Zofri, Metro, and Cotrisa). In particular, 11 water and sewage companies were created from the national water works. The firms that were sold in this period Table 5.5 Revenues from Privatization of Chilean Public Enterprises, 1985­89 (US$ million) Firm 1985 1986 1987 1988 1989 Total Electricity firms (13) 16.4 124.3 393.0 632.5 77.9 1,244.1 Telecommunications 0.9 55.6 35.5 344.0 192.1 628.1 firms (3) Soquimich 4.7 85.4 71.5 60.9 0.0 222.5 CAP 12.1 139.5 53.2 0.0 0.0 204.8 ECOM 3.2 0.2 0.0 0.0 2.8 6.2 IANSA 0.0 8.8 1.0 50.8 8.0 68.6 Lab. Chile 0.0 2.8 3.8 18.1 3.1 27.8 Schwager 0.0 0.0 6.1 2.2 7.0 15.3 ENAEX 0.0 13.4 0.0 0.0 0.0 13.4 Isegen 0.0 0.0 0.0 0.0 5.6 5.6 LAN Chile 0.0 0.0 0.0 7.0 75.9 82.9 Chilefilms 0.0 0.0 0.0 4.5 0.0 4.5 Isevida 0.0 0.0 0.0 0.0 8.8 8.8 Total 37.3 430.0 564.1 1,120.0 381.2 2,532.6 Source: Corporación de Fomento (CORFO) annual reports. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 211 were pre-1970 SOEs or firms that were spin-offs of pre-1970s SOEs that were created in order to be privatized. The state sold 12 pre-1970s SOEs and 14 firms that were spin-offs of SOEs, as well as two other firms that were acquired in the 1970s and had been kept for strategic reasons. Most of the firms sold in this period were utilities and included 13 electric and 3 telecommunications companies (table 5.5). Four different privatization mechanisms were used in this period. The first was through best price offers for the firm or for controlling packages in open international auctions. The second mechanism was the auction of noncontrolling packages of shares on the stock market. A third mechanism was the direct sale of shares to the workers of privatized companies, pub- lic employees, and small investors--the so-called labor and popular capi- talism. Workers and public employees financed the purchases of shares by using their severance benefits and loans from public institutions at subsi- dized interest rates. Private pension funds participated in the privatization process through the acquisition of packages of shares in the stock market. Finally, public utility users that needed to connect to the system or increase the capacity of their connection were required to pay for the infrastructure in return for shares in the company (Bitrán and Sáez 1994). Privatization during the 1990s The first elected government after the military regime (1990­94) stopped the privatization process almost completely, in contrast to the second elected government (1994­2000), which gave new life to the privatiza- tion process. From 1994 to 2001, 14 companies were privatized while 4 others were closed down. During the same period 12 new firms were cre- ated, 10 of them being subdivisions of Emporchi, the port authority. By late 2001, 38 firms remained in public hands; 14 of these were pre-1970 SOEs, and 24 had been created after 1980, mainly by splitting up tradi- tional SOEs. The current SOEs include the largest copper mining com- pany, the oil refinery, 9 regional water and sewage companies, the post office, the subway, a copper refinery, 10 ports, the post office, and a commercial bank. Between 1994 and 2000 the government used public auctions to sell the state-owned transportation companies: two shipping companies (Empre- mar and Transmarchilay), a cargo railway company in the northern part of the country (Ferronor), and the cargo railway company in the central zone (Fepasa). It also sold the remaining 27 percent of the national airline on the stock exchange. Ferronor bought the northern rail system, which consists of several lines that run from mines in the Andes to ports and carry minerals. It has been a successful company. Fepasa got the cargo concession in the rail system south of Santiago, but the lines were kept by the state (which also kept the money-losing passenger rail system). Unfor- tunately for this second company, the rail lines were in worse shape than 212 FISCHER, GUTIÉRREZ, AND SERRA expected, as was the case with the rolling equipment. Moreover, its hold- ing company had financial problems, and initially Fepasa made some busi- ness mistakes. Hence, it is only after several years in private hands that it has been able to achieve positive operational flows. During this period, the state also completed the privatization of the electric power sector. Edelnor was privatized in the 1991­94 period. In 1995 Codelco, the state-owned copper mining company, hived off and then sold its thermal power plant (Tocopilla). A 37.5 percent stake in Col- bún was sold in 1996. Before the sale just over 15 percent of Colbún was traded on the stock market. In December 1997 the government auctioned 4.65 percent of Colbún in the stock market, selling its remaining shares in 2001.10 The two most important privatizations of the 1990s were those of the three largest water and sewage companies. Unlike other public services, the military regime did not privatize sanitation services. The need to raise rates significantly before it became feasible to privatize these services was a hindrance to the sale of the water companies. The military government felt that privatization followed by a substantial price hike would have been politically unpopular. In fact, in the late 1980s, water rates were on aver- age less than half of what was needed to finance provision of the service, with prices covering less than 20 percent of outlays in the desertic north- ern regions. Before privatization, however, rates had to be raised so that the water companies could cover their costs. Modernization of the water and sewer sector began in 1977 with the creation of the Servicio Nacional de Obras Sanitarias (Sendos). This en- tity absorbed several agencies belonging to different ministries and made it possible to reduce the work force from 10,000 to 3,000. Apart from regulatory responsibilities for the whole sector, Sendos was charged with providing water services to the regions. In the same year, state-owned wa- ter companies were set up based on preexisting companies both in the Santiago metropolitan region (Emos) and in the central zone, or Region V (Esval). In 1989, 11 regional joint-stock companies affiliated with CORFO were created out of Sendos. In 1988 a new regulatory framework was set up for the sector, closely matching its electricity sector counterpart. The new rate system allowed for the self-financing of efficient firms. Pricing zones with relatively homoge- neous costs were also established. The new pricing system was introduced gradually in 1990, and charges rose by an average of 90 percent in real terms between 1990 and 1994, although the rate adjustment process was still not complete in all regions. The rise in prices was steeper in areas with higher costs, exceeding 500 percent in some cases, and by 1998 average re- gional water rates ranged from $0.43 to $1.21 per cubic meter. Arrears were cut from 7.9 percent in 1990 to 2.9 percent in 1994, as a result of a business-oriented approach and by the possibility of cutting off service to customers in arrears. In 1994 the average rate of return on equity among public water companies was 6.3 percent. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 213 The Frei administration decided to privatize water companies. How- ever, it wanted to strengthen the regulatory framework before selling the firms, since it was not totally satisfied with the way the regulation of pri- vatized public utilities was working. In 1995 the administration sent a bill to Congress improving the rate settlement process. Congress approved the bill in December 1997 after a prolonged and heated debate, since it was assumed that the bill was in preparation for privatization. The privatization of the water sector began in 1998. Since then the three major water and sewerage companies have been sold. A scheduled rate revision took place in the two major sanitation companies after pri- vatization. The revision resulted in a 20 percent increase in the rates of both firms. The increase is more or less in line with the 10 percent cost of capital estimated for the sector (the public firms had a 7 percent rate of re- turn on equity), and the sale prices reflect these numbers. The privatized firms are investing in sewage treatment plants and this will lead to further rate hikes. In 2002 the government was in the process of franchising its re- maining water and sewerage companies. Data on Privatized Chilean Firms Given the history of the privatization process described above, the data are difficult to obtain in usable form. We excluded from our analysis privati- zations that took place between 1975 and 1979, first, because most firms that were privatized during that period were managed by the government for only a few years. Second, the political and economic turbulence of the 1970s renders available information highly idiosyncratic. In fact, eco- nomic data from 1971 to 1973 show significant distortions, and the eco- nomic recovery did not start until 1976. Moreover, accounting standards were lax and were changed in 1982. Hence, in this chapter we focus on the 54 firms privatized from 1979 to 2001. Two of these are insurance com- panies and are thus excluded from the sample, which includes only nonfi- nancial firms. Two water companies that were privatized in 1999 have also been excluded, since there is only one year of postprivatization data. Of the remaining 50 firms, only 34 are publicly traded on the stock ex- change and thus are required to provide financial information to the pub- lic, while the other 16 have no public disclosure requirements.11 However, we were able to access the information of 3 of the firms in the latter group (Empremar, Fepasa, and Ferronor) and they have been included in the sample.12 This leaves 37 firms for which we have usable data. Data Problems The basic source of information is the so-called FECUs (Ficha Estadística Codificada Uniforme), the standardized quarterly reports that companies with publicly traded shares (plus some other firms designated by law) must 214 FISCHER, GUTIÉRREZ, AND SERRA provide the Chilean Securities and Exchange Commission. The December FECU typically includes the annual financial report and other informa- tion, including the number of workers in the firm.13 FECUs have been re- quired since 1982 and are available in digital form.14 Before the important changes in the accounting standards introduced in 1982,15 the accounting information of firms was not standardized; it is thus less descriptive of the true financial status of firms. An additional source of information is the annual company reports. However, the data in the annual reports are not standardized and are therefore less useful. No source exists for the following data at the firm level: number of white- and blue-collar workers, average wages, salary differentials, and output price indexes. Another important data limitation is the lack of read- ily available physical data, since the output of some products is described only in the annual reports; the products themselves change between annual reports; and, finally, we have no price index of these products. As an ex- ample, a steel company might produce steel in ingots and as sheets and bars, iron ore, and other products. Should we assume that physical pro- ductivity has gone down because steel ingots per worker have fallen, or does that decline result from a change in the demand for ingots as com- pared to steel bars and iron ore, for instance? In principle, one might look at sales per worker as an index of produc- tivity, and we do this. However, most of the firms we analyze either are reg- ulated or face very few competitors, so that prices are not determined in a free market environment. More sales per employee after privatization could therefore be attributable to either higher prices or higher productiv- ity, or to a combination of both factors. In our analysis, in addition to working with the whole sample of firms, we analyze the performance of the group of regulated and unregulated firms. Our definition of a regulated firm is slightly ad hoc: a firm is regulated if the government, through inter- ventions in the market or through rate regulation, has the ability to change its profitability.16 This implies that all electric distribution companies are regulated for our purposes, as are local and long-distance telephone com- panies.17 IANSA, the sugar refining monopoly, is a doubtful case, since high trade barriers protect it. IANSA was assumed to be unregulated, but results do not change if we group it among the regulated firms. We have defined investment as the difference between net physical assets (that is, accumulated depreciation is subtracted from accumulated invest- ment) at time t minus the same variable at t 1. To measure the impact of privatization on firms, we exclude the two years immediately before and after privatization. As described in La Porta and López-de-Silanes (1999), there is a potential cleansing effect in the ac- counts before the sale of the company, while the two years following priva- tization might not be representative of the postprivatization performance if the firm is still undergoing a reorganization process. Hence, we compare years three to five before privatization with years three to five after privati- THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 215 zation. Nevertheless, we have examined the data in the three years before privatization to get a feeling for the cleansing effect. Treatment of Mergers and Divestitures Using data three to five years before and after privatization creates its own set of problems. Some firms were spun off just a short period before being privatized. For instance, Chilectra was divided into three firms before pri- vatization. Six regional distribution companies, Colbún, and smaller gen- erating companies were spun off from ENDESA. The problem is that preprivatization FECUs for the newly independent firms do not exist, and determining the change in performance caused by privatization is there- fore not straightforward. Similarly, in the case of mergers, we do not have postprivatization independent FECUs. In these cases we have to rework the data in order to assign the assets of the original firms to each daugh- ter firm. Conversely, when the firms are merged, we have to "disassemble" the merged firm into its original constituents. The procedure we follow is to assign the different variables in propor- tion to their fraction of the merged firm at the time of privatization. For example, suppose that firm A splits off from firm Z and both are priva- tized. To obtain data on a variable prior to privatization, we take the data at privatization and consider the proportions of that variable for the com- bined firm. We then assign the data in the combined FECU or variables not in the FECU (before privatization) in those proportions. A similar procedure is used to analyze data for merged firms. Data Adjustments To eliminate the effect of economic conditions on the performance of firms, we also present normalized comparisons. Subtracting the average values of the performance ratios for the two-digit SIC group to which they belong normalizes firm performance ratios. Although the two-digit de- composition encompasses widely differing industries, going to more digits in the decomposition would not have been useful, since the firms in ques- tion represent most if not all of the industry at more detailed SIC levels. We have subtracted the two-digit averages rather than using ratios because of the extreme variations in these ratios, which would have given excessive weight to some observations. Moreover, the interpretation is simple: if an adjusted ratio for a privatized firm is negative, the ratio for that firm is worse than the average of that variable in its (two-digit) sector. Some two- digit average ratios have been treated differently because of the extreme variation in the data. For example, consider the ratio of net income to sales (or to property, plant, and equipment) for a small timber company in the control group that sells a forest. This is nonoperational income, there are very few sales, and the ratio of net income to sales is astronomical. In these 216 FISCHER, GUTIÉRREZ, AND SERRA cases we have taken the sum of net income for firms in the control group and divided it by total sales of the firms in the control group to obtain a more reasonable result. Effects of Privatization on Chilean Firms We analyze the firms before and after privatization both in terms of absolute performance and by comparing them to a benchmark given by the average behavior of their sector at the two-digit level, as mentioned previously. The first part of the analysis in each subsection is devoted to unadjusted data. Perhaps the most interesting result we obtain is the dif- ference between the performance of regulated and unregulated firms. Detailed tables and figures showing the pre- and postprivatization performance appear in the appendix. Profitability Before privatization, in contrast to Mexico (see La Porta and López-de- Silanes 1999), Chilean SOEs were fairly profitable. That was the case for most of the large SOEs that were privatized, as shown in table 5.6, so the firms did not have to go through the large changes that were required in other countries. Even though, on average, privatized firms were profitable, several smaller firms (and a few large ones such as ENDESA in 1985) did have losses before privatization. If anything, at the time there were com- plaints that the government was selling the crown jewels. Hence, the scope for efficiency benefits from privatization was relatively small. Table 5.6 Net Income to Equity for Privatized Firms before Privatization, 1970­86 Year Firm 1970 1974 1979 1983 1986 privatized CAP 10.9 0.5 0.2 0.7 2.3 1986 Chilectra 0.5 3.2 2.6 4.6 -- 1986 CTC 0.7 4.1 1.7 11.9 10.9 1987 ENDESA 0.3 4.3 2.4 6.4 4.9 1988 ENTEL 0.7 3.4 12.3 13 35.4 1988 IANSA 9.3 12.1 9.8 24 5.5 1988 Lab. Chile 4.1 7.9 0.5 196.4 12.8 1989 Soquimich 65.3 11.9 7.9 10.1 30.8 1986 -- Not available. Source: Sáez 1996. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 217 Using table 5A.1 of the appendix, which is analogous to the first part of table IV in La Porta and López-de-Silanes (1999), we can analyze prof- itability before and after privatization. If we consider the profitability variables--operating income to sales (OI/S), operating income to physical assets (OI/PPE, or property, plant, and equipment), net income to sales (NI/S), and net income to physical assets (NI/PPE)--we observe that there seems to be a significant change in the profitability ratios before and af- ter privatization. In particular, we observe that NI/S rose from less than 2 percent to 13 percent, on average, and that NI/PPE rose from less than 4 percent to more than 16 percent. Moreover, the profitability measures are strongly positive. This change in profitability, however, comes mainly from the change in the results of regulated firms. While the profitability ratios improve for the group of unregulated firms, there is enough variation in the results that we cannot show that the change is significant. The important ratio NI/PPE increases from a low of 4 percent to a fairly reasonable rate slightly above 12 percent. By contrast, in the case of regulated firms the change in profitability ratios is far more important. The ratio NI/PPE rises from 3.5 percent to 20.5 percent, and the same pattern of large in- creases in profitability of regulated firms occurs for the other profitabil- ity ratios. When we consider adjusted variables (appendix table 5A.4), obtained by normalizing the profitability ratios by subtracting the average ratio of their sector, we find that the improvement is less significant. This implies that part of the improvement observed in the unadjusted data can be explained largely by a simultaneous improvement in the average prof- itability of the sector. Nevertheless, the ratio NI/PPE for the sample of firms rises from 1.5 percent above the average in the sector to almost 10 percent. What is interesting is that the profitability ratios with respect to sales are not significantly different from those of the industry as a whole, which seems to indicate that the increased profitability is related to better use of physical assets or, alternatively, to overinvestment before privatization. As in the case of unadjusted variables, most of the change in prof- itability results from the increase in the profitability ratios of the regu- lated firms. While adjusted profitability of unregulated firms increases after privatization, the increase is insignificant. Moreover, the profitabil- ity of these firms is not significantly different from that of the other firms in their sector. By contrast, all adjusted profitability ratios except for NI/S increase significantly after privatization in the group of regulated firms. Moreover, these firms, which had average profitability similar to that of their sectors, became much more profitable afterward, seeming to indi- cate that the regulators were unable to pass the gains in efficiency on to consumers. 218 FISCHER, GUTIÉRREZ, AND SERRA Efficiency Efficiency is described by the ratios of cost per unit (cost/sales) and sales to physical assets (S/PPE), shown in appendix table 5A.2. The cost-per- unit ratio falls by a small but significant amount at the 10 percent level for the sample of privatized firms. The S/PPE ratio falls slightly but insignifi- cantly. Once again, there is a large difference in the behavior of regulated and unregulated firms. Cost per unit falls significantly for regulated firms, while it barely changes for unregulated firms. Similarly, the S/PPE ratio in- creases significantly (at 10 percent) for the regulated firms, whereas it falls for unregulated firms. When we examine adjusted efficiency ratios (appendix table 5A.5), we observe that there is no difference between the privatized firms and the cost per unit in their sectors, and there is no change postprivatiza- tion. Moreover, for this ratio there is no difference between regulated and unregulated firms. Things are different for the S/PPE ratio, since the privatized firms seem to have much higher ratios than the average for their sector. Assets and Investment Appendix table 5A.3 shows variables related to assets and investment. We use net investment, defined as the year-to-year difference in the depreci- ated stock of fixed assets. The average value of the logarithm of physical assets (the log of the geometric mean of PPE) shows a slight increase, which is not significant, after privatization.18 The change is concentrated in the regulated firms, where the effect is statistically significant. The in- vestment to sales ratio (I/S) fell after privatization but not significantly. This would seem to indicate that firms invested more productively. An al- ternative explanation is that SOEs that operated in a competitive setting were investing efficiently before privatization so there was not that much scope for improvement. The ratio of investment to physical assets (I/PPE) for the whole set of firms fell, but again, the effect is not statistically significant. There was a fall in this ratio for the unregulated firms and an increase for regulated firms (again, insignificant). The ratios of investment per employee (I/em- ployee) increased, but not significantly. Physical assets per employee (PPE/employee) increased significantly after privatization. This seems to imply that workers had access to better equipment, but what is interesting here is that it is in the unregulated firms that the increase is more signifi- cant (at 10 percent). When we consider adjusted ratios (appendix table 5A.6), no significant changes appear in the behavior before and after privatization, except for the investment-to-sales (I/S) ratio, which is lower than average in the sec- tor before privatization and approaches the average--but is still lower-- THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 219 after privatization. This effect is significant both for the whole sample of firms and for the regulated firms. Productivity In appendix table 5A.2 we show the labor productivity ratios of sales to em- ployees and operating income to employees. Both of these ratios show that productivity increased significantly as firms became private, as expected. Once again, however, most of the change was attributable to the behavior of regulated firms, for which sales-to-employment ratios increased by 88 percent and operating-income-to-employment ratios rose by 325 percent. We were unable to obtain employment data for all the firms in a sector at the two-digit level, so we have not been able to compare the growth in pro- ductivity of privatized firms with the other firms in their sectors. We have also examined physical productivity for firms for which we could obtain measures of physical product (tons, passengers per kilome- ter, gigawatt hours [GWh], and so forth) as shown in appendix table 5A.8. We have used these variables to construct productivity ratios, and then we have taken the percentage difference before and after privatiza- tion. The results show that firms increased their productivity by about 25 percent on average after privatization, but there is enough variation in the data that this is not significant for the whole sample nor for regulated or unregulated firms separately. Some caution is required in the use of these data, however: most firms have more than one line of production, and therefore a fall in physical productivity on the basis of one product may mean nothing. As an example, the airline LAN Chile seems to have de- creased its productivity in terms of passengers per kilometer after privati- zation. However, after privatization, the firm launched a successful cargo branch, the revenues of which are generally on par with those of the pas- senger segment of the company. Therefore, the data in this section may show that productivity has increased in physical terms, but without the prices of these different products and their production, this comparison is not very informative. Employment As can be seen from table 5.7, there is no evidence that firms fired work- ers during the 1983­92 period, which includes the years in which the firms were privatized. In fact, it appears that firms took on more workers on ag- gregate. Moreover, it is clear that SOEs reduced their employment levels several years before they were privatized (more than three years in most cases). However, different firms were privatized at different times, and therefore it is interesting to see if this continues to hold for the complete sample of privatized firms, considering the time at which they were priva- tized. To examine this issue, we use appendix table 5A.7. Again, there is 220 FISCHER, GUTIÉRREZ, AND SERRA Table 5.7 Employment Changes in Privatized Firms, 1970­92 Year Firm 1970 1973 1979 1983 1986 1992 privatized CAP 7,025 11,637 9,321 6,519 6,667 9,643 1986 Chilectra NA 4,250 4,196 3,846 4,133 4,712 1986 CTC 5,887 7,252 7,206 6,338 6,938 8,504 1987 ECOM 188 341 333 165 149 -- 1986 ENAEX 344 340 394 388 470 -- 1987 ENDESA 6,512 8,504 4,270 2,705 2,905 2,980 1988 ENTEL 1,161 1,458 1,236 1,338 1,402 1,748 1988 IANSA 2,827 2,881 1,597 1,079 2,027 1,561 1988 Lab. Chile 3,608 4,546 2,059 1,372 883 797 1989 Soquimich 10,814 10,684 7,109 4,096 4,704 3,242 1986 -- Not available. Note: ECOM went bankrupt before 1992. Source: Sáez (1996), except data for 1992, which is from the FECUs (Ficha Estadística Codificada Uniforme). Data for ENDESA (apart from 1970) from Hachette and Lüders (1994). Data for Chilectra from Sáez (1996), except for 1986 and 1992. Those years obtained by aggregation of all the firms that were originally part of the firms in 1980 using data in Hachette and Lüders (1994) for 1986. no evidence that firms fired workers after privatization. Employment in- creased slightly but not significantly after privatization: the average firm grew from 1,193 to 1,381 employees. Both regulated and unregulated firms grew in size. Note that on average, unregulated firms are larger. Privatization of Regulated Sectors: The Efficiency of Privatized Utilities In this section we provide an assessment of the privatization-cum- regulation process of electricity and telecommunications companies car- ried out between 1985 and 1989. The evaluation considers the aims of the privatization efforts, namely, to provide capital for expansion of util- ities that the state was not able to fund at the necessary level, to enhance the efficiency of enterprises, and to transfer those efficiency gains to con- sumers. In the previous section we showed that SOEs increased their profitability and efficiency after privatization. These changes, however, are in line with those of their corresponding industries in the case of un- regulated firms. But firms that provide regulated services stand apart. Their profitability increased more than did that of their unregulated counterparts. This difference must be attributed to the interplay between privatization and regulation. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 221 The Privatization Process The government privatized most of the telecommunications and electric- ity industries between 1985 and 1989. Some of the smallest companies were sold through public auctions. Larger firms were privatized through a variety of mechanisms: sale of shares on the stock market; the periodic auction of packages of shares on the stock market; and the direct sale of shares to employees of privatized firms (labor capitalism), public employ- ees, and small investors (popular capitalism). In the late 1970s the telecommunications industry was dominated by two public enterprises: CTC, which provided basic telephone service throughout almost the entire country, and ENTEL, the only international long-distance provider. These two companies shared the domestic long- distance market. The state also owned two regional local telephone com- panies (CNT and Telcoy) and Correos y Telégrafos, which provided tele- graph service. In 1982 the government sold CNT and Telcoy through public bidding. However, the privatization of the large telecommunica- tions firms started only in 1985. By the end of 1987, 25 percent of the equity of CTC was in private hands. In 1988 the government sold 45 per- cent of the ownership of the company to a foreign investor. In the case of ENTEL, in 1985 and 1986 the government sold 30 percent and 3 percent, respectively, of its shares, most of which were acquired by pension funds. In 1988 the state further reduced its stake in ENTEL to 37.7 percent. This time, company workers were the main purchasers (12.5 percent). The rev- enues from the privatization process of the main telecommunications firms appear in table 5.8. The privatization of the two largest electric power companies (EN- DESA and Chilectra) started in 1986. To create competition in the wholesale electricity market, they were restructured before privatization. The restructuring involved separating distribution from generation. EN- DESA, the largest company, was divided into six generating companies, six distribution companies, and two small, isolated companies combin- ing generation and distribution in the southern part of the country. Table 5.8 Privatization of Chilean Telecommunications Firms, 1984­89 (US$ million) Firm 1985 1986 1987 1988 1989 Total ENTEL 0.2 36.7 8.4 81.8 105.0 232.2 CTC 0.7 4.7 27.1 262.2 87.1 381.7 Telex 0 14.2 0 0 0 14.2 Total 0.9 55.6 35.5 344.0 192.1 628.1 Source: Corporación de Fomento (CORFO) annual reports. 222 FISCHER, GUTIÉRREZ, AND SERRA Table 5.9 Privatization of Electric Power Firms, 1984­89 (US$ million) Firm 1985 1986 1987 1988 1989 Total Distribution Chilmetro 10.0 36.0 83.3 0 0 129.3 Chilquinta 2.4 11.1 18.7 0 0 32.2 Emec 0 6.0 7.5 0 0 13.5 Emel 0 7.9 0 0 0 7.9 Emelat 0 0 9.7 0.9 0 10.6 Emelari 0 0 0 0 3.1 3.1 Eliqsa 0 0 0 0 4.8 4.8 Elecda 0 0 0 0 6.1 6.1 Generationa Endesa 0 0 180.0 585.4 63.8 829.2 Pullinque 0 0 62.0 0 0 62.0 Chilgener 4.0 22.2 31.8 33.8 0 91.8 Pilmaiquen 0 41.1 0 0 0 41.1 Integrated Edelmag 0 0 0 4.8 0.1 4.9 Total 16.4 124.3 393.0 624.9 77.9 1,236.5 a. Excludes Pehuenche, which was sold as a project for $7.6 million. Source: Corporación de Fomento (CORFO) annual reports. Chilectra was divided into three firms: a generating company and two distribution companies. Most of the firms were under private control by 1989. The revenues from the privatization process of the main electric- ity firms appear in table 5.9. The Regulatory Framework Some services provided by utilities were considered to be natural mo- nopolies, and therefore the development of regulatory institutions pre- ceded their privatization. Regulatory bodies were created in the late 1970s for each sector: the National Energy Commission (CNE) and the Undersecretariat of Telecommunications, respectively. They are re- sponsible for granting operating licenses, monitoring technical stan- dards, and setting rates for services where competition is insufficient. Regulation, operation, and, to some extent, policymaking had previ- ously been in the hands of the SOEs themselves. Moreover, new regu- latory legislation was introduced in 1982. The aims of these laws were to create the conditions for competition to arise whenever possible and to guarantee, in cases where there was insufficient competition, that THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 223 the efficiency gains expected from privatization would be transferred to consumers. Under these rules, concessions to operate utility services are not ex- clusive, and objective, nondiscriminatory criteria govern the granting of licenses. Only technical reasons, as in the case of mobile telephony, may limit the number of operators. At the same time, legislation mandates that service be provided within the area of the concession, defines conti- nuity and quality standards, and requires interconnection with other firms when regulators deem this to be necessary. Rates of regulated serv- ices are based on the long-term marginal cost of a hypothetical efficient firm. Prices are set every four (electricity distribution) or five (basic telephony) years, and within the price-setting periods they are indexed to the prices of the main inputs used to provide the service. The separation of rates from current costs is intended to create an incentive for firms to be efficient. Chile's regulations did provide for open access to essential facilities but did not initially regulate access charges. Moreover, Chilean legislation does not preclude vertical integration. Thus, in 1992 Enersis, a holding company that owned distribution companies supplying 44.4 percent of the market in the Central Interconnected System, took control of ENDESA, the largest power generation company, which in turn owned the main transmission system. It is fairly well known that regulated monopolies that are vertically integrated into unregulated segments may have an incentive to sabotage their downstream competitors (Beard, Kaserman, and Mayo 2001). Accusations by their competitors that integrated monopolies were discriminatory led to regulatory changes. In 1994 the telecommunications law was amended to mandate the regulation of access charges to the local telephone network. In 1997 the Antitrust Commission instructed EN- DESA to recharter its transmission subsidiary as a public corporation and open its ownership to the participation of other shareholders. ENDESA sold the transmission subsidiary in 1999. The Electricity Sector. Legislation regulating the electric power sector dis- tinguishes among three distinct activities: generation, transmission, and distribution. Only distribution firms need concessions. Distribution li- censes are granted for indefinite periods but may be canceled if the qual- ity of service falls below the legal standard. Power-generating firms and transmission companies within the same area must interconnect, and they must coordinate their operations through an economic load dispatch cen- ter (CDEC). This center's aims are to guarantee the most economical operation of all generating facilities, to guarantee the right of power- generation companies to sell energy at any point in the system, and to safe- guard the security of the system. All plants must be available for dispatch (refusal to provide energy when requested can lead to severe penalties), un- less maintenance has been scheduled. The optimal operation of the various 224 FISCHER, GUTIÉRREZ, AND SERRA facilities, independently of existing supply contracts, calls for transfers of energy to be made between power generators at the so-called spot price, which is the operational (or marginal) cost of the most expensive plant in operation at a given time. The Chilean regulatory system distinguishes between large and small customers. Large customers, with maximum power demands above 2 megawatts (MW), are free to negotiate the terms of their supply with the various generating firms. Small customers, in contrast, purchase energy from distribution companies at regulated prices, which are made up of two components: the node price, at which the distribution firms buy energy from power-generation firms, and the value added of distribution, which pays for distribution services. Distribution charges are computed for the various urban or rural areas in such a way that an efficient firm operating in an area with those characteristics would make a 10 percent return on the net replacement value of its assets. This charge is calculated as a weighted average of the findings of outside studies contracted by the in- dustry and the national energy commission, respectively, with the com- mission study accounting for two-thirds. This figure is applied to the real firms to calculate average profit levels for the industry over the net re- placement value of assets. If these average profit ratios are more than 14 percent or less than 6 percent, distribution costs are adjusted to the nearest of the two ranges. The node price, in turn, has two components: the price of energy and the price of peak power. To guarantee stable rates for small consumers, the price of energy is computed every 6 months as an average of the marginal costs expected over the next 48 months, using projections of demand, fuel prices, water reserve levels, generating plants under construction, and the indicative investment plan drawn up by the national energy commission. The price of peak power is defined as the annual cost of increasing power during peak hours with the least expensive type of plant. This cost is in- creased to take into account the reserve margin (or security level) of the system. Large customers (including distribution companies) are required to have contracts with generating companies. In turn, every power-genera- tion company must have the capacity to meet the yearly energy contracts, bearing in mind potential dry spells that would affect the hydroelectric plants and the average capacity of thermal-generation units. Power- generation firms must also be able to satisfy peak demand, measured as the average gross hourly demand they have undertaken to supply their customers at the system's peak times. A yearly determination is made of power and energy deficits or surpluses incurred by the generation com- panies with respect to their supply contracts that would give rise to trans- fers between producers. The terms of energy transfer arrangements are negotiated between the firms, while transfers of peak power are made at the price set by the regulatory commission. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 225 Finally, power-generating firms pay the marginal cost plus a fee (the "basic" fee) for the use of the transmission lines. Given that there are sig- nificant economies of scale in building lines, marginal-cost pricing does not allow for recovery of all transmission costs. The difference between the total cost of a line and the revenue collected through marginal costs is designated as the basic fee. Then, for each line, the basic fee has to be distributed among the various power-generating firms. The basic fee is negotiated between the transmission company and the generating com- pany, and disagreements must be settled by arbitration. The assignment of the basic fee cost of a line among the various generating firms is based on the firm's demand at the moment of maximum total demand. The foregoing criteria have no solid conceptual basis, particularly with re- spect to the assignment of the entire transmission cost to the generating firms. Telecommunications. The legislation governing telecommunications gen- erally provides for free market pricing of telecommunications services. However, rates are regulated for those services that the Antitrust Com- mission considers to be provided under conditions of inadequate compe- tition.19 The telephone companies themselves, on the basis of guidelines set by Subtel (Subsecretaría de Telcomunicaciones, a department under the Ministry of Transportation and Telecommunications), carry out the studies that are used to set rates. The companies hand in these studies to Subtel, which has 120 days to present its objections and counterproposals. A committee of three experts arbitrates disagreements between the com- panies and Subtel, with respect to both guidelines and objections. The company appoints one member of the committee, the regulator appoints the second one, and the two parties agree on the third. Although the reg- ulators make the final decision, they tend to follow the recommendations of the experts, since the companies are otherwise likely to go to court. The ambiguities of the 1982 law had created a legal monopoly in the long-distance service market. In 1989 a number of companies applied to Subtel for licenses to operate this service. The final decision was not handed down for several years because of the indecisiveness of the courts as to whether or not vertical integration of local and long-distance service should be allowed. Finally, in 1993 the Antitrust Commission authorized the participation of local telephone companies in the long-distance mar- ket. In 1994 the law was modified to introduce competition to long- distance service through a multicarrier system, and following the ruling of the commission, it imposed restrictions on local telephone companies that wished to operate in the long-distance market. In the first place, they had to do so through subsidiaries organized as independent joint-stock com- panies, subject to supervision by the Superintendent of Securities and In- surance (the Chilean Securities and Exchange Commission). The law also required that local telephone firms not discriminate among long-distance 226 FISCHER, GUTIÉRREZ, AND SERRA carriers with respect to quality of service and information on long-distance traffic demand. Moreover, the charges to access the local network were regulated. In 1988 the government set standards for mobile telephone service, al- though an early entrant had had a concession since 1981. The new regu- lations created two concession areas for mobile service, with two licenses in each area, to be granted on a first-come, first-served basis. In Novem- ber 1996 Subtel granted three nationwide personal communications sys- tem (PCS) licenses, using a "beauty contest" in which geographic coverage was the key bidding variable. Until 1999 subscribers had to pay the same fee for both the calls they made and the calls they received, which was a disincentive to the use of mobile phones. In February 1999 the regulator introduced a "calling party pays" principle, under which callers pay for all charges (including the regulated access charges) when using mobile phones. Evaluation of Privatization in the Regulated Sectors An assessment of the privatization efforts carried out at electricity and telecommunications companies between 1985 and 1989 should consider the objectives of the process. As we mentioned earlier, the goals were to increase the efficiency of these firms and to invest in new capacity. There- fore, a complete evaluation of the privatization-cum-regulation process would require, as a counterfactual, a prediction on how the privatized firms would have developed had they remained in the public sector. Here we take a more modest approach. We analyze, for each sector, the post- privatization evolution of a set of variables and relate their behavior to the regulatory changes. In particular, the comparison between regulated utili- ties and those that operate in competitive markets makes it possible to draw inferences regarding the effectiveness of the regulatory system. In some cases, the differences are so significant that inferences can be drawn despite the obvious limitations of this approach. The Electricity Sector. Between 1988 and 2000 electricity generation grew from 16,914 GWh to 39,142 GWh, and installed capacity rose from 4,016 MW to 10,045 MW. In the Central Interconnected System, capacity grew less than electric generation, as peak demand grew at a lower rate during those years because of the use of peak-demand pricing.20 Moreover, power-generating firms have generally invested earlier than required un- der the government's indicative investment plan. Despite the installation of new capacity ahead of the plan, there have been periods of energy shortages in the system because of its heavy dependence on hydroelectric power (in rainy years, such as 1992, 97 percent of generation is provided by hydroelectric power). These outages, however, seem to have been caused by regulatory failures. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 227 Small customers face regulated (forward-looking) energy prices. This price inflexibility, however, should not be an obstacle to market clearing even in very dry years. If generating companies were to compensate users by an amount equal to the outage cost per unit of undelivered energy dur- ing a severe drought, this would make users indifferent to a reduction in their energy consumption, so that in theory, the supply deficit should be eliminated.21 Unfortunately, this compensation mechanism has never been used. Changes to the law (introduced at the suggestion of the largest hy- droelectric operator) eliminated these compensations when the drought is more severe than in the driest year that is used in computing the node price, leaving an incomplete price system. Moreover, no procedures were introduced to deal with that case. After the 1998 blackouts the law was modified and now imposes compensations under all circumstances. In re- sponse, generators have not renewed their contracts with distributors to supply energy at the node price, leading to an impasse. Labor productivity in the privatized companies has improved consider- ably. In ENDESA power generated per worker rose from 2.2 GWh in 1989 to 18.1 GWh in 2001 (table 5.10). If we consider only employees working in the holding company and in the generation subsidiaries, the power generated per worker rose from 6.3 GWh in 1991 to 28.7 GWh in 2000. Labor productivity in electricity distribution also grew substantially after privatization. For example, Chilectra, the largest distributor, has more than doubled its annual sales of electricity since privatization, from 3,612 GWh in 1987 to 9,253 GWh in 2001, and its customer base has grown from 973,000 to 1,289,000. The number of workers, meanwhile, fell from 2,587 to 722, and the number of clients per worker grew from 376 in 1987 to 1,785 in 2001. In addition, energy losses fell from 19 per- cent to 5.4 percent in the same period (table 5.11). Table 5.12 shows the node prices in the two main interconnected sys- tems (Sistema Interconectado Central and Sistema Interconectado del Norte Grande) in current dollars and pesos. There has been a clear down- ward trend in energy prices since generating firms were privatized. In con- stant pesos, the drop is approximately 33 percent for the central system and 73 percent for the northern system. This drop is explained primarily by the decline in the prices of fuels used at the thermoelectric plants (partly due to the appreciation of the peso), which play a part in determining marginal prices. In the central system, the fall was particularly sharp from 1997 on- ward, owing to the anticipation of the arrival of natural gas supplies from Argentina (recall that the regulated price of energy is forward-looking). A greater load factor as a fraction of installed capacity (that is, the system op- erates closer to capacity) and the transfer to consumers of these gains in productivity also help explain the lower prices. The profits of the main power-generating company have increased moderately since privatization, reaching a peak of 15.7 percent return on equity in 1995 and declining in the following years. This decline is a result of unfavorable hydrological 228 Table 5.10 ENDESA: Investment, Power Generation, and Labor Productivity Investment, Investment, Domestic domestic foreign generation Local workers Labor productivitya Year (US$ million) (US$ million) (GWh) All Generationb All Generation 1988 -- -- 7,420 -- -- -- -- 1989 110 -- 6,649 2,980 -- 2.2 -- 1990 -- -- 6,608 2,883 -- 2.3 -- 1991 131 -- 8,521 2,445 1,357 3.5 6.3 1995 180 119 11,783 2,255 1,038 5.2 11.4 1999 301 362.4 13,672 1,383 711 9.9 19.2 2000 145 78 15,346 888c 574 17.3 26.7 2001 -- -- 15,741 -- -- -- -- -- Not available. Note: GWh gigawatt hours. a. In GWh per worker. b. Assumes that 30.5 percent of employees work in transmission in 1991 and 1992 (the 1993 figure). c. The reduction in the labor force is partially explained by the sale of the transmission subsidiary. Source: ENDESA (Empresa Nacional de Electricidad) annual reports. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 229 Table 5.11 Chilectra: Sales, Employees, Labor Productivity, and Energy Loss Item 1987 1988 1989 1990 1995 2000 2001 Sales (GWh) 3,612 3,844 4,070 4,230 6,676 8,854 9,523 Customers 973 1,008 938 935 1,100 1,262 1,289 Employees 2,587 2,565 2,144 2,159 1,801 867 722 Labor 376 393 437 433 610 1,455 1,785 productivitya Sales per 1.4 1.5 2 2 3.7 10.2 12.8 workerb Energy losses 19.8 18.8 16 14 9 5.2 5.4 (percent) Note: GWh gigawatt hour. a. Customers per worker. b. GWh per worker. Source: Chilectra annual reports. Table 5.12 Change in Node Prices and Residential Rates Node price Residential Year Centrala Northa Centralb Northb Centralb 1987 22.2 53.4 2.4 5.9 14.73 1988 26.1 49.7 2.9 5.6 15.87 1989 26.8 51.7 3.3 6.4 16.97 1990 22.6 59.5 3.0 7.9 18.15 1991 19.3 47.8 2.6 6.5 15.83 1992 18.9 36.9 2.9 5.6 15.72 1993 20.4 37.8 3.1 5.7 15.08 1994 21.1 34.7 3.7 6.1 15.31 1995 18.4 23.1 3.7 4.6 15.44 1996 15.4 23.5 3.1 4.8 14.65 1997 12.7 18.8 2.7 4.0 13.77 1998 10.7 14.0 2.1 2.8 12.16 1999 11.4 11.4 2.1 2.1 12.16 2000 14.9 13.7 2.6 2.4 -- -- Not available. a. October 2,000 pesos per kilowatt hour. b. Current US cents per kilowatt hour. Source: National Energy Commission (CNE). 230 FISCHER, GUTIÉRREZ, AND SERRA conditions and the fact that the installation of more efficient combined-cycle gas turbines and the arrival of natural gas from Argentina have reduced the economic value of existing plants.22 Note that the fall in profitability in generation led to further labor productivity gains in the late 1990s. Regulation of distribution firms has been less successful. According to data from the Ministry of Economics, the value added (that is, the charge to consumers) of distribution for Chilectra fell by 18 percent in the rate- setting process of 1992 and by an additional 5 percent in the rate-setting process of 1996. However, this price reduction does not match the effi- ciency gains achieved after privatization. This situation led to increases in the profits of distribution companies. The return on equity obtained by Chilectra increased from 8 percent in 1988 to 32 percent in 1996­98. In the 2000 rate-setting process, rates were reduced by a further 18 percent, which led to lower profit margins at first. In response, Chilectra increased labor productivity substantially. The rest of the industry has gone through a similar process. The profit levels in distribution are much higher than those of the gen- erating companies, which in any event are subject to greater risks, both for lack of a secure market (they operate under competition) and because of the potential for droughts (table 5.13). Some of the distribution industry profits come from unregulated services that are unlikely to become com- petitive, because they are closely related to regulated services, as in the case of the renting of meters. Distribution companies also obtain significant re- turns by allowing phone and cable TV companies to hang cables on their poles. However, these returns are not considered when estimating the in- come of the efficient firm, so, in effect, consumers can pay more than twice for the same infrastructure. Telecommunications. Since privatization, the telecommunications sector has experienced rapid growth, as shown by all indicators. Between 1987 and 2001, the number of lines in service rose by a factor of almost six, so the line density rose from 4.7 to 23.1 lines per 100 inhabitants (table 5.14). In the main local phone company (Telefónica), which accounts for 76 per- cent of all subscribers, average installation time was reduced from 416 days in 1993 to 6 days in 2001, and the waiting list, which in 1987 included 236,000 people, had been reduced to 32,000 by 2001, having reached a peak of 314,000 in 1992. Digital conversion rose from 36 percent in 1987 to 100 percent in 1993. Long-distance traffic also grew significantly. Out- going international traffic rose by a factor of 10, from 21.2 million minutes in 1987 to 241.0 million minutes in 2001. Growth was especially rapid af- ter the introduction of competition in long-distance services (table 5.14). The number of lines per worker in the largest telecommunications firm grew from 74 in 1987 to 845 in 2001 (table 5.15). This period also saw the emergence of new services, such as beepers, data transmission, private networks, and the Internet. However, the new Table 5.13 Profits of the Main Electric Sector Companies: 1987­2000 (percent) Distribution Generation Year Chilectra CGE Chilquinta Saesa ENDESA Gener ElectroAndina Edelnor 1987 -- 18.5 8.8 17.6 5.2 3.1 -- 7.7 1988 7.4 19.7 12.4 19.9 13.7 7.8 -- 2.8 1989 21.3 17.8 19.5 25.9 7.7 8.4 -- 0.7 1990 22.9 17.5 19.5 25.2 6.4 9.4 -- 3.3 1991 19.4 16.5 21.7 26.6 10.4 7.4 -- 3.0 1992 17.3 16.7 42.3 24.9 13.5 7.3 -- 3.4 1993 14.5 18.3 15.7 27.1 11.0 8.6 -- 3.4 1994 17.9 17.1 7.9 22.5 15.7 8.4 -- 7.2 1995 27.6 21.1 9.5 24.8 14.5 11.6 -- 2.3 1996 32.1 22.0 19.8 26.3 12.7 9.5 -- 0.1 1997 31.8 20.0 11.8 22.2 9.9 10.3 5.6 2.5 1998 31.6 20.2 9.3 18.6 3.6 5.9 8.2 2.6 1999 20.6 16.9 111.3ª 16.4 13.5 0.8 6.2 0.9 2000 16.0 15.3 8.9 29.2 9.1 0.3 8.8 3.9 -- Not available. a. Profits of Chilquinta in 1999 include nonrecurring profits from sales of shares. Source: Authors' computations from FECUs (Ficha Estadística Codificada Uniforme). 231 232 FISCHER, GUTIÉRREZ, AND SERRA Table 5.14 Telecommunications Statistics, 1987­2000 International Density traffic Lines in service (lines/100 Mobile phones (millions of Year (thousands) inhabitants) (thousands) minutes) 1980 363 -- -- 8.0 1985 537 -- -- 13.4 1988 631 4.9 -- 27.5 1989 689 5.4 4.9 29.9 1990 864 6.5 13.9 38.8 1991 957 7.8 36.1 47.0 1992 1,283 9.6 64.4 53.1 1993 1,521 10.9 85.2 59.5 1994 1,634 11.6 115.7 63.5 1995 1,891 13.2 197.3 113.6 1996 2,264 15.6 319.5 144.2 1997 2,693 18.3 409.7 198.8 1998 2,947 20.4 964.2 215.0 1999 3,109 20.6 2,260.7 210.2 2000 3,365 22.0 3,401.5 222.5 2001a 3,581 23.1 5,271.5 241.0 -- Not available. a. December 2001 (estimated). Source: Subtel. service that has had the greatest impact is mobile services. At the end of 1997, 16 years after the entry of the first operator, there were only 410,000 subscribers. This number rose sharply with the fall in prices brought about by the entry of new PCS concessionaires. By mid-1998, the number of subscribers had risen to 650,000. With the introduction of the "calling party pays" system in February 1999, the number of subscribers jumped again. By the end of 2001 the number of subscribers had reached 5.3 million.23 This explosion in the number of mobile phones is partially explained by the high level of the access charge to mobile companies set by the regulator. This charge was set too high, so mobile phone companies have been willing to give away the phones in order to benefit from the ac- cess charge paid on incoming calls. Real residential local telephone charges have increased by about 5 percent since privatization (table 5.16). However, there was a simulta- neous rebalancing of rates, which makes it difficult to reach definite conclusions about the evolution of charges. Before 1993 phones rates were much higher for business subscribers than for residential clients. Moreover, clients have benefited from the extension of basic phone THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 233 Table 5.15 Telefónica-CTC: Basic Fixed Phone Statistics Lines Lines per Installation Waiting list Year (thousands) Share Workersa workera time (days) (thousands) 1980 360 99.3 6,911 52 -- 150 1985 505 94.1 6,894 73 -- 181 1988 592 93.7 7,518 79 -- 236 1989 646 93.8 7,366 88 -- 284 1990 812 94.0 7,530 108 -- 308 1991 997 94.3 7,994 125 -- 241 1992 1,213 94.5 7,991 152 -- 314 1993 1,437 94.5 8,133 177 416.0 198 1994 1,545 94.6 7,424 208 208.9 117 1995 1,754 92.8 7,449 235 169.8 52 1996 2,056 90.8 7,073 291 55.4 72 1997 2,394 88.9 6,898 347 38.6 97 1998 2,650 89.9 6,917 383 35.4 58 1999 2,592 83.4 5,649 459 15.4 27 2000 2,701 80.3 4,639 582 4.3 10 2001 2,723 76.1 3,223 845 5.7 32 -- Not available. a. Excludes employees working in subsidiaries. Source: Subtel and CTC annual reports. zones. In fact, many calls that were previously considered long-distance calls are now considered local calls. In 1993 rates were 8.6 percent higher than they were in 1987 as a result of the 1988 rate-setting process. In the 1993 rate-setting process residential rates were further increased by 9.8 percent. This rate hike was explained by the need to compensate for the partial elimination of subsidies from long-distance service to local service and the unification of residential and commercial rates. Starting in 1994 access charges for long-distance calls were sub- stantially reduced. The 1999 rate-setting process can be considered a turning point. Basic phone rates were reduced by 11 percent, at the same time that access charges to the local network were reduced by an average of 72 percent. As a result, in 2001 basic phone rates for an average family were still 5 per- cent higher than in 1987, but access charges for long-distance calls were much lower. The deregulation of long-distance service in 1994 eliminated the need for rate-setting in that market. Deregulation coupled with the re- duction in access charges to the local network led to a dramatic fall in long-distance rates. This is illustrated by the value of a one-minute call to the United States, a route that represents 42 percent of international traf- fic. In 1987 the average per minute cost of a call to the United States was 234 FISCHER, GUTIÉRREZ, AND SERRA Table 5.16 Cost of Local Monthly Telephone Service for the Average Family Year (May) US$ May 1987 pesos 1987 11.62 9,853 1988 11.00 9,151 1989 11.24 8,347 1990 13.44 9,475 1991 15.69 10,213 1992 17.75 10,156 1993 18.91 10,817 1994 19.96 11,742 1995 24.36 11,584 1999 23.57 11,432 2000 19.43 10,137 2001 17.11 10,340 Note: Fixed charge plus variable consumption, including value added tax. Source: National Institute of Statistics (1987­98), authors' estimations for 1999­2001. $1.51. If the regulated rate-setting process had remained in place, the price today during normal hours would have been $2.40, instead of the current price of about $0.10.24 Since long-distance companies, as well as other telecommunications operators, require access to local networks in order to provide service, it became crucial to regulate fees for access to the public network. In the 1994 rate-setting process, the regulator established the rule that the access charge for incoming and outgoing domestic long-distance calls and outgoing international calls would be 0.63 times the charge for a lo- cal call, which is higher than the real cost of providing the service. Even worse, the per minute access charge for incoming international calls was set at a rate that was 14 times the local rate during normal hours and 84 times the local rate during reduced-rate hours. High access charges to lo- cal networks, coupled with strong competition in the industry on the part of the market leader, Telefónica-CTC, meant that many long- distance operators had serious financial difficulties during the 1994­99 period. The local telephony companies, which were allowed to operate in the long-distance market through subsidiaries, had incentives to charge below cost on long-distance calls, since by lowering rates long- distance traffic would increase and the companies would benefit from the higher revenue arising from access charges to the local network, a reward that the other long-distance companies did not have. In response, the 1999 rate-setting process reduced access rates by an additional 62.7 per- cent on national and international outgoing calls. In the case of incoming THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 235 international traffic, the charge was reduced by 97.5 percent in normal hours and by 99.6 percent in off-peak periods (from the previous high levels). The average reduction in regulated access charges was about 72 percent. Prices of mobile telephony have also declined sharply with increased competition. At the end of 1997 subscribers paid a fixed charge of 15,000 pesos plus 130 pesos per minute for calls made as well as calls received. The entry of ENTEL PCS in March 1998 led to a marketing war among operators that entailed heavy spending on advertising and brought about a significant decline in prices. In early 1998 Telefónica-CTC offered 60 free calling minutes for a fixed monthly charge of 7,080 pesos and billed additional outgoing minutes at 124 pesos during normal hours and 80 pe- sos during reduced-rate hours. In addition, customers who signed a two- year contract received the mobile phone for free. Other plans offered 200 free calling minutes for 16,000 pesos. Clearly, increased competition sig- nificantly reduced rates. The profitability of Telefónica-CTC increased after privatization and remained high until 1997, as shown in table 5.17. Regulators were unsuccessful in passing Telefónica-CTC's efficiency gains on to cus- tomers. This state of affairs changed in 1998, as Telefónica-CTC began to shift investment toward competitive sectors such as mobile telephony and long-distance. The rate-setting process of 1999 that lowered local rates and access charges to the local network also had a large impact on Table 5.17 Profits of Telecommunications Enterprises, 1987­2000 (percent return on equity) Year CTC CNT Telcoy ENTEL Telex BellSouth 1987 11.5 20.1 20.5 56.4 n.a. n.a. 1988 12.7 26.7 23.9 73.6 n.a. n.a. 1989 17.8 18.7 26.2 73.8 57.4 n.a. 1990 12.9 20.2 15.6 52.7 21.9 n.a. 1991 16.2 22.7 16.7 50.5 14.5 n.a. 1992 19.4 29.2 22.8 49.7 28.3 n.a. 1993 23.0 30.2 30.4 37.4 58.9 n.a. 1994 18.7 24.9 32.2 17.2 16.5 0.0 1995 17.3 13.7 29.2 8.4 10.2 70.4 1996 20.9 21.0 37.3 2.4 5.6 250.3 1997 18.7 18.6 39.0 5.1 29.9 1.0 1998 10.8 24.1 47.8 3.8 41.5 62.6 1999 3.8 24.6 36.3 7.0 30.1 3.0 2000 8.5 15.7 20.0 6.3 45.1 1.1 n.a. Not applicable. Source: Authors' figures, based on companies' annual reports. 236 FISCHER, GUTIÉRREZ, AND SERRA Telefónica-CTC. Moreover, Telefónica-CTC suffered from the devalu- ation of the peso (20­30 percent) that began in 1998, since it had not hedged its dollar-denominated debt. Another negative effect was the decline in demand growth and the increase in nonpaying clients attribut- able to the economic slowdown that began in late 1998. Finally, Telefónica- CTC is responsible for the access charges to mobile phone companies of its nonpaying clients, and these access charges are 20 times higher than those that Telefónica-CTC can charge. Telefónica-CTC has implemented a strict cost reduction plan, which in- cludes a drastic reduction in the number of workers. This cutback results from the elimination of inefficiencies in the firm as well as the reduction in planned investment made necessary by the decline in profitability brought on by the new rates and slower economic growth. This increase in effi- ciency allowed the company to achieve modest profits in 2001 after two years of large losses. Since severance payments are large, part of the ex- planation for the losses is the cost of scaling back the level of employment in the company. The two basic regional telephone companies that are dominant in their respective areas (Telcoy and CNT) were less affected by the rate- setting process and have maintained their profit levels. ENTEL is in the opposite situation. While it was a regulated monopoly provider of long- distance services, its rates were much higher than the cost of providing service, which allowed it to have profit levels of above 50 percent for several years. Deregulation led to dramatic falls in long-distance rates and profits (see table 5.17). In 1998 the company reported losses result- ing from the strong competition in long distance, its restructuring costs, and the cost of entry into the mobile telephony market, where it has be- come one of the key players. In 1999 and 2000 the firm showed profits once again, in part because of the asymmetric access rates between the fixed and mobile networks and its successful marketing approach in mobile telephony. In short, the telecommunications sector has been one of the most active in the last few years, and only since 2001 has a slowdown become noticeable. The increased competition in the sector has had a favorable effect on consumers, who are spoiled for choice. Even in local calls, a market that is monopolized in most countries, the market share of Tele- fónica-CTC has declined from 94 percent in 1987 to 76 percent in the year 2001. There are substantial unresolved regulatory problems, how- ever. The most important revolve around the principles that should guide the regulation of access charges. The central issue is that, apart from the direct effect on the profitability of the company, high access rates present a cost on competitors. One dilemma is whether access charges should be based on costs adjusted for the demand facing a company--a principle that Telefónica-CTC claims represents a subsidy to the competition and that the competition deems essential for survival--or if rates should be THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 237 symmetric for identical services. The inclusion of fixed costs is also an is- sue in this regard: should they be included in the cost calculations? Finally, a further issue is whether time-metering of calls or access is ap- propriate when capacity is not a constraint (at least for standard tele- phone calls) due to technological change. The underlying problem is that in the case of access charges, price com- petition can break down. Consider a situation in which users value out- going calls more than incoming calls and Telefónica-CTC's access rates are set low, while its smaller rivals have high regulated rates due to their higher costs. A Telefónica-CTC client would pay the regulated rate for calls within the Telefónica-CTC network, but a higher rate for calls to Telefónica-CTC's rivals. Conversely, a rival's client faces a cheaper rate to call a Telefónica-CTC phone. Since Telefónica-CTC is by far the largest company, with 76 percent of all telephone lines, most of the other providers' calls end up on Telefónica-CTC's network in any case. Since the competitors pay a low access rate for these calls, they might be able to charge a low rate for phone service even though their own networks are more expensive. Thus, the rivals can gain market share at the expense of Telefónica-CTC by having high access charges. Of course, if users also value incoming calls, this incentive to raise access charges is smaller since clients will not appreciate the fact that people are reluctant to call them be- cause it is expensive. Nevertheless, on balance, heavy users of outgoing calls are more attractive to firms than clients who put more weight on incoming calls. At the same time, the last mile is an essential facility in telecommu- nications. Last-mile technology is any telecommunications technology, such as wireless radio, that carries signals from the telecommunications facility along the relatively short distance (hence, the "last mile") to and from the home or business. Cable companies usually have access to the last mile, but other operators (long-distance, mobile telephony, Internet access providers) require access to the local telephone network (or to the cable network) to reach consumers. A wireless fixed system, WLL, has not been as successful as expected. Since Telefónica-CTC faces reg- ulated rates, it has incentives to become a monopoly in the competitive sectors (Beard, Kaserman, and Mayo 2001). It can achieve a monopoly by nonprice discrimination against the other operators. To reduce this risk and preserve competition in the other markets, the regulator may prefer to incur the social cost of having more than one last-mile service provider. Which option is better depends on the extent of economies of phone-line density. Infrastructure Franchises By the early 1990s continuous high growth rates for the preceding years had led to congestion and severe quality problems in highways, seaports, 238 FISCHER, GUTIÉRREZ, AND SERRA and airports. Even though the government had increased the expenditure in infrastructure several times over the minuscule amounts spent during the 1980s, they were insufficient. Therefore, franchising became the hope for rehabilitating and expanding public infrastructure. In 1992 a franchise law was passed allowing the private sector to finance and operate high- ways, airports, and other infrastructure. Franchises have other advantages in addition to solving the problems of governments that do not have the resources (financial, managerial, and supervisory) to provide for infrastructure needs.25 First, when the same firm is in charge of construction and maintenance, it has better in- centives to invest in nonverifiable quality; second, justifying cost-based tolls is politically easier when the project is a private concession; third, the cost of the project is imposed on users and not on the rest of soci- ety; and fourth, there is a built-in screening mechanism against socially wasteful projects, since a project with a negative private return will most likely also have a negative social return. Moreover, when franchise auctions are open and competitive, tolls or user prices should be close to average cost, which is second-best optimal in the presence of economies of scale. From 1994 and 2004, 32 projects were auctioned for a total of about $5 billion; about 18 are already operational. In addition, in 1997 a law was passed allowing franchises of the infrastructure of public ports. Currently, more than 2,000 kilometers of interurban highways together with the main airports and seaports are privately managed. Even though the system has been remarkably successful, several challenges remain. One of these is how to incorporate flexibility in order to react to changed conditions (for instance, unexpected permanent increases in traffic that require widening a road or raising the toll) while at the same time keeping a reputation for not renegotiating contracts when the fran- chise is losing money or for not expropriating money-making fran- chises. Another problem is that most of the profitable private projects have been franchised, and the projects that remain require government subsidies to attract interested bidders. The existence of government sub- sidies, however, negates many of the advantages of infrastructure fran- chises, and the optimal approach to franchising in this case is equivalent to the traditional approach of franchising the building of the road out to the lowest bidder and financing it up-front with public funds (Engel, Fischer, and Galetovic 2002). Highways and Airports The private sector has financed the construction of new highways and airports through build-operate-transfer concessions (table 5.18). More recently the government has extended the range of concession contracts to the building of water reservoirs for irrigation and to penal complexes THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 239 Table 5.18 Concessions in Operation Investment Franchise length Project name Project origin (US$ million) (years) Northern access to Public 214 28 Concepción Access to Santiago's Public 9 12 Airport La Serena Airport Public 4 10 Route 78, Santiago-- Public 172 23.7 San Antonio Road of La Madera Public 31 25 Road Nogales-- Public 12 22 Puchuncaví Road Santiago-- Private 131 28 Los Andes Route 5, Chillán-- Public 192 21 Collipulli Route 5, Los Vilos-- Public 244 25 La Serena Route 5, Santiago-- Public 251 23 Los Vilos Route 5, Talca-- Public 171 12.5 Chillán Route 5, Temuco-- Public 211 25 Río Bueno Carriel Sur Airport, Private 25 16.5 Concepción El Loa Airport, Private 4 12 Calama El Tepual Airport, Public 6 12 Puerto Montt Iquique Airport Public 6 12 El Melón tunnel Public 50 23 Source: Ministerio de Obras Publicas. (appendix tables 5A.9 and 5A.10).26 In general the auction process for concessions has operated as follows. The government sets the minimum technical specifications of the project and grants a concession for 20 or 30 years to the bidder offering to charge the lowest user price for build- ing, operating, and maintaining the project. Bidders must first go through a technical vetting process that qualifies them to make an economic bid. A ceiling and a floor price are imposed. If the ceiling is reached, the bid- ders compete on the minimum subsidy requested. If the floor price is 240 FISCHER, GUTIÉRREZ, AND SERRA reached, the firm that offers the largest payment to the state wins the concession. The first project, a $42 million tunnel, was put out to tender at the end of 1992, was completed on time at very close to the budgeted cost, and was inaugurated in 1995. The most important franchised highway project has been the improvement of the Pan-American Highway, with a total in- vestment estimated at $2.4 billion and total length of 1,511 kilometers. The project was divided into eight segments and put out to tender, with concessions awarded over a two-year period. The final stretch, adjudi- cated in May 1998, runs from Santiago to Talca, with an estimated cost of $750 million. The concession will last 25 years. Starting in 1995 con- tracts for building and operating the cargo and passenger terminals of the eight main airports were awarded in a public bidding process. Airport con- cession-holders have invested about $271 million, of which $200 million were spent in Santiago. Concessions raise important regulatory issues. There are end-point problems, especially regarding maintenance close to the end of the con- cession period. The length of infrastructure concessions and the rigidity of the contract rules pose a major dilemma. For instance, in cases of conges- tion, welfare maximization may require increases in the user fee set in the original contract. The question is how to share the increased income, since the firm bid on a lower price and unless it gets a fraction of the increased revenue, prefers to keep the lower price. However, when bidders expect contracts to be renegotiated, the benefits of competitive bidding are largely lost, since the firm's ability to negotiate, or its lobbying capacity, counts as much as or more than its efficiency (Williamson 1976). Short- ening the concession period is not a solution, since concessionaires would not have enough time to recoup their investments, necessitating govern- ment subsidies. The Melón Tunnel illustrates these problems. Although it had no cost overruns and was built on time, it has not been successful and is unlikely to recoup the original investment (the firm has been incurring annual losses of about $1.5 million). The successful bidder fell victim to the winner's curse (fairly common in a newly developed system such as infrastructure concessions), having offered to pay substantially more than the runner-up. The concessionaire overestimated the demand for the road at the toll ceil- ing, since a significant percentage of drivers choose the old alternative road.27 The winner claims that the lower-than-estimated demand results from the construction of new alternative roads and offers to reduce the toll if the government lowers the annual payment. Such an agreement would almost certainly be socially beneficial in the short run, but the government has refused to renegotiate on the grounds that it would set a dangerous precedent. Franchise-holders have discovered that willingness to pay is less than anticipated when an alternative free route is available, even if an economic THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 241 computation shows that the savings in time and wear and tear on the ve- hicle compensate for the toll. Even in cases in which alternatives are not competitive at all, demand can be highly variable and depend on macro- economic and regional effects. Moreover, the traffic on a specific road depends on the other links in the highway network. Thus, government may affect the demand on a particular route when it alters the rest of the network, and government flexibility in this respect is obviously required.28 The government dealt with this problem by introducing min- imum traffic guarantees, which promise that if traffic flows fall below predetermined levels (usually equivalent to the toll revenues that would pay 70 percent of estimated construction costs), the government will make up the shortfall. Giving guarantees to concession-holders makes it easier for bidders to obtain loans in the financial system, which translates into a larger num- ber of bidders and therefore greater competition. At the same time, state guarantees have some disadvantages. First, they increase the chances that projects that are neither privately nor socially profitable will be under- taken. Private investors might push for higher estimated construction costs so that the guarantee covers all of their actual construction costs, even though they know that actual traffic would probably be much lower (given that the guarantee covers 70 percent of estimated construction costs). Second, it is inconvenient to eliminate all risks from the conces- sion-holder during the highway operation period, because it would mean losing the benefits of private management. Third, guarantees create con- tingent liabilities for the government, but they are seldom valued and are excluded in the year-to-year budget or counted as government debt. However, guarantees carry a seldom-observed political economy advan- tage: if traffic falls far below expectations, the government can always point to the guarantees as a way of reducing the pressure to renegotiate the franchise contract. To avoid some of the problems associated with standard auctions, the government has been experimenting with a new mechanism advocated by Engel, Fischer, and Galetovic (forthcoming) for auctioning infrastructure concessions.29 In their proposal, the regulator sets the maximum toll that the concession-holder can charge and then awards the concession to the firm demanding the least present value of revenue (PVR) for building and then operating the highway until the required revenue is collected through toll payments. Hence, the duration of the concession is endogenous. This auction mechanism reduces the risk faced by the franchise-holder, because the present value of the total income the concession-holder will receive is known in advance. An additional advantage of PVR auctions is that they are inherently flexible. Early termination of a concession is not a problem, if required, for instance, to widen the highway. If the government compensates the operator with the amount remaining to be collected minus estimated 242 FISCHER, GUTIÉRREZ, AND SERRA savings on maintenance and operation costs, this is a fair compensation to the franchise-holder. Hence, PVR greatly reduces the scope for dis- agreement. The authority could also adjust the toll charged by the fran- chise-holder to more closely approximate the optimal toll given the level of congestion.30 The highway linking Santiago and Valparaíso was auctioned using the PVR method. In February 1998, a Spanish consortium won the con- cession. It sought a present value revenue of UF (Unidades de Fomento) 11,938,207, or approximately $400 million, an amount it expects to collect in 15 years. The price-cap for the toll is 1,800 Chilean pesos (about $3). In this instance, the rules required that bidders seeking a minimum guaranteed income would have had to pay the government for this guarantee. Remarkably, two out of four bidders, including the winner, did not seek the guarantee. Thus, in principle, the state did not assume any risk. Starting in 1999 the government has awarded four urban highway concessions in Santiago that represent a total investment of about $1.4 billion (two franchises for the improvement of a road that rings Santi- ago and two that cross Santiago in the north-south and east-west direc- tions). Auctioning urban highways has proven more difficult than expected. First, in urban highways the range of government decisions influencing traffic is much broader than in interurban highways. The construction of access roads, the use of complementary or substitute routes, the expansion of the subway system, or the introduction of tolls on congested streets can affect traffic patterns. Moreover, the construc- tion of highways generates urban problems. The construction of large- capacity urban expressways can cause the deterioration of the sur- rounding area. In Santiago people living in a well-to-do residential area adjacent to a proposed highway mounted a strong campaign against its construction. Although they could not prevent its construction, they forced major changes that increased the cost substantially. Ecologists have opposed urban highways because they believe highways will en- courage car use and thereby increase pollution; instead they favor in- vesting in public transport. While the argument is correct when the highways are free, it no longer holds when these highways have tolls that depend on the level of congestion. Concessions for Port Management and Operation There are 10 state-owned ports and 22 private ports in Chile. The state- owned seaports have natural advantages because of their better geograph- ical locations. In general the private ports are used for bulk cargo, so they need less infrastructure than the state-owned ports, which tend to special- ize in general cargo (normally in containers). This type of cargo requires THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 243 calm waters for loading and unloading, so extensive works usually protect the state-owned ports. Private participation in state-owned ports started in 1981 when pri- vate firms were allowed to perform the duties of loading and unloading ships at the docks, as well as on-port storage services. This change greatly increased efficiency in cargo handling, making it unnecessary to invest in expanding these ports, even though previously, under public operation, the ports were always congested. The port authority (Empresa Portuaria de Chile, or Emporchi), however, retained the management of all state- owned port infrastructure, that is to say, docking sites and storage facili- ties (Foxley and Mardones 2000). In the mid-1990s, it became evident that rapid growth in foreign trade would, in the short term, render inadequate the cargo transfer ca- pacity of state-owned ports. This was particularly true of ports located in the central zone of the country where, for geographic reasons, there is little potential for development of new ports. Chile has only a few well-protected bays and inlets, and most of these lie in the middle of ur- ban centers. Expanding the number of docking sites at existing ports is possible, at least in the case of the port of San Antonio, but costly. In addition to these stumbling blocks, a dearth of stacking and storage space at the ports further compounds problems, since urban growth and sprawl have severely limited the ability of these port service areas to ex- pand.31 Finally, having multiple private operators conspired against the coordination of activities and investment in specialized gantry cranes for containers. The government feared that inefficient port operations would have a multiplier effect on the costs of the transportation chain. Ships range from large, fast, and expensive types to slower, more sea-worn vessels. Since an efficient ship costs tens of millions of dollars, from the point of view of a shipping company the main cost of an inefficient port is not the fees for docking, loading, and unloading but rather the capital cost of the ship. In- efficient ports tend to be able to receive only slower, older, and smaller ships with higher operational costs. Hence, even in addition to having high docking and loading costs, inefficient ports raise the total transport cost of traded goods by much more and render a country uncompetitive in the in- ternational market for its goods. This was one of the fears of the Chilean government, since Chile is an open economy that depends on remaining competitive for its future growth. The government believed that it was possible to expand the transfer capacity of the state-owned ports by increasing private participation in port administration and operation. Moreover, the government began to think of ports as consisting of terminals, known as frentes de atraque, which combine groups of docking sites and storage space that could func- tion as independent units. By coordinating activities and internalizing 244 FISCHER, GUTIÉRREZ, AND SERRA all the benefits of investment in new equipment, a single operator would be the best way of operating a terminal. Based on this assessment, a bill was introduced in Congress for modernization of the state-owned port sector, which was enacted in December 1997. The law split Emporchi into 10 different SOEs or port authorities, one for each state-owned port, which were granted the power to award concessions to multiple or single private companies for the administration and operation of port infrastructure. Granting concessions for state-owned port administration and opera- tions posed certain risks to the sector's competitiveness. There are only three ports in the central zone of Chile (Region V), which is where most of the general cargo enters and leaves the country. Two of these ports are state-owned (Valparaíso and San Antonio), while one (Ventanas) is pri- vately owned. Altogether, these three ports are endowed with seven frentes de atraque, but not all of these are able to berth large vessels. Additionally, it is necessary to consider that some terminals are especially built for trans- fer of containers, others for bulk cargo (where there is no lack of compe- tition), and still others, for standard cargo. In other countries, large-scale port users, mainly shipping compa- nies, own their own cargo terminals, because such an arrangement pro- vides operational advantages. In Chile, however, so few frentes de atraque are available that only a small number of users of significant size would be able to own their terminal. This would, of course, place other users at a great disadvantage. Even though regulations make it mandatory for prices to be made public and set on a nondiscriminatory basis, concessionaires can use subtle ways of discriminating against nonintegrated shippers that are difficult to prove and therefore to pe- nalize. These methods include assigning the choice spaces in the hold- ing areas to one company over another, providing one company with better service than others, using insider information, and manipulating docking reservations. In drafting the port modernization law (Law 19.542), legislators took these problems into account and included several clauses designed to safe- guard competition in the sector. First, the law requires that concessions be awarded through public auctions and for no more than 30 years. Second, concessionaires must be incorporated as publicly owned companies that are engaged in a single line of business. Third, the rates set by concession- aires must be made public and established on a nondiscriminatory basis. Fourth, proposed by-laws and internal regulations for concessions are re- quired as an integral part of the rules of bidding. These rules must conform to objective technical and nondiscriminatory standards, especially with re- gard to assignment of spaces and reserve capacity. The two port authorities in Region V put up for simultaneous pub- lic bidding three out of the six docking areas they owned. Two of these were the frentes de atraque capable of berthing the largest vessels at THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 245 each port, while the third was the bulk terminal of San Antonio. The port authorities, in consultation with the Antitrust Prevention Com- mission, imposed additional conditions on concession-holders to pre- vent risks of abuse of a dominant position, as provided for by the law. Their conditions included ceilings on horizontal integration, restric- tions on vertical integration, additional rules of transparency, reserving the right to set maximum prices to prevent low bidder turnout, and quality standards. The rules specified, for instance, that significant users (defined as those that shipped more than 15 percent of the cargo in the region) should not own more than 40 percent of the stock or voting shares in the firms that operated the port franchises. According to the port authorities, it was nec- essary to limit significant users to a minority position in the company in or- der to reduce the possibility of discrimination. Concessionaires are required to grant any interested party expeditious access to information such as cargo contracts, service priorities, and types of cargo and consignees, so that all of the interested parties would have the same information. Finally, the port authorities can impose penalties for low quality of service. Mini- mum transfer speeds and maximum waiting times for ships are specified in the concession contracts. The concessions were awarded in July 1999. In principle, each one was to be awarded to the bidder that offered the lowest maximum port transfer rate index, which was an average of four transfer charges. Nonetheless, in fairness to private port competitors, the rules of bidding for each docking front specified a minimum rate floor index. Moreover, the minimum rate floor has the beneficial effect of creating ex-post rents for the nonintegrated port, which implies that the incentives for under- hand integration with a shipper and then discriminating against its com- petition are reduced. In the event that more than one bidder offered the minimum rate index established in the rules, a tie-breaking payment was to be offered in addition.32 This payment was over and above the leasing payment that was established in the rules of bidding for the port infrastructure and was calculated on the basis of the economic value of the property. The bidding attracted a great deal of interest, and a total of 21 bids were tendered by consortia made up of leading domestic and foreign com- panies, of which 19 included the minimum rate index, plus the additional tie-breaking payment. All terminals were awarded in the end on the basis of the tie-breaking payment amount. Consequently, the average rates for port services were reduced by more than 10 percent in the frentes de atraque that were awarded in concession, and the government was also able to take in revenues of $267 million, tripling its expectations (Foxley and Mardones 2000). The results of the first years of operation have also satisfied the government's expectations, as can be seen with data for the Port of 246 FISCHER, GUTIÉRREZ, AND SERRA Table 5.19 Valparaíso: Time Spent Loading and Unloading and Transfer Speed Productivity measure 1999 2001 2002 (est.) Loading, unloading (hours) 45.0 26.3 21.0 Transfer speed (containers per hour) 25.5 43.7 54.8 Note: Loading and unloading time is for a Eurosal vessel with 1,150 cargo movements. Source: Empresa Puerto Valparaíso. Valparaíso (table 5.19). The efficiency in port services increased sub- stantially. Similarly, the transfer speed at the port of Iquique increased by 41 percent in just half a year.33 Finally, at the franchised terminal at San Antonio, the main port, the transfer speed rose from 475 tons an hour to 635 tons an hour, an increase of 34 percent.34 At the Valparaíso concession, investment in new cranes, computer software, and other equipment during 2001 topped $8 million, with another $27.5 million expected by 2006. Conclusions This section collects our results and presents some hypotheses that offer consistent explanations for these results. Unfortunately, the lack of infor- mation limits the verifiability of these hypotheses. First, we find that pri- vatized firms experienced significant improvements in efficiency, but this improvement is no different from the change experienced by other private firms in their respective economic sectors. This allows us to conclude that Chilean SOEs were efficient before privatization, at least if we compare them with private firms in their respective sectors. Hachette and Lüders (1994) have noted this previously. This conclusion is consistent with the fact that employment levels in privatized firms were stable for several years before privatization and rose afterward. This is not surprising given that many years before privatization, state-owned enterprises had under- gone reorganizations that were especially geared toward reducing the number of workers (see table 5.7). Second, we find that there are significant differences in the postprivati- zation performance of regulated and unregulated firms. Hence, we report separate results for each group. We give an account of adjusted results, that is, where we have normalized the performance of firms with respect to the performance of their economic sectors at the 2-digit SIC levels. We focus first on the behavior of firms that were not regulated. Results for this group show no major changes in efficiency measured as unit costs and sales over physical assets after privatization. Since these firms operated in THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 247 competitive sectors and their efficiency did not grow compared with other firms in their sectors, adjusted profitability should not show major changes after privatization, as is the case. The postprivatization performance of regulated firms is quite different. The profitability of regulated firms grew after privatization. In fact, the ratio of net income over PPE rose substantially, while the ratio of income to sales also increased, though at a more modest rate. These firms enjoyed efficiency gains after privatization, but those gains are not statistically sig- nificant. Similarly, the cost per unit indicator shows a slight decrease, while the ratio of sales to PPE shows a modest increase. These results are consis- tent with efficiency gains resulting primarily from a more efficient use of capital (and probably a minor increase in regulated prices). There is some evidence that before privatization, regulated SOEs had overinvested in physical assets. This implies that privatization should result in higher prof- its rates, as observed. The implication is that Chile's approach to incentive regulation has paid off, promoting efficiency in regulated firms. As shown, the efficiency gains in the regulated sectors that were privatized do not lag behind those of the unregulated privatized sectors. However, regulators have been un- able to transfer all these efficiency gains to consumers. This should not come as a complete surprise. It is well known that regulation is an imper- fect substitute for competition, and Chile is no exception. Moreover, the ability of regulators to transfer gains to consumers has recently improved substantially. Certain aspects of the Chilean regulatory legislation and practice should be improved. In particular, the transparency of the rate-setting process should increase. Currently, regulators in the electricity and telecommunications sectors can exchange the information used to set rates only with the regulated companies; this prevents consumer organizations from countering the lobbying pressures of the regulated enterprises. The recent law regarding the water and waste treatment sector takes the op- posite approach: all the information used in setting rates must be made public. However, it is not clear that this new law has been effective in restraining lobbying while at the same time limiting the possibility of reg- ulatory takings. The regulatory process requires improvement in access to information on the regulated firm. It also necessitates modeling an efficient firm, but this in turn requires information that is uniquely available to the real firm, since costs depend, among other factors, on topography, geographic density of customers, and demand. Regulators have encountered major problems in gaining access to company data, because legislation does not provide spe- cific penalties for failure to deliver information or for submitting false information. Currently, when a company refuses to hand over information, the regulator must go to the courts, where the process is lengthy and penal- ties are low. 248 FISCHER, GUTIÉRREZ, AND SERRA Another lesson that may be learned from the Chilean experience is the importance of properly regulating essential facilities. A 1982 law, for ex- ample, required the dominant local telephone operators to provide in- terconnection access for other operators requesting it, with the cost of access to be negotiated by the parties. However, the negotiation of these charges led to prolonged lawsuits that made it difficult for new compa- nies to enter the market. A 1994 law solved the problem by regulating all interconnection charges. Similarly, Chilean legislation guarantees power- generating firms access to the transmission system, but the fact that the largest power-generating company owned the transmission system, com- bined with the fact that transmission tolls are negotiated, created some problems. In June 1997 the Antitrust Commission ruled that the power- generating company should divest its transmission assets to an indepen- dent company. Other types of natural monopolies were also privatized: those related to infrastructure. In those cases, the rate-setting problem was solved by competition for the field (Demsetz 1968), auctioning franchises to the firms that asked for the lowest user fee. It has been a successful system. The main highways have been completely overhauled and their capacity has increased substantially, reducing internal transportation costs and making the country as a whole more efficient. Only a few potential prob- lems remain. First, the traffic guarantees offered by the government to successful bidders created unaccounted liabilities; and, second, fran- chise-holders might be successful in lobbying the government for changes in the terms of their contracts. In fact, many contracts have al- ready been renegotiated because the highway projects were awarded through agreements that omitted important details and thus had to be re- vised. This has meant a substantial (but not overwhelming) increase in the cost of the projects. At the same time, the government was able to re- main firm when Tribasa (a company that had received three important highway concessions) failed to complete one of its projects on time, and this is an encouraging sign. Seaport franchises have also been successful so far: investment has in- creased, port efficiency is higher, and ships require much shorter periods for loading and unloading. There have been no complaints from ship- ping companies that they are discriminated against in the franchised ports, so it appears that the horizontal and vertical integration restric- tions on the port operators have served their purpose. Despite these fa- vorable results, however, it is too soon to have a fair evaluation of the port franchises. To summarize, privatization has benefited the country, even in the case of the regulated sectors. Market imperfections mean that it is not always easy to align the interests of private providers in regulated sectors and those of society as a whole. However, regulation has partially succeeded in this goal, in part because of ongoing fine-tuning. Appendix 5A Table 5A.1 Changes in Profitability of Privatized Firms t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Operating 37 0.1182 0.2051 3.6298* 2.5459* 3.6213* 7.6211* income/sales 0.0835 0.1512 0.0000* 5,12E­09* Operating 37 0.0969 0.2000 2.5620* 2.9784* 3.7906* 4.7129* income/PPE 0.0624 0.1269 0.0000* 5,12E­09* Net income/sales 37 0.0172 0.1303 3.2708* 3.2811* 0.6074 3.8679* 0.0192 0.1459 0.1620 5,42E­07* Net income/PPE 37 0.0386 0.1636 3.3633* 3.5189* 2.8923* 4.7555* 0.0214 0.0968 0.0494* 6,17E­08* Unregulated Operating 18 0.1113 0.1797 1.2527 0.9808 1.8682 3.8725* income/sales 0.0763 0.1294 0.0038* 0.0006* Operating 18 0.1206 0.2139 0.9345 0.7277 2.4296* 2.5125* income/PPE 0.0711 0.0788 0.0038* 0.0006* Net income/sales 18 0.0331 0.0461 0.9706 1.0124 0.7044 0.7687 0.0299 0.0456 0.2403 0.0154* Net income/PPE 18 0.0510 0.1273 0.9072 0.5379 2.1953* 2.0011 249 0.0402 0.0476 0.0481* 0.0037* (Table continues on the following page.) 250 Table 5A.1 (continued) t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 Regulated Operating 19 0.1247 0.2291 4.7276* 2.8465* 4.0062* 7.9657* income/sales 0.0872 0.1726 0.0004* 1,91E­06* Operating 19 0.0744 0.1868 5.8026* 3.6347* 4.4425* 8.2944* income/PPE 0.0549 0.1649 0.0004* 1,91E­06* Net income/sales 19 0.0649 0.2101 5.1091* 3.3720* 2.1687* 9.6619* 0.0143 0.2030 0.3238 1,91E­06 Net income/PPE 19 0.0268 0.1981 4.5417* 4.2478* 1.9231 6.7673* 0.0019 0.1607 0.3238 1,91E­06* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. Source: Authors' calculations. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 251 Figure 5A.1 Cost per Unit before and after Privatization, Adjusted and Unadjusted a. Cost per unit (mean) b. Cost per unit (median) 0.74 0.82 0.72 0.80 0.78 0.70 0.76 0.68 0.74 0.66 0.72 0.70 0.64 0.68 0.62 0.66 0.60 0.64 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t­5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Cost per unit, adjusted (mean) d. Cost per unit, adjusted (median) 0.08 0.14 0.06 0.12 0.10 0.04 0.08 0.02 0.06 0.00 0.04 ­0.02 [t-5: t-3][t-3: t-1][t+1: t+3][t+3: t+5] 0.02 0.00 ­0.04 ­0.02 ­0.06 [t-5: t-3][t-3: t-1][t+1: t+3][t+3: t+5] ­0.04 ­0.08 ­0.06 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. 252 Table 5A.2 Changes in Operating Efficiency of Privatized Firms t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Cost per unit 37 0.7049 0.6606 1.8235 1.2595 22.8569* 21.9694* 0.7769 0.7149 0.0000* 0.0000* Sales/PPE 37 1.1686 1.1570 0.2074 1.1838 4.3087* 7.2498* 0.5787 0.7487 0.0000* 0.0000* Sales/employee 30 69284.4870 102795.0097 3.9218* 2.1290* 5.4755* 6.5359* 42013.6095 82924.2615 0.0000* 0.0000* Operating income/ 30 14705.0275 25088.0736 2.8042* 2.7795* 2.9264* 4.2550* employee 4087.5993 13788.9867 0.0000* 0.0000* Unregulated Cost per unit 18 0.6979 0.6747 0.4239 0.3797 17.4061* 14.3750* 0.7234 0.6978 0.0000* 0.0000* Sales/PPE 18 1.7011 1.0906 1.0359 5.7266* 3.2401* 4.2907* 0.6083 0.6532 0.0000* 0.0000* Sales/employee 12 1.07647.9371 134806.9036 1.3016 0.4041 4.0110* 3.7855* 85164.2136 107110.7778 0.0002* 0.0032* Operating income/ 12 29483.4651 39012.3794 0.8572 0.1732 2.5880* 2.8487* employee 6807.3028 22053.4960 0.0002* 0.0193* Regulated Cost per unit 19 0.7116 0.6473 2.7041* 1.4159 14.9651* 16.5470* 0.8083 0.7257 0.0000* 0.0000* Sales/PPE 19 0.6641 1.1394 1.6805 1.8539 6.3889* 6.1465* 0.5589 0.8323 0.0000* 0.0000* Sales/employee 18 43708.8536 81453.7471 4.0624* 2.9108* 6.2884* 8.6217* 34940.3242 82345.8544 0.0000* 0.0000* Operating income/ 18 4852.7357 15805.2031 5.8470* 3.8915* 3.9250* 6.8893* employee 3448.3215 13788.9867 0.0007* 0.0000* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. Source: Authors' calculations. 253 254 FISCHER, GUTIÉRREZ, AND SERRA Figure 5A.2 Investment as a Fraction of Physical Assets (PPE) before and after Privatization, Adjusted and Unadjusted a. Investment over PPE (mean) b. Investment over PPE (median) 1.20 1.20 1.00 1.00 0.80 0.80 0.60 0.60 0.40 0.40 0.20 0.20 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Investment over PPE, adjusted (mean) d. Investment over PPE, adjusted (median) 0.50 0.50 0.40 0.40 0.30 0.30 0.20 0.20 0.10 0.10 0.00 0.00 ­0.10 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] ­0.10 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] ­0.20 ­0.20 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: PPE property, plant, and equipment; t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. Table 5A.3 Changes in Investment and Assets in Privatized Firms t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Log(PPE) 36 16.4534 16.7571 1.1599 0.7658 41.4029* 38.2008* 16.3349 16.6948 0.0000* 0.0000* Investment/sales 28 0.1612 0.1074 0.6339 0.5899 2.6560* 2.1901* 0.0977 0.0265 0.1725 0.0436* Investment/ 24 4885.4701 18064.6755 1.2613 0.3093 0.9709 2.0507 employees 3106.3502 2235.6167 0.1537 0.0758 Investment/PPE 28 0.0537 0.0322 0.4064 0.0655 1.3310 1.6093 0.0443 0.0448 0.1725 0.0436* PPE/employees 30 233653.7571 330056.5131 2.3351* 0.7097 2.2491* 2.5458* 60346.6975 70591.5865 0.0000* 0.0000* Unregulated Log(PPE) 17 16.0596 16.2138 0.2260 0.3272 21.9867* 19.9756* 16.0273 16.5830 0.0000* 0.0000* Investment/sales 12 0.1771 0.0930 0.3470 0.4041 1.3215 0.9619 0.0302 0.0159 0.3872 0.6128 Investment/ 6 6420.8692 54646.7804 1.0025 0.9608 0.4668 2.3891* employees 10070.6767 65872.1966 0.3438 0.3438 Investment/PPE 12 0.0493 0.0049 0.3395 0.1155 0.5256 0.1229 255 0.0250 0.0268 0.3872 0.6128 (Table continues on the following page.) 256 Table 5A.3 (continued) t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 PPE/employees 12 470839.7510 670100.9661 1.6964 0.7506 1.8877 2.2080* 173215.2970 283717.7356 0.0002* 0.0002* Regulated Log(PPE) 19 16.8058 17.2432 3.5192* -0.8321 44.4567* 43.4982* 16.5665 17.1940 0.0000* 0.0000* Investment/sales 16 0.1492 0.1182 0.4954 0.9422 3.6193* 2.4036* 0.1132 0.0265 0.0384* 0.0106* Investment/ 15 4100.1476 8183.0048 0.3273 0.0622 3.0199* 2.1897* employees 3242.7176 2307.4099 0.0592 0.0176* Investment/PPE 16 0.0570 0.0600 0.1086 0.1131 3.4090* 3.7712* 0.0494 0.0448 0.0384* 0.0106* PPE/employees 18 75529.7611 103360.2111 0.6359 0.8542 5.7723* 4.6236* 57560.8634 70591.5865 0.0000* 0.0000* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. Source: Authors' calculations. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 257 Figure 5A.3 Investment as a Fraction of Sales before and after Privatization, Adjusted and Unadjusted a. Investment over sales (mean) b. Investment over sales (median) 2.50 2.50 2.00 2.00 1.50 1.50 1.00 1.00 0.50 0.50 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Investment over sales, adjusted d. Investment over sales, (mean) adjusted (median) 0.60 0.40 0.60 0.20 0.40 0.00 0.20 ­0.20 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] 0.00 ­0.40 ­0.20 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] ­0.60 ­0.40 ­0.80 ­0.60 ­1.00 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. 258 Table 5A.4 Changes in Profitability of Privatized Firms, Adjusted t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Operating 36 0.0570 0.0128 2.0203 1.1374 2.0925* 0.5885 income/sales 0.0576 0.0313 0.0005* 0.1214 Operating 36 0.0223 0.1093 2.1198* 1.9934* 1.1384 2.6297* income/PPE 0.0030 0.0296 0.2025 0.0662 Net income/sales 36 0.0621 0.0355 0.8480 0.3941 2.6489* 1.3586 0.0405 0.0328 0.0326* 0.0326* Net income/PPE 36 0.0149 0.0964 2.2858* 1.6555 1.0682 3.0114* 0.0110 0.0141 0.3088 0.1214 Unregulated Operating 18 0.0170 0.0301 1.0079 0.4113 0.3470 0.8091 income/sales 0.0072 0.0005 0.1189 0.5927 Operating 18 0.0408 0.1410 1.0144 1.0124 1.1243 1.7721 income/PPE 0.0030 0.0094 0.4072 0.2403 Net income/sales 18 0.0879 0.0487 0.5670 0.1898 2.0494 1.0150 0.0150 0.0242 0.2403 0.1189 Net income/PPE 18 0.0211 0.0804 0.7612 0.2214 0.8746 1.4181 0.0150 2.3359E­05 0.4072 0.5927 Regulated Operating 18 0.0970 0.0559 1.8235 1.4553 4.4782* 2.8959* income/sales 0.1085 0.0714 0.0006* 0.0481* Operating 18 0.0037 0.0776 3.4631* 1.9299 0.2543 3.0121* income/PPE 0.0031 0.0756 0.2403 0.1189 Net income/sales 18 0.0363 0.0222 0.4705 0.2847 1.9511 1.0098 0.0442 0.0368 0.0481* 0.1189 Net income/PPE 18 0.0087 0.1124 2.9345* 1.9932* 0.5940 3.6010* 0.0140 0.0783 0.1189 0.0481* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. Source: Authors' calculations. 259 260 FISCHER, GUTIÉRREZ, AND SERRA Figure 5A.4 Net Income as a Fraction of Physical Assets before and after Privatization, Adjusted and Unadjusted a. Net income over PPE (mean) b. Net income over PPE (median) 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Net income over PPE, adjusted d. Net income over PPE, adjusted (mean) (median) 0.12 0.14 0.10 0.12 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 -0.02 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] -0.02 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] -0.04 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: PPE property, plant, and equipment; t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. Table 5A.5 Changes in Operating Efficiency of Privatized Firms, Adjusted t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Cost per unit 36 0.0017 0.0055 0.2858 0.1126 0.0679 0.2343 0.0275 0.0322 0.4340 0.1215 Sales/PPE 36 0.6096 0.6121 0.0093 1.0812 2.3561* 4.0088* 0.1922 0.2864 0.3089 0.0662 Unregulated Cost per unit 18 0.0652 0.0419 0.4396 0.6328 1.7328 1.0239 0.0441 0.0093 0.1189 0.2403 Sales/PPE 18 0.8943 0.4995 0.6096 0.3797 1.7690 2.0071 0.0600 0.1074 0.4073 0.4073 Regulated Cost per unit 18 0.0617 0.0529 0.3107 0.8226 2.2441* 2.8640* 0.1164 0.0548 0.0481* 0.0038* Sales/PPE 18 0.3248 0.7248 1.7461 1.4237 3.1721* 4.0149* 0.2820 0.4787 0.1189 0.0481* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. 261 Source: Authors' calculations. 262 FISCHER, GUTIÉRREZ, AND SERRA Figure 5A.5 Operating Income as a Fraction of Physical Assets before and after Privatization, Adjusted and Unadjusted a. Operating income over PPE (mean) b. Operating income over PPE (median) 0.25 0.20 0.18 0.20 0.16 0.14 0.15 0.12 0.10 0.10 0.08 0.06 0.05 0.04 0.02 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Operating income over PPE, adjusted d. Operating income over PPE, adjusted (mean) (median) 0.16 0.12 0.14 0.10 0.12 0.08 0.10 0.06 0.08 0.06 0.04 0.04 0.02 0.02 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] -0.02 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: PPE property, plant, and equipment; t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. Table 5A.6 Changes in Investment and Assets in Privatized Firms, Adjusted t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Variable N Median before Median after mean median median before 0 median after 0 All firms Log(PPE) 35 0.7909 0.7846 0.4744 0.3230 39.4821* 39.0845* 0.7845 0.7779 0.0000* 0.0000* Investment/sales 28 1.8959 1.5611 2.0772* 2.4089* 11.7751* 14.2995* 2.0177 1.6611 0.0000* 0.0000* Investment/PPE 28 0.6250 0.6034 0.2502 0.8193 11.1709* 13.1927* 0.5748 0.5465 0.0000* 0.0000* Unregulated Log(PPE) 17 0.7785 0.7628 0.4644 0.4650 23.5227* 22.6188* 0.7845 0.7779 0.0000* 0.0000* Investment/sales 12 1.4619 1.3864 0.1834 0.6351 6.1159* 7.0902* 1.6319 1.3423 0.0032* 0.0002* Investment/PPE 12 0.6013 0.6354 0.1415 0.3464 5.4303* 6.0524* 0.5748 0.5435 0.0032* 0.0002* Regulated Log(PPE) 18 0.8026 0.8052 0.3869 0.1898 33.6094* 35.8317* 0.7866 0.7699 0.0000* 0.0000* Investment/sales 16 2.2214 1.6922 2.8174* 2.9397* 12.0912* 14.3819* 2.2513 1.6893 0.0000* 0.0000* Investment/PPE 16 0.6428 0.5793 0.8667 0.6030 11.6052 27.7174* 0.5759 0.5501 0.0000* 0.0000* 263 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number and PPE property, plant, and equipment. Source: Authors' calculations. 264 FISCHER, GUTIÉRREZ, AND SERRA Figure 5A.6 Operating Income as a Fraction of Sales before and after Privatization, Adjusted and Unadjusted a. Operating income over sales (mean) b. Operating income over sales (median) 0.25 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms c. Operating income over sales, adjusted d. Operating income over PPE, adjusted (mean) (median) 0.04 0.02 0.02 0.00 0.00 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] -0.02 [t-5: t-3] [t-3: t-1] [t+1: t+3] [t+3: t+5] -0.02 -0.04 -0.04 -0.06 -0.06 -0.08 -0.10 -0.08 -0.12 -0.10 -0.14 -0.12 All firms Unregulated firms All firms Unregulated firms Regulated firms Regulated firms Note: t year of privatization; the number represents the number of years before or after privatization. Source: Authors' calculations. Table 5A.7 Employment in Privatized Firms t-statistic for Z-statistic for Test for Test for Mean before Mean after change in change in mean before 0 mean after 0 Type of firm N Median before Median after mean median median before 0 median after 0 All firms 32 1193 1381 1.3374 0.1880 3.8391* 3.4071* 380 360 0.0000* 0.0000* Unregulated 15 1557 1764 0.6502 0.1867 3.1965* 2.7635* 754 874 0.0000* 0.0000* Regulated 17 871 1044 1.2049 0.3272 2.2127* 2.0096 179 236 0.0001* 0.0001* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number. Source: Authors' calculations. 265 266 Table 5A.8 Physical Productivity before and after Privatization Percent Firm Before After variation Unit of measurement Unregulated CAP 0.0718 0.0629 0.1233 Tons of steel forged (MT/worker) Colbún 17.0733 Electroandina 12.4271 ENDESA 2.6206 3.6677 0.3995 Energy generated (GWh/workers) Fepasa 0.4554 0.4123 0.0947 (Millions of tons km/workers) Ferronor 0.2891 0.5084 0.7584 (Millions of tons km/workers) Gener 5.7859 4.7388 0.1810 Energy generated (GWh/workers) Laboratorios Chile 1394.5421 LAN Chile 1.7089 1.0008 0.4144 (Passengers per kilometer/workers) Pilmaiquén 7.1905 8.6277 0.1999 Energy generated (GWh/workers) Telex 39.0387 Regulated Mean 0.0778 Median 0.0947 Standard deviation 0.4002 Number 7 Chilectra 1.3412 1.7036 0.2702 Energy purchased (GWh/workers) Chilquinta 0.8038 1.6152 1.0093 Energy purchased (GWh/workers) CTC 69.3534 151.1731 1.1797 Number of lines in operation (units/ workers) Edelaysén 0.3208 Edelmag 0.6296 0.8611 0.3676 Energy purchased (GWh/workers) Elecda 0.9252 1.4479 0.5649 Energy purchased (GWh/workers) Eliqsa 1.6372 2.0288 0.2391 Energy purchased (GWh/workers) Emec 1.7957 1.4135 0.2128 Energy purchased (GWh/workers) Emel 1.2992 0.8345 0.3577 Energy purchased (GWh/workers) Emelari 1.0657 1.6876 0.5836 Energy purchased (GWh/workers) Emelat 1.5176 1.5749 0.0377 Energy purchased (GWh/workers) Emos 0.1911 IANSA 278.0696 266.4219 0.0419 Sugar produced per worker (metric tons/ workers) Saesa 0.6055 0.9196 0.5188 Energy generated (GWh/workers) Telefónica del Sur 39.0387 Mean 0.3466 Median 0.3189 Standard deviation 0.4612 Number 12 All firms Mean 0.2475 Median 0.2391 Standard deviation 0.4485 Number 19 Note: MT metric tons, GWh gigawatt hours, and km kilometers. Source: Authors' calculations. 267 268 Table 5A.9 Concession Projects under Construction Bidder's budgeted Percent built (as of Project expense Franchise Project name December 2001) origin (US$ million) length (years) Cerro Moreno Airport, Antofagasta 100 Private 8 10 International Airport, Santiago 100 Public 170 15 Punta Arenas Airport 98.95 Public 10 9 Santiago-Valparaíso--Viña del Mar highway 77.96 Public 383 25 Litoral Central Road 0 Public 67 30 Route 5, Collipulli--Temuco 91.79 Public 256 25 Route 5, Río Bueno--Puerto Montt 99.3 Public 236 25 Route 5, Santiago--Talca and southern access 33.78 Public 698 25 to Santiago North--South urban highway, Santiago 1.08 Public 517 30 Costanera Norte urban highway 33.4 Public 405 30 El Bato reservoir, Illapel Auctioned 2001 NA 37 25 Alternate Melipilla road Auctioned NA 19 25 Américo Vespucio South urban highway Auctioned 2001 NA 28 30 Américo Vespucio North urban highway, Auctioned 2001 NA 250 30 Santiago International road 60 Auctioned January 2002 NA 165 30 Route Talcahuano--Penco Auctioned 2001 NA 19 25 Group 1 jails (Iquique-La Serena-Rancagua) Auctioned 2001 75 15 to 20 Source: Ministerio de Obras Publicas. Table 5A.10 Projects in Concession Process Status December Project cost Franchise Project name 2001 (US$ million) length (years) Auction date Bidding date New regional airport, Atacama Call for bids 25 20 October 2001 April 2002 Jails, Group 2 (Concepción, Valdivia) Call for bids 50 22 October 2001 May 2002 Commuter Rail Melipilla--Santiago-- Call for bids 300 18 July 2002 Batuco Chiloé bridge Call for bids 350 30a August 2002 Airport, Arica NYA 10 10 June 2002 September 2002 Jails, Group 3 (Santiago1, 2, region V, NYA 80 20 June 2002 October 2002 interior) International airport, IV region NYA 45 20 November 2002 Intermediate Tech., Recoleta-- NYA 171 20 June 2002 December 2002 Independencia Ecological complex, Santiago NYA 50 30 2nd half, 2002 North-West access Santiago NYA 160 30 1st half, 2002 2nd half, 2002 Exchange stations, Quinta Normal, NYA 60 30 2nd half, 2002 1st half, 2003 Gran Avenida Land Port, Los Andes NYA 16 25 2nd half, 2002 1st half, 2003 Intermediate highway ring, El Salto-- NYA 32 30 2nd half, 2002 1st half, 2003 Kennedy Convento Viejo Reservoir NYA 210 20 to 25 2nd half, 2002 1st half, 2003 Improvement, Route 5: La Serena-- NYA 105 17a 2nd half, 2002 1st half, 2003 Caldera 269 (Table continues on the following page.) 270 Table 5A.10 (continued) Status December Project cost Franchise Project name 2001 (US$ million) length (years) Auction date Bidding date Exchange Stations Pajaritos, Santos NYA 60 -- 1st half, 2003 2nd half, 2003 Dumont Jails, Group 4 (Santiago 2, region V) NYA 50 15 to 20 1st half, 2003 2nd half, 2003 New airport, Region IX NYA 48 20 to 25 1st half, 2003 2nd half, 2003 Maintenance, Route 66 NYA 64 18 1st half, 2003 2nd half, 2003 -- Not available. NYA Not yet auctioned. a. Approximate. Source: Ministerio de Obras Publicas. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 271 Notes The authors wish to thank Florencio López-de-Silanes for many helpful com- ments. They also thank Pablo González, Manuel Cruzat, and Ángel Gajardo for their valuable assistance. 1. During the 1980s taxes were reduced and the government was able to finance the transition to a private pension system without going into deficit. Privatization revenues were not used to delay fiscal adjustment, as happened in Argentina. 2. In a personal communication, Rolf Lüders observed that he had not been able to detect improved performance in privatized firms discussed in Hachette and Lüders (1994). His explanation was that managers of state-owned firms were ide- ologically committed to efficiency during the 1980s. For further evidence of this, see appendix table 5A.1. 3. We do not include the water sanitation companies, which were sold during the late 1990s. 4. A terminal is an autonomous operational unit within a port that consists of adjoining berthing spaces and their associated support and service areas, thereby making it possible to auction them as separate units. 5. Of the 12 companies in the unregulated sector, 4 face no domestic compe- tition (CAP, ENAEX, LAN Chile, and Soquimich), while Colbun, ENDESA, and Gener represent almost all of the generating capacity in the interconnected central electric system. 6. For example, in 1965, value added in state-related firms was just 14.5 per- cent of GDP. 7. However, CORFO may have only displaced private investment in those sectors. 8. Related firms are those that belong to the same conglomerate. 9. In other words, the Central Bank exchanged fresh money for a claim on profits. 10. The government also privatized a radio company. The only valuable assets in this company were the rights to the FM spectrum, which were sold separately under allegedly questionable circumstances. The remaining AM frequencies were not valuable, and the company went bankrupt very shortly thereafter. 11. Some of these 16 firms had never been publicly traded, while others were taken off the stock market (that is, became private or "closed") after privatization. 12. This is a very slow process, since the firms are not required to provide the information. Obtaining data for Empremar, Fepasa, and Ferronor took almost three months because it required obtaining information not only from the firms but also from the original state-owned firms, which no longer exist. 13. Another source of information at the plant level is the Instituto Nacional de Investigación Agropecuaria survey, which registers quarterly data on many of the variables of interest for this study. Unfortunately, privacy considerations sur- rounding information provided by the National Institute of Statistics, which owns the survey, make it impossible to use the data for our purposes. 14. The digital form does not include the number of workers, which must be reconstructed from the FECUs in paper form. 15. Circular 239, Superintendencia de Valores y Seguros, Santiago, 1982. 16. We have included the electric-generating companies among unregulated firms because they sell a large fraction of their energy through unregulated long- term contracts. 272 FISCHER, GUTIÉRREZ, AND SERRA 17. Telex, a long-distance operator, was not regulated during the period under consideration. 18. PPEs appear in the FECUs. The PPE increase is significant for the change in medians. 19. The law excludes mobile telephony (except for access charges) from the re- quirement that the Antitrust Commission should decide whether the market is competitive, so user prices are free. 20. Initially, the creation in the early 1990s of an independent interconnected system in the northern part of the country helped to increase the use of existing ca- pacity in that area. However, business mistakes in the late 1990s led to overcapac- ity and large losses. 21. When computing future marginal costs, and very dry conditions are simu- lated, the regulated or node price is the outage cost (that is, the cost to users of long- run supply failures). 22. Chile has no stranded-cost principle, so the introduction of new technology may reduce the value of existing power plants to zero. 23. The majority of the mobile phones are sold as calling card phones and do not have fixed monthly contracts. 24. The price drop has not been as sharp on other routes. Carriers pay so-called accountancy rates to their foreign counterparts for traffic imbalances on interna- tional routes. On those routes where outgoing traffic exceeds incoming calls, the marginal cost of providing service should include the accountancy rate. 25. For more details on these arguments, see Engel, Fischer, and Galetovic (forthcoming). For a different perspective, see Gómez-Lobo and Hinojosa (2000). 26. Some doubts exist about the rationale for this last type of concession con- tract, since it appears to be a means of evading the standard budgetary process. 27. The franchise was awarded to the firm that had the highest score on an in- dex that weighed (mainly) the toll and the payment to the government. Because of poor auction design, payments to the government had a higher relative weight, so the bidders set tolls at the ceiling and bid positive payments. 28. If the government promised to compensate the franchise-holders for each change in the network that was claimed to affect their traffic flows, there would be endless and expensive negotiation of the impact of the changes. An example of the effects of such restrictions is Orange County's Riverside Freeway, which is terminally congested because its contract with the private 91 Express Lanes does not allow ex- pansion without permission from the owner of the private franchise. See Engel, Fis- cher, and Galetovic (forthcoming). 29. A similar approach had already been used in the United Kingdom when awarding the franchises of the Second Severn Bridge and the Queen Elizabeth II Bridge over the Thames. The main difference with the U.K. approach is that there were no auctions for the bridges. 30. This requires that the contract specify a minimum toll in order to avoid the threat of expropriation of the franchise-holder. 31. San Antonio is an exception and has been able to expand these support areas. 32. Ex ante rents are dissipated through this cash payment. See Engel, Fischer, and Galetovic (2001) for a detailed analysis. 33. Report of the President of Empresa Portuaria de Iquique, 2000. 34. Empresa Portuaria San Antonio. References Beard, T. Randolph, David L. Kaserman, and JohnW. Mayo. 2001. "Regulation, Ver- tical Integration and Antitrust." Journal of Industrial Economics 49 (3): 319­33. THE EFFECTS OF PRIVATIZATION ON FIRMS: THE CHILEAN CASE 273 Bitran, Eduardo, and Raul Sáez. 1994. "Privatization and Regulation in Chile." In Barry Bosworth and others, eds., The Chilean Economy: Policy Lessons and Challenges, pp. 329­77. Washington, D.C.: Brookings Press. Corporación de Fomento de la Producción. n.d. "Privatización de empresas y ac- tivos, 1973­1978." Gerencia de Normalización de Empresas, Santiago. Demsetz, Harold. 1968. "Why Regulate Utilities?" Journal of Law and Econom- ics 11: 55­66. Engel, Eduardo, Ronald Fischer, and Alexander Galetovic. 2001. "How to Auction an Essential Facility When Underhand Integration Is Possible." NBER Working Paper 8146. National Bureau of Economic Research, Washington, D.C. ------. 2002. "Highway Franchising with Subsidies." Cowles Foundation Discus- sion Paper 1354; Economic Growth Center Discussion Paper 840. Yale Uni- versity, New Haven, Conn. ------. Forthcoming. "Privatizing Roads: An `Old' New Approach to Infrastruc- ture Provision." Regulation. Fischer, Ronald, and Pablo Serra. 2000. "Regulation of the Electric Market in Latin America." Economía 1 (1): 155­98. Foxley, Juan, and Jose Luis Mardones. 2000. "Port Concessions in Chile: Contract Design to Promote Competition and Investment." Public Policy Journal. Issue 223. World Bank, Washington, D.C. Galetovic, Alexander. 1998. "Desatando a prometeo: Reformas microeconómicas en Chile 1973­1989." Perspectivas en política, economía y gestión 2 (1): 131­56. Gómez-Lobo, Andrés, and Sergio Hinojosa. 2000. "Broad Roads in a Thin Coun- try: Infrastructure Concessions in Chile." Technical Report 2279. World Bank, Washington, D.C. Hachette, Dominique. 2000. "Privatizaciones: Reforma estructural pero incon- clusa." In F. Larraín and R. Vergara, eds., La transformación económica en Chile. Santiago: Centro de Estudios Públicos. Hachette, Dominique, and Rolf J. Lüders. 1994. La privatización en Chile. Santi- ago, Chile: Centro Internacional para el Desarrollo Económico. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits from Priva- tization: Evidence from Mexico." The Quarterly Journal of Economics 114: 1193­242. Larraín, Felipe, and Rodrigo Vergara. 1995. Macroeconomic Effects of Privatiza- tion: Lessons from Chile and Argentina. Washington, D.C.: World Bank. Larroulet, Cristián. 1984. "Reflexiones en torno al estado empresario en Chile." Estudios Públicos (14): 129­51. Mamalakis, Markos J. 1976. The Growth and Structure of the Chilean Economy: From Independence to Allende. New Haven, Conn.: Yale University Press. Meller, Patricio. 1996. Un siglo de economía política chilena (1890­1990). Santi- ago: Editorial Andrés Bello. Rosende, Francisco, and Andrés Reinstein. 1986. "Estado de avance del programa de reprivatización en Chile." Estudios Públicos (23): 251­74. Sáez, Raul E. 1996. "Las privatizaciones de empresas en Chile." In Oscar Muñoz G., ed., Después de las privatizaciones: Hacia el estado regulador. Santiago: Dolmen-CIEPLAN. Sanfuentes, Andrés. 1984. "Los grupos económicos: Control y políticas." Colec- ción Estudios Cieplan (15): 131­70. 274 FISCHER, GUTIÉRREZ, AND SERRA Valenzuela, Mario. 1989. "Reprivatización y capitalismo popular en Chile." Estu- dios Públicos (33): 175­217. Vergara, Rodrigo. 1996. "Privatización de la banca: La experiencia chilena." Es- tudios Públicos (63): 335­45. Williamson, Oliver. 1976. "Franchise Bidding for Natural Monopoly­In General and with Respect to CATV." Bell Journal of Economics 7 (spring): 73­104. 6 Privatization in Colombia: A Plant Performance Analysis Carlos Pombo and Manuel Ramírez IN THE EARLY 1990S, THE COLOMBIAN government launched an economic liberalization program through the promotion of market competition and institutional deregulation. The economic openness package included ma- jor structural reforms encompassing foreign trade policy, the exchange rate regime, capital flow controls, central bank independence, privatiza- tion, labor legislation, foreign investment legislation, and social security and pension regimes.1 Historically, the size of the public sector in Colombia has been below the average of other Latin American countries such as Argentina, Brazil, Peru, and Venezuela. Yet revenues from privatization have had an im- portant impact on the government's short-run fiscal policy, financing the majority of investment in social programs from 1994 to 1998. During the 1993­98 period, the privatization program in the productive sector was dominated by the sale of assets in the power, natural gas trans- portation, manufacturing, and, to a lesser degree, the water and sewer- age sectors. Reforms in the telecommunications sector have induced new private investment in the public enterprises rather than outright changes in ownership. Colombia's most important institutional and regulatory reform dur- ing the 1990s took place in public utilities, where free entry was granted to private sector providers. This implied the establishment of modern and independent regulatory commissions for electricity and natural gas, water and sanitation, and telecommunications. Hence, economic dereg- ulation in Colombia was part of a comprehensive long-term strategy to promote new roles for the public and private sectors. However, a decade later, economic liberalization has not been well documented or analyzed on a sectoral basis.2 275 276 POMBO AND RAMÍREZ Some examination has been undertaken of the privatization process, however, including the work of Zuleta and others (1993) and Montene- gro (1994, 1995). These papers document in a preliminary manner the motivations that induced the government to rely on privatization as an economic instrument for promoting market competition, but they do not provide any empirical analysis of efficiency performance after privatiza- tion. Papers by Gutierrez and Berg (1999) on telecommunications and Pombo (2001a) on electric utilities document the regulatory reforms in these two sectors and present the evolution of some indicators that pro- vide a partial evaluation of such reforms. Thus, the documentation of Colombia's privatization programs and regulatory reforms during the 1990s is still incomplete; empirical evidence is required to gauge the suc- cess of the design and implementation of those economic policies. This chapter seeks to fill that gap by providing an ex post performance analysis of the privatization programs based on a representative sample with emphasis on manufacturing and power plants, following the bench- mark approach of Megginson, Nash, and van Randenborgh (1994) and La Porta and López-de-Silanes (1999). The objectives of the study are twofold. First, it seeks to measure the changes in performance indicators for a sample of manufacturing and power firms that underwent privatiza- tion, were restructured because of new regulations, or started operations under the new regulatory environment. Second, the chapter aims to model technical efficiency and profitability variables controlling for industry and plant characteristics, ownership type, and regulatory variables in order to evaluate the effect of privatization on plant performance. The chapter is organized as follows. The next section analyzes privati- zation efforts within the context of overall deregulation, private invest- ment in public infrastructure, and the promotion of free market policies. The chapter then examines the privatization programs by economic sec- tor. It begins by analyzing the divestiture program of former Institute for Industrial Promotion (Instituto de Fomento Industrial, or IFI) enterprises from 1986 to 1997 and then continues with a brief summary of the di- vestiture program for the state oil company, which affected the natural gas and regional gasoline distributing companies. The section ends with an analysis of the regulatory reform of Colombia's power sector, which was greatly affected by privatization. Next we present the core results of the chapter, evaluating the null hypothesis of structural changes in indicator mean and median regarding firm profitability, efficiency, investment, payroll size, and sales. The analysis takes into account industry-adjusted indicators by specific control group for the newly privatized firms in man- ufacturing and power utilities. It also looks at thermal power generation as a measure of technical efficiency--based on Data Envelopment Analy- sis (DEA) techniques--before and after the 1994 regulatory reform in the power sector. The chapter then presents the empirical evidence of firms' efficiency and profitability indicators, controlling for plant characteristics, PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 277 industry-specific variables, ownership structure, and variables related to regulatory policy. The last section offers some concluding remarks. The Deregulation and Privatization Program in Colombia Privatization in Colombia was originally approached as a tool for eco- nomic deregulation and promotion of market competition. The objective of the privatization program designed during the 1990s was to create in- centives for and redirect private investment in public infrastructure and network industries. This was to be achieved through concession contracts, sales contracts, and sectoral regulatory reforms. Concession contracts are instruments for promoting the involvement of private investment in public works and residential services. Concessions had been virtually abolished in Colombia since 1930, when nationaliza- tion and direct government involvement in the market economy became more prevalent. Before that, concessions had been widely used during the 19th century in railroads, mining, and crude oil exploitation. The eco- nomic deregulation policy of the 1990s restored concessions as a favored instrument for enhancing investment in strategic infrastructure sectors such as railroads, ports, airports, and highways. In 1991 the constitution was amended to introduce new rules for property rights regarding resi- dential public services and the development of public infrastructure, cre- ating a legal basis for implementing concessions. In 1993 new legislation (Law 80) set up a flexible legal framework regarding public contracting and concession regimes. The main objectives of the law were to introduce equal treatment into the awarding of state contracts to private and public firms and to extend the length of contracts. Specifically, the law allows the signing of contracts of more than 20 years in duration. At roughly the same time, the 1990 Government Development Plan was addressing the new economic agenda: economic deregulation, trade liberalization, and sectoral regulatory reforms. Afterward, a series of documents from the National Council for Economic and Social Policy (Concejo Nacional de Política Económica y Social, or CONPES), as well as the laws govern- ing residential public services, electric power, telecommunications, and privatization, set forth specific rules and guidelines regarding private in- vestment, the upcoming privatizations, and regulatory reform of network industries.3 Concession-type contracts were used in public works infrastructure projects such as maritime ports, road construction and maintenance, airports, aqueducts and sewers, railroads, and mobile phone networks. The recent studies of Alonso and others (2001) and Bonilla and others (2000) document the most important concession contracts by economic sector in Colombia. The former focuses on the contracts' characteristics 278 POMBO AND RAMÍREZ and incentive mechanisms, providing a preliminary assessment. The lat- ter analyzes the provision of utilities and transportation infrastructure in the largest cities of Colombia's Atlantic coast region. This study is important because concessions have been more active in those cities, where a history of poor local governance had translated into low- quality residential public services for decades before private providers were allowed to enter the market. According to the results of those studies and several follow-up CON- PES documents, one can conclude that concessions have not been widely implemented. By 1998, 35 concession contracts were signed (see appendix table 6A.1 for a list of concessions). Out of 1,400 municipal and rural aqueducts within the country, only 4 contracts were written for water companies; out of a possible 20 airports, only 3 contracts were written. The only concessions in telecommunications have been in mobile teleph- ony. Local phone companies have implemented joint-venture contracts with private investors for network expansion, as has the public long- distance carrier, TELECOM. Railroad concessions have been limited to cargo transportation, mainly to one operating concessionaire in coal transportation. In fact, in 1998 the rail network in operation was only half the size of the national network in 1970.4 Despite their limited coverage, concessions have been important in promoting private investment in road maintenance, maritime ports, and the construction of new gas pipelines. Concessions were not the only facet of the privatization program. In addition, the program involved the outright sale by local, regional, or na- tional public institutions of equity shares in several enterprises in the man- ufacturing, network utilities, natural gas distribution, and banking indus- tries.5 The privatization program was centered on the IFI equity transfers, the Colombian state oil company (ECOPETROL) divestiture program, and the direct sales of municipal or regional public utilities (MPUs and RPUs), most of which were in the power sector. Table 6.1 displays a com- plete list of the number of privatization contracts that took place in the productive sector between 1986 and 1998. Three comments are worth making. First, the number of sale contracts as well as their amount is too low in comparison with other international experiences. Aside from public utilities, privatization sales added up to $547 million.6 This amount is equivalent for instance, to the privatization sale of a government shareholding in a single company such as Volkswa- gen, Singapore International Airline, or Elf Aquitane.7 Privatization con- tracts were generally limited to the sale of equity shares of mixed-capital enterprises in manufacturing, natural gas and gasoline distribution, and, to a lesser degree, in services and mining, according to data up to 1998. Second, privatization in Colombia was not a centerpiece of policy, but rather a policy instrument designed to complement economic deregula- tion. In that sense, privatization was intended to ease industrial restruc- turing processes. Third, privatization of network industries arose as one PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 279 Table 6.1 Privatization Program in the Real Sector in Colombia, 1986­98 Total Number of Divesting public sale Industry contracts institution (US$ million) Manufacturing 27 IFI 288.1 Consumer goods 8 IFI 7.3 Intermediate goods 12 IFI 220.8 Capital goods 7 IFI 60.0 Mining 4 IFI 3.5 Natural gas 2 ECOPETROL 205.5 Gasoline distribution 5 ECOPETROL 41.2 Fishing 1 IFI 1.5 Services 6 IFI 6.9 Power sector 12 MPU 5,060.0 Water and sanitation 1 MPU 2.9 Total 58 5,610 Note: IFI Instituto de Fomento Industrial, ECOPETROL Compañía Colombiana de Petróleos, MPU municipal public utility, and RPU regional public utility. The privatized firms of IFI and ECOPETROL shareholdings were in all cases mixed-capital enterprises rather than state-owned enterprises. Source: ECOPETROL, requested files; IFI, requested files; DNP 1993, 1997b, 1997c; Dager 1999; Bonilla and others 2000; Alonso and others 2001; Pombo 2001b. instrument for promoting market competition. It came as part of ongoing sectoral regulatory reforms aimed at enhancing industry efficiency, chan- neling private investment, and deregulating market entry, especially in the provision of residential public services. The power sector has been the leader by far in accumulated privatization sales (90 percent of all privati- zation sales) followed by sales in manufacturing (5.1 percent) and natural gas transportation and gas distribution (3.6 percent). Hence, one can conclude that privatization in Colombia, in contrast to the experiences of other Latin American countries such as Argentina, Chile, Mexico, and Peru, was not a comprehensive process. Privatization in Manufacturing The privatization program in manufacturing was centered around the sale by the Instituto de Fomento Industrial of shares from its investment port- folio in a group of manufacturing and nonmanufacturing enterprises. The IFI was founded by Decree-Law 1157 of 1940 and became a strategic tool for state promotion of industrialization. The IFI's main objectives are to provide long-term credit to private enterprises and to advance risk capital 280 POMBO AND RAMÍREZ to industrial investment projects. Typically, the IFI's resources come from domestic saving through the issue of certificates of deposit and long-term bonds. In the international market, the IFI leverages loans from multilat- eral agencies and commercial banks. The role of the IFI in creating new manufacturing was central during the 1950s and 1960s. Today Colombia's largest private capital enter- prises in the steel, chemical, paper, fertilizer, metalworking, and automo- bile sectors are former IFI-associated companies. The IFI's larger projects were oriented to capital-intensive industries and producers of intermedi- ate materials and were an integral part of Colombia's import substitution industrialization policy, which sought to generate a new supply of manu- factured goods for the domestic market. The IFI firms to a large extent drove Colombia's industrialization process during the postwar years. Thus, the formation of mixed-capital enterprises channeled private-sector investments into new activities. IFI also guaranteed a degree of stability in foreign investment participation. The IFI's founding statutes are spe- cific in ordering the sale of equity shares once the government considers the new enterprises to be established in their respective markets. Hence, privatization traditionally has been a financial instrument used by the Colombian government. The role of the IFI, however, made for a differ- ent type of privatization than that carried out in other Latin American countries, as most firms in the Colombian case were mixed-capital enter- prises rather than state-owned enterprises. CONPES document 2378 (DNP 1988) set forth an accelerated timetable for the privatization of IFI enterprises. In so doing, the policy placed more emphasis on the transfer of assets than on the IFI's new in- vestments. In December of 1987, the IFI had capital shares in 45 manu- facturing and nonmanufacturing enterprises. Thirty of them were in oper- ation and the others had already begun a liquidation process. In addition, there were investments in 6 ongoing projects.8 Table 6.2 summarizes the IFI's sale program. The equity transfer program involved three steps in each privatization contract: the selection criteria, the stock assessment, and the method of sale. The selection criteria singled out for sale of equity shares all those op- erating enterprises that were not subject to special legal procedures, as well as ongoing projects that had not started business operations within three years of initial disbursements.9 The stock assessment process sought to de- termine the value of the firms' net assets and the stock price. The assess- ment studies took into account several parameters such as the present value of company cash flows, asset benchmarking, asset book values, stock exchange prices, and reposition and liquidation costs. In addition, all stocks were listed on the domestic stock exchange markets as well as at the National Stock Registry Office to lend transparency to the process. Regarding the sale method, the IFI used several bidding procedures: pri- vate offers to current shareholders or to the company's employee unions Table 6.2 IFI Privatization Program, 1986­97 Stock Stock Total Total IFI assessment sale sale sale Privatization share (current (current (million (million Sector Company name date (percent) pesos) pesos) pesos) US$) Selling method Fishing COPESCOL July 1991 49.0 1,000 6,505 956 1.5 Public bid Manufacturing EMPACA S.A. May 1986 29.2 10 150 54 0.3 Public bid Manufacturing SUCROMILES S.A. May 1986 15.6 100 2,400 247 1.3 Public bid Manufacturing VIKINGOS S.A. July 1986 35.5 10 16 113 0.6 Stock market Manufacturing UNIKA S.A. March 1988 3.4 10 95 105 0.4 Stock market Manufacturing FORJASCOL S.A. Dec. 1988 n.a ASSETS 1,700 5.7 Public offer Manufacturing SOFASA Feb. 1989 49.8 1,000 18,362 19,935 52.1 Public offer Manufacturing CICOLSA March 1990 17.4 100 100 14 0.0 Private offer Manufacturing AICSA S.A. April 1990 49.0 10 144 190 0.4 Public offer Manufacturing ING RISARALDA S.A. July 1990 11.7 100 421 972 1.9 Public offer Manufacturing PAPELCOL S.A. Aug. 1990 22.7 ASSETS 16,218 32.3 Public offer Manufacturing COLCLINKER S.A. Oct. 1990 15.7 1,000 16,160 1,909 3.8 Private offer Manufacturing RIOCLARO S.A. Dec. 1990 10.3 100 430 2,185 4.4 Stock market Manufacturing CCAa Dec. 1990 0.0 0 0 0 0.0 Private offer Manufacturing COSEDA June 1991 20.0 1,000 1,277 255 0.4 Private offer Manufacturing ASTIVAR Aug. 1991 31.0 100 2,800 130 0.2 Private offer Manufacturing TEXPINAL Sep. 1991 32.4 5 160 3,534 5.6 Private offer Manufacturing PROVICA Sep. 1991 13.2 1,000 1,414 67 0.1 Private offer Manufacturing CONASTIL Jan. 1992 59.9 1,000 1,000 1,014 1.5 Private offer Manufacturing FERTICOL April 1992 0.7 10 10 1 0.0 Preferential offer 281 Manufacturing PENNWALT Nov. 1992 40.7 10 158 1,223 1.8 Private offer (Table continues on the following page.) 282 Table 6.2 (continued) Stock Stock Total Total IFI assessment sale sale sale Privatization share (current (current (million (US$ Sector Company name date (percent) pesos) pesos) pesos) million) Selling method Manufacturing FATEXTOL Feb. 1993 16.0 1,000 2,250 540 0.8 Stock market Manufacturing FRIGOPESCA Dec. 1994 47.4 100 440 2,512 3.2 Public bid Manufacturing INTELSA April 1995 15.7 1,500 16,500 130 0.2 Public offer Manufacturing COSECHAR Oct. 1995 1.4 500 695 8 0.0 Public offer Manufacturing QUIBI S.A. April 1996 20.7 10 45 578 0.6 Public offer Manufacturing CERRO MATOSO Feb. 1997 47.7 100 28,264 155,814 150.3 Preferential offer/ public bid Manufacturing NITROVEN Dec. 1997 10.3 1,000 702,933 21,088 20.3 Preferential offer/ public bid Mining FOSFONORTE S.A. Jan. 1989 1.1 1,000 1,250 1 0.0 Private offer Mining FOSFOBOYACA S.A. Feb. 1990 6.4 1,000 1,000 9 0.0 Private offer Mining PROCARBON Sep. 1991 0.1 100 270 9 0.0 Stock market Mining PRODESAL Oct. 1991 11.6 100 921 2,164 3.5 Stock market Services PROHOTELES S.A. May 1986 10.8 10 39 43 0.2 Stock market Services CIAC S.A. March 1989 0.5 10 38 4 0.0 Private offer Services COLAR LTDA. Aug. 1989 n.a ASSETS 100 0.3 Public offer Services CORFERIAS S.A. Oct. 1989 5.6 10 65 276 0.7 Private offer Services CORFIDESARROLLO Sep. 1993 16.1 100 217 3,295 4.8 Stock market Services COKOSILK S.A. Jan. 1997 16.2 690 690 876 0.8 Preferential offer/ public bid Manufacturing 23.7 230,536 288.1 Mining 4.8 2,184 3.5 Services 9.8 4,593 6.9 Fishing 49.0 956 1.5 Total 238,269 300.1 Note: IFI Instituto de Fomento Industrial. The stock assessment value is the equity book value after the assessment study, which represents the preprivatization price. After 1995 all privatization contracts were subject to Law 226. a. CCA: Equity shares seized by Banco Colombia's trust fund in 1986. Source: IFI, requested files; Dager 1999. 283 284 POMBO AND RAMÍREZ and retirement associations, public bids, and the domestic stock exchange. The first method consisted of one preferential offer to individuals or com- mercial partners who had a direct involvement in either the company's ownership or control. These transfers followed the logic that block-holders represent the company's best interests. The second method followed a public bid scheme. Specifically, either the IFI or the investment bank in charge of underwriting the sale called for offers from strategic investors. Similarly, the listing on the stock exchange was intended to give small shareholders the opportunity to benefit from privatization. In addition, before 1995 at least 15 percent of the public equity shares in a company had to be offered to cooperatives, investment funds, union associations, and the company's employees. From May 1986 to December 1997, there were 38 privatizations af- fecting IFI enterprises. Three aspects merit further comment. First, in all but one case, the shares held by the IFI accounted for less than half of the firm's net worth. In one firm (CONASTIL), the IFI reported a 59 percent shareholding. In three cases the sale was based on book values of assets of firms undergoing liquidation. One of them was a liquidating paper-mill project (PAPELCOL) that never started operations. These data confirm that the role of the IFI was oriented toward promoting technology trans- fers and fostering entrepreneurship rather than exerting direct control on managing policies. Second, the data suggest that the sales process was suc- cessful in the sense that stock prices were in all cases greater than or equal to the preprivatization nominal stock price. However, there is no evidence to ascertain if fixed assets were correctly valued before privatization. Third, the total sales figure for manufacturing was $288 million, which re- flects the modest government involvement in manufacturing by the end of the 1980s. Privatization of Natural Gas and Gasoline Distribution The state oil company's sale of equity shares from its investments in the natural gas and distribution industry as well as the gasoline retail distri- bution networks represented the privatization program in the natural gas and gasoline industry. Privatization was restricted to the sale of those as- sets that were not directly related to crude oil exploration, transportation, and refineries, and other investments in nonoil businesses.10 Table 6.3 summarizes the divestiture program of ECOPETROL up until mid-1999. During this period two gas companies were sold. Law 226 of 1995 reg- ulated these transactions and required that privatization sales must give a preferential offer to the solidarity sector, which includes the former com- pany's labor union, worker associations, and cooperative firms. For that reason, these sales had two rounds. First was the preferential offer to co- operative associations. The offer price was equal to the assessment price. Cooperatives purchased around 10 percent of shares of Gas Natural and Table 6.3 ECOPETROL Privatization Program, 1993­99 Stock Stock assessment sale Value Value ECOPETROL price price sale sale Privatization shareholding (current (current (million (US$ Company name Activity year (percent) pesos) pesos) pesos) million) Sale method Natural gas companies Gas natural Transportation 1997 60.6 6,667 19,902 168,494 147.8 Preferential-offer distribution stock market COLGAS Distribution Ongoing 16.2 143 Preferential-offer stock market Promigas Transportation 1997 28.8 2,800 3,362 56,830 49.8 Preferential-offer stock market INVERCOLSA Distribution Ongoing 24.8 64 Preferential-offer stock market Surtigas Distribution No sale 15.4 256 Gases Guajira Transportation No sale 6.2 n.a distribution Gasoline companies Terpel Sabana Distribution 1993 40.0 n.a 6,562 4,200 5.3 Direct offer Terpel Bucaramanga S.A. Distribution 1993 36.1 n.a 7,691 14,477 18.4 Stock market Terpel del Centro S.A. Distribution 1993 49.7 n.a 221 10,386 13.2 Stock market Terpel Sur S.A. Distribution 1993 45.6 n.a 6,502 1,705 2.2 Stock market Terpel Norte S.A. Distribution 1993 18.0 n.a 705 1,615 2.1 Stock market Total natural gas 197.6 Total gasoline 41.2 285 n.a. Not applicable. Note: After 1995 all privatizations were subject to Law 226; ECOPETROL stopped the sale of IVERCOLSA in 2000. COLGAS had not been sold as of 2001. Source: ECOPETROL Planning Office, requested files; Decree 829 of 1999; DNP 1993b, 1997b, 1997c. 286 POMBO AND RAMÍREZ 2 percent of Promigas at those prices. The second round consisted of a si- multaneous first price auction on the country's three stock markets among previously registered bidders. These auctions were successful, particularly for Gas Natural, where the sale price was three times greater than the base price.11 Privatization of the gasoline distribution companies took place in 1993, so no preferential offer was made. The sale was through the stock ex- change, except in one case where the sale was through a direct offer. Do- mestic investors bought 100 percent in three TERPEL companies and 50 percent in the other two cases.12 ECOPETROL also offers financing for these purchases. On average, 36 percent of the shares was financed through direct credits, and the remaining 64 percent was paid in cash. Two observations are worth mentioning. First, except for Gas Natural, ECOPETROL's shareholdings in the privatized companies were less than 50 percent on the privatization date. Therefore, like the IFI enterprises, firms were not directly subordinated to or controlled by ECOPETROL manage- ment policies. Moreover, those companies were independent in their invest- ment expansion plans, and company wage policy was set independently of ECOPETROL. Second, the sale of the TERPEL gasoline stations was the first privatization transfer after the privatization was authorized in 1993. TERPEL was traditionally the competitor of private retailers. Thus, this transfer implied that gasoline retail distribution became wholly privately owned and run, in contrast to other oil producers in Latin American coun- tries such as Ecuador, Mexico, and Venezuela, where gasoline distribution remains vertically integrated with the state oil company. Industry Restructuring and Privatization in the Power Sector Regulatory reform in Colombia's electricity supply industry is supported by the Electric Law (Law 143) and by the Residential Public Services Law (Law 142) of July 1994. This reform has been the most important and comprehensive since 1967, when the national grid company, Inter- conexión Eléctrica S.A. (ISA), was established. The reform changed the structure of the vertically integrated industry. The new regulatory insti- tutions started to operate one year later. The reform's core elements fol- lowed the schemes adopted in Great Britain: separating generation, transmission, and distribution markets; setting up an electricity spot market or pool; and developing a long-term contract market for elec- tricity.13 Law 143 created the Regulatory Commission for Energy and Gas (Comisión de Regulación de Energía y Gas, or CREG) and rules re- garding the sector's planning and expansion plans, the regulatory scheme, power generation, transmission and grid operation, grid access fees, the rate-setting regime for electricity sales, concession contracts, and environmental issues. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 287 The power sector reform sought to introduce new competition and set up an independent regulatory system. In that sense, the main purpose was to set the basis for the expansion and diversification of power generation sources, improving both the sector's efficiency and its reliability. Political willingness to support this plan was high after 1992, when the country was caught in the middle of a generalized power shortage and electricity rationing schedules were imposed. The generating system had to be made less vulnerable to abnormal hydrological conditions (namely, El Niño) and more reliant on thermal generation from either coal or natural gas. There are several points to make about the separation of electricity gen- eration, transmission, and distribution. First, the split among power ac- tivities implied the divestiture of the main power holdings that were verti- cally integrated monopolies. The same happened with the national grid company, which had to sell all of its power-generating units in 1995. The new regulatory framework seeks to promote market entry and competi- tion among generators. They compete openly by sending their bids one day ahead to the pool. The sale price is based on an hour of use, and it dif- ferentiates between peak and off-peak hours. The National Dispatch Center, which is located at ISA headquarters, combines information regarding the system's constraints, such as hydrological factors, reservoir levels, and transmission bottlenecks, with final commercial demand in order to de- termine the dispatch orders. Thus, the market price that the pool sets is the highest marginal bid that clears the market each hour. Based on the above, the pool administrator runs the next-day merit order dispatches.14 Electric power is bought either through direct purchases from the pool or through contracts signed directly between generators and final users. However, the pool administrator runs the invoicing generated by all financial agree- ments. That is, that office pays and collects bills derived from contracts. The regulatory scheme treats power transmission as a natural monop- oly, and thus CREG guarantees equal access to the grid to all providers. ISA is not allowed to have an equity share in either power-generating or -distributing companies. Power distributors face two types of regulation. The first one is price regulation. CREG currently sets the markup formula for distributors, as well as determining the nature of the costs that can be passed through to final users. These include the direct purchase costs such as the pool sale price and transportation charges, capacity charges, and costs of the reserve provisions that stabilize the system and prevent bot- tlenecks in the transmission system. Price regulation at this stage differs from most systems that have moved toward electricity markets that have adopted price-cap rules. The second type of regulation concerns quality control, whereby companies are subject to sanctions if their service fails to meet minimum quality standards. The reform was designed to protect two types of final users. Residential users are mainly consumers whose elec- tricity prices are set by a markup formula, which includes past inflation. The reform also covered large clients, mainly commercial and industrial 288 POMBO AND RAMÍREZ users that use at least half a megawatt of electricity a month. Large clients may enter into purchase agreements with power distributors, wholesale re- tailers, or generators, enabling them to hedge against pool price volatility, a sensitive variable especially in hydro-based systems. The reforms and regulations led to a general divestiture across electric- ity holdings to fully separate power generation, transmission, distribution, and the setting up of new commercialization activities. Thus, privatization arose as one instrument for promoting market competition and industry restructuring, and it became a complementary policy within a broad deregulatory context. Table 6.4 describes the privatization process in the power sector. Privatization in the power sector had two phases. The first one was the 1996­97 privatization round, which focused on the sale of thermal plants and hydroelectric stations. Sales reached $3.9 billion and covered half of overall generating capacity. The most important transaction was the sale of 48 percent of the Bogotá Power Company's net worth, which also in- cluded the transfer of the local distribution network and the regional grid. The buyers were two holding companies owned by ENDESA and CHILECTRA, Chile's largest power generators. The second phase of the privatization program took place in 1998 and focused on the capitalization and sale of the CORELCA holding, which covered Colombia's northern Atlantic region. The restructuring involved splitting the holding into several independent companies according to power activity: generation, transmission, and distribution. The national grid company ISA bought 65 percent of the new transmission company's equity share. A holding company formed by American and Venezuelan utilities purchased a 65 percent equity share of the two distribution utili- ties founded after CORELCA's restructuring. The two transactions added up to $1.16 billion. Performance Analysis This section studies firm performance within a sample of former IFI man- ufacturing enterprises and the privatized power holdings. It also provides an efficiency analysis of the thermal generation sector in which new regu- lations led to privatization, restructuring, and the entry of new entities. The approach follows the general framework of Megginson, Nash, and van Randenborgh (1994) and La Porta and López-de-Silanes (1999) for performance analysis within the privatized power holdings; in the case of manufacturing, it follows as closely as possible the definition and con- struction of the performance variables of the benchmark cases. Nonethe- less, we include other performance variables regarding efficiency, market power, technology, and profitability indicators that follow standard methodologies in industrial economics and are based on a combination of Table 6.4 Privatization in the Power Sector, 1995­98 Sale Buyer Investor Capacity (US$ share country Utility (megawatts) Type million) Seller Buyer (percent) origin Batania 500 Hydro 497 ICEL ENDESA 100 Chile Chivor 1,000 Hydro 645 ISA CHILGENER 100 Chile Tasajero 150 Thermal, coal 30 ICEL Cooperative 58 Colombia TermoCartagena 180 Thermal, coal 15 CORELCA Electricidad-Caracas 15 Venezuela Cooperative 85 Colombia EPSA-generation 772 Hydro 535 CVC Houston Industries 56 United States 210 Thermal, gas EPSA-distribution Electricidad-Caracas Venezuela EEB-generation 2,312 Hydro 810 EEB Capital-Energia Holdinga 48.5 Chile, Spain 104 Thermal, coal (EMGESA) EEB-distribution 1,085 EEB Luz-Bogota Holdingb 48.5 Chile, Spain (CODENSA) EEB-transmission 141 EEB Capital-Energia Holdinga 5.5 Chile, Spain 141 EEB Luz-Bogota Holdingb 5.5 Chile, Spain (EEB-Head Quarters) CORELCA ElectroCosta- CORELCA Houston Inc. - Electricidad 65 United States, distribution Caracas Venezuela 289 (Table continues on the following page.) 290 Table 6.4 (continued) Sale Buyer Investor Capacity (US$ share country Utility (megawatts) Type million) Seller Buyer (percent) origin ElectroCaribe- 980 CORELCA Houston Inc. - Electricidad 65 United States, distribution Caracas Venezuela Transelca- 180.5 CORELCA ISA 65 Colombia transmission Total generation 5,228 2,532 Total distribution 2,065 Total transmission 462.5 Total privatization 5,060 Note: EEB Empresa de Energía de Bogota, EPSA Empresa del Pacifico S.A. (formerly CVC), CVC Corporación Autónoma del Cauca, ICEL Instituto Colombiano de Energía Eléctrica, CORELCA Corporación Eléctrica de la Costa atlántica, and ISA Interconexión Eléctrica S.A. a. Capital Energía ENDESA (Chile) ENDESA-Desarrollo (Sapin). b. Luz Bogota CHILECTRA (Chile) ENERSIS (Chile) ENDESA-Desarrollo (Spain). Source: MME 1996, 1998; reports to the Congress; ISA 1998, 1999. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 291 physical and financial series at the plant level, which have a similar or equivalent interpretation to those constructed from financial statements. Data Sets There were two reasons for taking such measures. One was the lack of avail- ability of data at the plant or firm level. In particular, there were no consis- tent records of the financial statements at IFI headquarters that would allow us to assemble a data set similar to that of La Porta and López-de-Silanes (1999). The second reason was the quality of data. The most comprehensive longitudinal data set is the Annual Manufacturing Survey (Encuesta Anual Manufacturera, or EAM) of Colombia's National Statistics Department (Departamento Administrativo Nacional de Estadística, or DANE). This survey is a census of medium and large manufacturing enterprises that has been undertaken annually since 1958. The variables and industrial classifi- cations in this data set have been relatively unchanged in format since the 1970 industrial census. In general, the EAM includes around 140 variables, covers 94 specific groups (at the four-digit international standard industrial classification level), and surveys an average of 6,000 plants.15 The survey re- ports variables such as gross output, number of employees, wages and bene- fits, raw materials, electricity consumption, sales, gross investment, and some financial variables such as asset book value and accounting depreciation. The manufacturing data set consists of 30 former IFI enterprises that were at some point either publicly owned or mixed-capital companies for which IFI was a founding partner or strategic investor. Nineteen of the 30 started business operations between the 1950s and mid-1970s. Among them were the country's largest steel mills, tire and tube plants, pulp and paper mills, and basic industrial chemical plants. Twenty-one firms were part of the 1986­97 IFI transfer program, accounting for 75 percent of total accumu- lated privatization sales. Four firms in the data set were exiting firms that were liquidated after 1992. The remaining firms are cases in which either the companies were transferred to the private sector before 1987 or the sale was postponed for strategic reasons. The data set is an unbalanced panel that records individual information from 1974 to 1998. Hence, the panel permits analysis of market dynamics at the firm level by tracking entry and exit flows. That feature makes the study sample appealing because of the robustness and length of this data set in contrast to the data sets used in other privatization studies, which at most have available time series with three or four obser- vations before and after privatization (Megginson and Netter 2001).16 Table 6.5 shows a summary of the basic variables for the IFI sample and total manufacturing before and after privatization. The sample con- sists of larger capital-intensive plants in Colombia. For instance, the sam- ple accounts for 5 percent of manufacturing's value added, 3 percent of manufacturing employment, and most important, 20 percent of the total capital stock as well as power consumption of manufacturing. 292 POMBO AND RAMÍREZ Table 6.5 Average Changes in Manufacturing Basic Variables after Privatization for IFI Sample and Total Manufacturing (US$ million at 1995 prices, unless otherwise specified) Manufacturing before Manufacturing after privatization, privatization, 1974­89 1990­98 IFI Total Ratio IFI Total Ratio Variable 1 2 1/2 3 4 3/4 Gross output 1,203 20,145 6.0 1,751 31,052 5.6 Value added 497 9,030 5.5 697 13,495 5.2 Total employment 20,631 495,404 4.2 15,806 622,594 2.5 Gross investment 127 956 13.2 92 1,372 6.7 Capital stock 2,016 9,679 20.8 2,934 14,450 20.3 Number of plants 25 6,356 0.39 27 7,475 0.36 Electricity consumption 1,060 4,953 21.4 1,594 8,299 19.2 (gigawatts per year) Note: IFI Instituto de Fomento Industrial. Definitions of each variable and its methodology can be found in appendix table 6.C1. Source: Authors' estimations based on DANE, various years. For the privatized power holdings, we were able to collect the financial reports as far back as 1983 from several sources; this data set allowed us to replicate similar measures of profitability, efficiency, assets and invest- ment, sales, and employment as in the benchmark study done by La Porta and López-de-Silanes (1999). See appendix table 6D.1 for a complete description of the power sector data sets. Changes in Performance in Manufacturing The study of IFI enterprises seeks to analyze changes in economic per- formance before and after privatization. The postprivatization period for the manufacturing sector has coincided with the economic openness pol- icy that began in the 1990s. In fact, 30 out of 37 IFI privatization contracts have taken place since March 1990. Thus, the analysis relies on the meas- urement of five types of indicators of performance and strategic competi- tion: efficiency and productivity, profitability and market concentration, labor, assets and investment, and sales or total output. The proxies are measured at the firm level. For incumbent firms the preprivatization period is 1974­89. Thus, changes in performance cap- ture two effects: privatization and economic deregulation. For entrants, the time series starts with the first recorded observation, which in most cases coincides with the start-up year of commercial operations. The PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 293 sample has four exiting firms, which shut down operations within the pri- vatization period (1990­98). The time period covered by the data set allows us to assume that in most cases firms have had enough time to complete re- structuring processes after privatization. Changes in performance are tested by using the Mann-Whitney (1947) rank sum test of means. The reading of the results is not straightforward because one has to analyze two forces behind the tests. On one hand, testing differences in sample means and assuming equal variances shows the direction of pri- vatization effects. On the other hand, by testing for differences in medi- ans, one is evaluating a change in the distribution shape, which may or may not coincide with the direction of the change in means. For exam- ple, increases in the sample mean with a negative change in the median show that a few individuals in the sample might explain overall varia- tion. Thus, privatization effects are not equally distributed or might have opposite results across firms. The basic results regarding performance changes in manufacturing are summarized in table 6.6, which depicts the raw indicators, and table 6.7, which presents the industry-adjusted indicators. The latter are ratios rela- tive to specific industry classification groups. These results call for several comments. The effects of structural changes on profitability are mixed. The traditional ratios of operating income to sales, operating income to capital stock, and net income to sales show that changes in means are not statistically significant and that the companies therefore were not neces- sarily unprofitable before privatization. The industry-adjusted ratios show that changes in medians rather than means are significant, in all cases at the 5 percent level. The median of the three indicators for the privatizing firms fell 25 points, on average, relative to their private competitors. At first glance, this outcome suggests that at least half of the IFI companies became less profitable after privatization. We analyzed those results further by constructing a set of complemen- tary profitability indicators that either are similar or have the same inter- pretation. These nontraditional indicators are the Lerner index as a proxy for the markup rates, the gross margin, and market share.17 According to the Lerner index, IFI enterprises were highly profitable before privatiza- tion. For example, the mean of the Lerner index as a proxy of the markup rate is 14.8 percent. That rate fell to 12.7 percent during the 1990s. More- over, the adjusted indicator shows that, on average, IFI firms were more than twice as profitable as their private competitors before privatization. After privatization this ratio fell slightly. The median of the markup rate, although statistically insignificant, increased from 6.6 percent to 10 per- cent, meaning that there was a positive convergence within the sample, partly due to the exit of less profitable plants after 1990. Changes are sta- tistically significant at the 5 and 10 percent levels. This result is contrary to most privatization studies, which have found that unprofitable public enterprises constituted a central argument in favor of equity transfers. 294 POMBO AND RAMÍREZ Table 6.6 Changes in Performance for the Sample of Privatized IFI Firms N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Profitability Traditional indicator Operating income/sales 394 227 0.2289 0.2206 0.37 0.2476 0.2403 0.43 Operating income/ 393 227 0.8684 0.8286 0.28 capital stock 0.4580 0.3045 1.96** Net income/sales 394 227 0.1328 0.1793 1.33 0.1953 0.2033 1.19 Nontraditional indicator Lerner index 394 227 0.148 0.127 1.59** 0.066 0.101 1.07 Gross margin 394 227 0.557 0.534 0.34 0.608 0.689 3.16*** Market share 394 227 0.135 0.108 2.50*** 0.073 0.079 1.32 Efficiency Traditional indicator Cost per unit 394 227 0.5757 0.5159 1.35* 0.5452 0.5312 1.45* Log(sales/employees) 394 227 10.562 11.037 6.59*** 10.527 11.177 6.28*** Log(sales/capital stock) 393 227 0.442 0.313 1.03* 0.777 0.435 2.41*** Nontraditional indicator Capital: partial 394 227 6.02 4.68 1.10 productivity 1.29 0.75 2.65*** Labor: partial 394 227 30,487 43,837 4.97*** productivity 21,161 29,207 4.33*** Translog index TFP 394 227 118.85 140.39 2.36*** 93.76 81.89 2.38*** Labor Log(workers) 394 226 5.59 5.47 1.07 5.84 5.62 1.58 Log(technicians) 354 221 3.04 2.89 1.34* 3.00 2.75 1.12 Log(administrative 394 222 4.53 4.56 0.34 employees) 4.71 4.54 0.33 Assets and investment Log(capital stock) 393 246 8.98 9.44 2.42*** 9.20 10.00 2.34 *** PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 295 Table 6.6 (continued) N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Capital/labor 394 227 173,184 228,968 1.28* 16,446 32,928 4.08*** Investment/value added 394 225 0.861 0.187 1.02 0.069 0.055 2.41*** Investment/before 393 225 0.091 0.059 3.53*** capital stock 0.059 0.025 5.27*** Machinery/investment 389 221 0.860 0.757 0.41 0.724 0.703 0.78 Investment/total 393 225 8,360 6,418 0.76 employees 1,533 1,529 1.13 Output Log(gross output) 394 227 16.69 17.07 2.61*** 16.89 17.48 2.74*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: IFI Instituto de Fomento Industrial, N sample size, and TFP total factor productivity. This table presents raw results for 28 manufacturing enterprises that were included in the 1988 IFI divestiture program. The data set includes 4 plants that shut down operations after 1990. The panel is unbalanced by construction, and the sample size refers to firm-year observations for two time periods: 1974­89, the years before privatization, and 1990­98, the years after privatization. The maximum number of firm-year observations before privatization is 394; after privatization, 227. The table presents for each empirical proxy the number of usable observations, the mean, and the median values before and after 1990. The table reports t-statistics and Z-statistics (Mann-Whitney nonparametric rank sum) as the test for significance for the change in mean and median values, respectively. Capital and labor productivities and capital/labor and investment/employee ratios are in 1995 US$. Definitions of each variable and its methodology can be found in appendix table 6C.1. Source: Authors' calculations. Part of the explanation can be found in the economic openness pro- gram and the corporate structure of these firms. IFI firms were a central piece within the mixed strategy of import substitution and export diversi- fication implemented since the mid-1960s.18 These firms enjoyed high effective protection until 1990. The drop observed in the mean of firm market share from 13.5 percent to 10.8 percent reflects the increase of im- ports within industry-specific domestic supply. Figure 6.1 depicts the markup rate evolution for the IFI sample and total manufacturing. Clearly, profitability shows a decreasing trend, although IFI firms still show above-average profitability in manufacturing. 296 POMBO AND RAMÍREZ Table 6.7 Industry-Adjusted Changes in Performance for the Sample of Privatized IFI Firms Industry-adjusted N N Mean Mean t-statistic Variable before after Median Median Z-statistic Profitability Traditional indicator Operating income/sales 394 227 0.9165 0.8788 0.26 0.9401 0.8018 2.46*** Operating income/ 393 227 2.1877 1.8446 0.95 capital stock 1.1061 0.6317 2.53*** Net income/sales 394 227 0.6348 0.6212 0.03 0.9157 0.7554 2.51*** Nontraditional indicator Lerner index 394 227 2.181 1.936 1.35* 1.416 1.398 0.20 Gross margin 394 227 0.190 1.281 1.23 1.022 1.000 3.62*** Market share 394 227 0.197 0.150 2.35*** 0.052 0.063 1.32 Efficiency Traditional indicator Cost per unit 394 227 1.012 0.964 0.89 0.948 0.984 0.49 Log(sales/employees) 393 227 0.998 1.008 1.99** 1.005 1.016 2.11** Log(sales/capital stock) 393 227 0.535 1.705 1.87** 1.018 1.127 2.05** Nontraditional indicators Capital: partial 393 227 8.681 7.197 0.85 productivity 1.733 0.986 2.72*** Labor: partial 394 227 1.321 1.385 0.76 productivity 1.069 1.022 0.98 Translog index TFP 391 158 1.408 1.609 1.66* 1.043 1.026 0.22 Labor Size 394 227 5.584 6.265 0.93 3.643 2.961 0.59 Size-workers 394 227 5.628 6.370 1.02 3.739 3.128 0.48 Assets and investment Capital stock 393 227 10.576 11.433 0.45 2.504 3.327 0.70 PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 297 Table 6.7 (continued) Industry-adjusted N N Mean Mean t-statistic Variable before after Median Median Z-statistic Capital/labor 393 227 4.915 4.153 0.77 0.674 0.966 2.12** Investment/value added 394 225 2.580 1.399 0.99 0.736 0.560 1.83* Investment/capital stock 393 225 1.186 1.068 0.49 0.577 0.382 3.57*** Investment machinery/ 393 225 1.612 1.190 0.90 investment 1.018 0.974 1.61* Investment/employees 393 225 1.744 1.646 0.22 0.746 0.616 1.13 Output Scale 394 227 6.313 7.846 1.81** 3.377 2.674 0.46 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: IFI Instituto de Fomento Industrial, N sample size, and TFP total factor productivity. The industry control group is the four-digit international standard industrial classification group to which a specific firm belongs. For each year and firm we compute industry-adjusted indicators by taking the ratio of the value of the indicator for the IFI firm to its industry control group. See table 6.6 for an explanation of the data set and significance tests on which this table is based. Definitions of each variable and its methodology can be found in appendix table 6C.1. Source: Authors' calculations. The gross margin rate is an indicator of working capital that shows how firms are restricted by payroll structures.19 The change in medians of this indicator is positive and significant at the 5 percent level, increasing from 60.8 percent before privatization to 68.9 percent afterward. This result indicates that half of the distribution was able to adjust its payroll structure to efficiency parameters. Most former IFI firms have had strict union convention clauses. The 1990 labor market reform (Law 50) elimi- nated wage rigidities such as the retroactive severance pay system and the mandatory reinstatement regime for workers with more than 10 years on the payroll.20 Thus, IFI firms could ease their payroll constraint and speed up the benefits derived from the 1990 labor reform after privatization. However, the industry-adjusted margin rate shows an opposite change in medians, which moved from 1.02 to 0.92 times, meaning that despite the IFI firms' new contracting flexibility, private competitors increased their gross margin faster than IFI firms during the 1990s. 298 POMBO AND RAMÍREZ Figure 6.1 Markup Rates for IFI Sample and Total Manufacturing, 1970­98 Markup rates 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 1970 1974 1978 1982 1986 1990 1994 1998 IFI Manufacturing Source: Authors' estimations based on DANE, various years, and Pombo 1999a. Corporate structure is another explanatory factor of IFI firms' prof- itability levels. As was pointed out earlier, the IFI companies were mixed- capital enterprises. At the time of privatization, the share of IFI in manu- facturing company equity was on average 24 percent and the accumulated sales were under US$300 million (see table 6.2). These numbers under- score the state's limited participation in manufacturing, where the IFI's policy of capital rotation implied a sale of equity once companies became incumbents and mature within the market. This limited state role in the economy was a sharp contrast to other Latin American countries. State- owned enterprises (SOEs) in Mexico, for example, represented 38 per- cent of the economy's capital stock by 1982, and there were SOEs in almost all manufacturing groups. In Chile SOE sales represented 14 per- cent of gross domestic product in 1965, and the government controlled the key foreign export sector of copper mining, as well as most large manufacturing enterprises located in industries such as paper and pulp manufacturing.21 PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 299 Efficiency gains underpin part of the pro-competitive pricing strategy that followed trade liberalization. In contrast to the profitability results, the efficiency outcomes are in the expected direction: there is evidence that IFI firms improved technical efficiency during the 1990s. The average (median) cost per unit decreased 6 percent (1.4 percent), while the mean and the median ratio of sales to employees rose 48 percent. At the same time, the median ratio of sales to capital stock fell 34 percent. These changes are all significant at either the 5 or the 10 percent level. The in- dustry-adjusted ratios of these indicators behave in different directions. On one side, the average and median ratios of sales to employees rise only slightly, but on the other side the average of sales to capital stock triples from 0.54 to 1.71 times. The median increases more moderately, moving from 1.02 to 1.13 times relative to their control group. These changes are significant at the 5 percent level. Last, neither the mean nor the median change in the industry-adjusted cost per unit is significant. These results indicate at least two things. First, an efficiency gain resulted from the large increase in labor productivity that improved firms' total factor productivity (TFP). Second, evidence of gains is not conclusive concerning capital productivity in the raw indicator of sales-to-capital ratio. Nonetheless, the change in the industry-adjusted indicator implies that even if IFI firms did not improve their capital productivity, they did much better than their private peers, which experienced a large drop in capital input productivity after 1990. To pin down the source of efficiency gains, we constructed nontraditional indicators of partial and total factor productivity. Ratios of value added to capital or labor are direct measures of partial input productivity. These measures are more accurate than sales- to-input ratios because changes in quality in raw materials are in fact a source of a firm's efficiency gains that might influence the direction of these performance indicators. The measuring of partial and total productivity shows that average (median) firm labor productivity increased 43 percent (13 percent) at con- stant prices,22 and the average TFP index rose by 22 basis points, which is equivalent to a 2.4 percent annual productivity growth rate. The TFP me- dian fell 11 basis points, meaning that efficiency gains were asymmetric across plants. The median for capital partial productivity contracted 58 percent, moving from $1.29 per unit of installed fixed capital to $0.75 at constant prices. These changes are significant at the 5 percent level. These results illustrate the direction of plant restructuring in these com- panies. IFI firms represent 20 percent of Colombia's manufacturing capital stock (see table 6.5). The study sample consists of large capital-intensive plants. In fact, the measurements of capital-to-labor ratios show that be- fore privatization, IFI firms were 4.9 times more capital intensive than their counterparts in the private sector. That number dropped to 4.1 after 1990. The behavior of the TFP index for the IFI sample closely follows the cycle for total manufacturing (figure 6.2). Productivity for IFI companies 300 POMBO AND RAMÍREZ Figure 6.2 Total Factor Productivity Indexes for IFI Sample and Total Manufacturing, 1970­98 Indexes 110 100 90 80 70 60 50 1970 1974 1978 1982 1986 1990 1994 1998 IFI Manufacturing Source: Authors' estimations based on DANE, various years, and Pombo 1999a. plummets by 40 basis points between 1979 and 1983, according to the TFP index. This means a ­10 percent TFP growth per year, whereas the average efficiency loss in manufacturing was 2.6 percent per year. This productivity shock implied that even 15 years later the surviving firms had not been able to reach the TFP levels of the mid-1970s. One factor that explains this cycle in productivity is the crisis in capital productivity of the 1980s and its slow pace of recovery. The adjusted indicator shows that for IFI firms the median of capital productivity was 1.8 times that of their private competitors. During the 1990s the number con- verged to 0.99 times, meaning that productivity was about the same in both groups and reaching industry benchmarks. Moreover, investment plum- meted because of an overinvestment problem. In fact, the mean (median) rate of capital accumulation dropped from rates of 9.1 percent (5.9 percent) per year before privatization to rates of 5.9 percent (2.5 percent) per year while the median investment rate fell from 6.9 percent to 5.5 percent per PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 301 Figure 6.3 Investment Rates for IFI Sample and Total Manufacturing, 1970­98 Manufacturing IFI 0.18 0.80 0.16 0.70 0.14 0.60 0.12 0.50 0.10 0.40 0.08 0.30 0.06 0.04 0.20 0.02 0.10 0.00 0.00 1970 1974 1978 1982 1986 1990 1994 1998 Manufacturing IFI Source: Authors' estimations based on DANE, various years. year after 1990. These changes are significant at the 5 percent level. At the same time, the mean of investment per employee decreased from $8,360 to $6,418 per year at constant prices, while the median remained unchanged. The change in the last indicator is not statistically significant. The changes in the adjusted indicators of investment rates are in the same direction. Figure 6.3 illustrates the overinvestment problem for IFI companies, where investment rates were 2.5 times the level for total manufacturing until 1989. Thereafter the gap decreased to 1.4 times. Thus, IFI firms strongly rationalized capital spending in order to pin down excess capac- ity. Various factors help explain the sample's overinvestment. One is associated with the macroeconomic disequilibriums of the late 1970s gen- erated by coffee boom prices that led to the appreciation of the Colombian peso. Temporary trade liberalization, which enhanced capital goods imports, was another factor. Microeconomic factors also came into play. During the 1977­83 period, the IFI and private investors undertook the two largest industrial investment projects since the 1960s. One was a cement mill (Compañía Colombiana de Clinker, or Colclinker) that began 302 POMBO AND RAMÍREZ operations in 1977 and still is the country's largest. The second, the establishment of one of the largest nickel processing plants in Latin Amer- ica (Cerromatoso), began operations in 1983. As a founding partner, IFI had a 45 percent equity share in this nickel plant. Similar results have been found in other country-specific studies of pri- vatization. Equity transfers to private holders do not necessarily boost investment, as many government officials argue when calling for privati- zation. By the late 1980s, most IFI-associated firms clearly had an excess capacity problem, which became a bottleneck once the domestic market started to peter out as a source of demand growth.23 Nonetheless, even with decreasing trends in overall investment rates, one might expect in- vestment to become more selective through spending on new machinery to replace worn-out capital equipment. This was not the case for the IFI com- panies. The mean (median) embodied investment rate--the ratio of in- vestment spending on machinery to total investment--declined from 86 percent before 1990 (72 percent) to 76 percent (70 percent) after privati- zation. However, these changes are not statistically significant. The ad- justed embodied investment rate shows that the median changed from 1.02 to 0.97 times after privatization and is significant at the 10 percent level. These results on investment behavior suggest that one aspect of firm restructuring after privatization was the reduction of excess capacity. Labor productivity is the other component explaining the direction of the firms' restructuring after privatization. As was pointed out, IFI firms were able to increase their TFP growth by 2.4 percent per year after 1990. How- ever, those rates have not allowed a full recovery from the dramatic loss in capital productivity that those companies experienced during the 1980s. The mean (median) of labor productivity rose from $30,487 ($21,161) to $43,837 ($29,207) at constant prices during the 1990s. This change means a 43 percent real increase in value added per worker. As reported in table 6.2, around 4,800 workers were laid off, representing a 23 percent total payroll reduction within IFI companies after privatization. The layoff com- position was 3,700 workers, 750 administrative employees, and 340 tech- nicians. However, the changes in means and medians of the labor series by type of occupational category are not statistically significant. The adjusted labor indicators given by plant size show an increase in the mean from 5.58 to 6.26, while the medians decrease from 3.64 to 2.96. These measurements also show the asymmetric effects of labor lay- offs across plants, but the direction of change in medians implies a lower plant operating scale. Again these changes are not statistically conclu- sive; therefore, one might have some caveats about the labor adjustment direction. In any case, labor cuts were small in comparison with other national experiences. The Mexican manufacturing payroll was halved during privatization (La Porta and López-de-Silanes 1999), for example, while the privatization of British Telecom involved the layoff of more than 5,000 workers (Armstrong, Cowan, and Vickers 1994). PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 303 Figure 6.4 Labor Productivity Indexes, IFI Sample and Total Manufacturing, 1970­98 Manufacturing IFI 0.18 0.80 0.16 0.70 0.14 0.60 0.12 0.50 0.10 0.40 0.08 0.30 0.06 0.04 0.20 0.02 0.10 0.00 0.00 1970 1974 1978 1982 1986 1990 1994 1998 Manufacturing IFI Source: Authors' estimations based on DANE, various years. Figure 6.4 depicts the labor productivity indexes for the IFI sample and for total manufacturing. One sees a positive gap between IFI firms and the average for total manufacturing, starting in 1984 and coinciding with the turning point for the IFI firms' productivity trends. The average value of the labor productivity gap was 20 points during the 1980s and nearly triple that in the 1990s. Thus, IFI firms made greater efforts in increasing labor productivity relative to their private competitors. This is partly explained by the moderate levels of layoffs but also by adjustment in plant efficiency scales, which by definition eliminates any diseconomies in production. The positive but lower rates of capital accumulation plus the payroll contraction after 1990 explain changes in the ratio of plant capital to la- bor. The mean (median) capital units per employee rose 32 percent (100 percent) at constant prices after privatization. The adjusted indicator also shows that IFI firm plants became more capital-intensive relative to their control group. The median changed from 0.67 to 0.97 times--that is, these firms effectively substituted capital for labor input, adjusting 304 POMBO AND RAMÍREZ their relation to industry benchmarks. The changes are significant at the 5 percent level. Another important result is that IFI firms were able to increase their sales despite the payroll contraction and the lower rates of capital accu- mulation. Mean (median) gross output increased 38 percent (59 percent). This increase implies a 4.2 percent (6.5 percent) output growth rate per year during the 1990s. The adjusted indicator shows that, on average, out- put rises from 6.3 to 7.8 compared with the control group. These results complement those found regarding plant size. On one hand, plant size adjustment implies a correction in plant scale but on the other hand, IFI firms were able to exploit new economies of scale. Thus, part of the ob- served TFP growth is explained by economies of scale and the rest might be attributable to changes in technology.24 Another issue that is important to analyze is the role of labor costs in explaining cost reductions that allow firms to set lower prices and behave more competitively. According to the redistribution hypothesis, one can expect a drastic fall in wages after privatization because of lower labor productivity and a renegotiation of unions' convention clauses. In the case of the IFI firms, we do find that the mean (median) of the industry- adjusted value added per worker before privatization was 1.3 (1.07) (see table 6.7). Hence, the argument does not apply. Table 6.8 presents the wage data by type of worker. The striking result here is the large increase in real wages of workers in the privatized IFI firms. The increase in means (median) in the per capita wage for blue-collar work- ers was 30 percent (17 percent) after privatization in real terms. For white- collar workers, the mean (median) increase was 36 percent (27 percent) at 1995 prices. More interesting are the changes in the industry-adjusted per capita wages. Wages for both blue- and white-collar workers increased rela- tive to private competitors. La Porta and López-de-Silanes (1999) document similar results for Mexico. All tests are significant at the 5 percent level. To summarize, the analysis of changes in performance indicates that IFI firms followed pro-cyclical trends relative to their control group. That is, there was no asymmetric performance of these companies in contrast to their private peers. Part of the explanation for this is that IFI firms were mixed-capital enterprises and followed profit-maximizing pricing rules rather than pursuing second-best prices or net transfers through subsidized sale prices. As a result, management strategies fol- lowed private-sector benchmarks, although rate-setting policy favored, at some points and for some cases, the dominant position of IFI firms within domestic markets. Because IFI's role was the strategic but sup- portive one of promoting firm capitalization, it never sought to control company management policies. The results regarding plant efficiency are consistent with the expected effects, but in this case, privatization helped to speed up firm labor restructuring, which implied a strong change in magnitude in the labor productivity indexes. Finally, the wage data PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 305 Table 6.8 IFI Firms: Role of Transfers from Workers N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Total wages/sales 394 227 0.1154 0.0988 2.33*** 0.0914 0.0691 4.06*** Log(total employees) 394 227 5.98 5.90 0.74 6.12 5.96 1.28 Real wages per 394 226 338.4 437.1 5.96*** blue-collar worker 312.5 363.1 3.98*** Industry-adjusted real 394 226 1.06 1.18 3.28*** wages per blue-collar 1.04 1.09 2.30*** worker Real wages per 394 222 543.43 738.71 6.34*** white-collar worker 492.02 625.07 5.01*** Industry-adjusted real 394 222 1.09 1.26 3.80*** wages per white-collar 1.04 1.14 3.14*** worker Real wages per worker 394 227 386.7 500.1 6.83*** 355.8 448.4 5.23*** Industry-adjusted real 394 227 1.16 1.34 3.32*** wages per worker 1.05 1.12 1.79* * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: IFI Instituto de Fomento Industrial and N sample size. The series of per capita wages are expressed in US$ at 1995 prices. See tables 6.6 and 6.7 for an explanation of the data set and significance tests on which this table is based. Definitions of each variable and its methodology can be found in appendix table 6C.1. Source: Authors' calculations. indicate that we can reject the hypothesis of transfers from workers to shareholders after privatization. Changes in Performance in the Power Sector Our analysis of the performance in the power sector takes into account the effects of the 1994 reform on firm entry, market competition, and ef- ficiency gains. In that sense, the analysis focuses on firm changes in means and medians of direct measures of profitability, efficiency, assets and in- vestments, and sales of the privatized power holdings. The study sample covers the equity transfers in three of five regional power systems where privatization took place in both power generation and power distribu- tion, as described earlier. They are the former Bogotá Power Company, Cauca Valley Corporation (CVC), and the Corporación Regional de la 306 POMBO AND RAMÍREZ Costa Atlántica (CORELCA) holding. The control group is Public Enter- prises of Medellín (EPM), which is a municipally owned company and has been traditionally the most efficient public utility. Series were chained after 1995 keeping the prereform holding structure in order to have com- parable statistics. Tables 6.9 and 6.10 present the main results regarding the performance effects of privatization on the power holdings. Several facts are worth not- ing. First, the reform has had a direct and positive effect on operating efficiency. The average cost per unit dropped 45 percent at constant prices after reform, compared with the preprivatization rate. The mean (median) ratio of sales to assets (property, plant, and equipment, or PPE) rose 17 percent (18 percent), while the mean (median) ratio of sales to employees rose 20.3 percent (15.7 percent). The same happened with the ratio of operating income to employee, where the mean (median) increase was 63 percent (48 percent) at constant prices after the reform. Changes are sig- nificant at the 5 percent level. There are at least three important sources of these efficiency gains. First, utilities made an effort to reduce both power losses and the tariff- collecting problem in power distribution. This was the case for the Bogotá Power Company in particular, which drastically reduced its power loss in- dexes from 53 percent in 1985 to 22 percent in 1996.25 Second, the reform and privatization induced new investment in incumbent firms, in contrast to what was observed for manufacturing. All investment rates at least dou- bled, on average. Notice that capital stock remained unchanged, but this finding is not statistically significant. Total assets usually have several bi- ases depending on the depreciation schedules. For that reason a more ac- curate indicator is the current investment rate. Notice that in most cases the industry-adjusted changes in performance of operating efficiency and investment-adjusted indicators are not statistically significant, meaning that despite their efforts, the newly privatized power holdings could not match EPM's efficiency changes. Third, employment cuts were not as sig- nificant as in the case of manufacturing. The four electric holdings alto- gether had, on average, 13,300 employees before the reform. This number only decreased to 11,600 employees during the 1995­99 period. Thus, the observed 23 percent real increase in labor productivity resulted from an in- crease in sales rather than drastic employment cuts. In fact, mean (median) sales increased by 16.4 percent (21.1 percent). The new regulation has used two instruments to encourage market entry. One is a guaranteed minimum return on installed capacity. The second in- strument is the power purchase agreement (PPA). These agreements are long-term contracts that allow generators to hedge against unexpected changes in demand and distributors to hedge against system constraints. One type of PPA initially implemented in Colombia is to pay what is gen- erated, which involves an advance purchase of plant capacity by a power distribution company Most thermal generators are marginal producers Table 6.9 Changes in Performance in the Sample of Privatized Power Utilities and Public Enterprises of Medellín N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Profitability Operating income/ 48 20 0.3208 0.1891 3.093*** sales 0.3587 0.2262 2.410** Net income/sales 48 20 0.1382 0.0882 0.693 0.1992 0.0998 0.794 Operating income/ 48 20 0.0562 0.0288 3.060*** PPE 0.0556 0.0397 2.544** Operating income/ 48 20 0.0997 0.0463 3.155*** net worth 0.0958 0.0452 3.876*** Operating efficiency Cost per unit 48 20 0.0292 0.0207 1.790** 0.0226 0.0194 1.561 Log(sales/PPE) 48 20 1.2574 1.4289 3.260*** 1.2278 1.4101 2.907*** Log(sales/employees) 48 20 2.0020 2.2035 4.469*** 2.0021 2.1578 3.957*** Operating income/ 48 20 112.82 183.91 4.367*** employees 105.54 156.68 4.321*** Labor Log(employees) 48 20 3.4354 3.3987 0.4701 307 3.5205 3.4255 0.8080 (Table continues on the following page.) Table 6.9 (continued) 308 N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Assets and investment Log(PPE) 48 20 4.1807 4.1733 0.1135 4.1513 4.1509 0.1750 Investment/salesa 48 8 0.0039 0.0066 1.9950** 0.0033 0.0063 1.1710 Investment/ 48 8 0.4869 0.8374 1.6493* employeesa 0.2721 0.6909 1.5220 Investment/PPEa 48 8 0.0742 0.1579 3.1909*** 0.0647 0.1521 1.9900** Log(PPE/total 48 20 0.7453 0.7746 0.4286 employees) 0.7817 0.7960 0.2690 Output Log(sales) 48 20 5.4382 5.6023 2.5933*** 5.4771 5.6886 2.6250*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N sample size and PPE property, plant, and equipment. This table presents raw results for three privatized power holdings and for public enterprises of Medellin. The data set is a balanced panel by construction, and the sample size refers to firm-year observations for two time periods: 1983­94, the years before the regulatory reform, and 1995­99, the postreform years. The maximum number of firm-year observations before the reform is 40; after the reform, 20. The table reports for each empirical proxy the number of usable observations, the mean, and the median values before and after the sector regulatory reform (1995), and the t-statistic and Z-statistic (Mann-Whitney nonparametric rank sum) as the test for significance of the change in mean and median values. Value variables before transformations in logs are in millions of pesos at 1995 prices. Definitions of each variable as well as details on Colombia's power sector data sets and definitions can be found in appendix table 6.C1. a. Postreform years are 1995­96 due to unavailability of appropriate data for later years. Source: Authors' calculations. Table 6.10 Industry-Adjusted Changes in the Performance of Privatized Power Utilities N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Profitability Operating income/ 36 15 0.7122 0.4088 2.308** sales 0.8264 0.6109 1.757* Net income/sales 36 15 0.1677 0.0504 0.341 0.1949 0.0076 0.537 Operating income/PPE 36 15 0.7458 0.3206 2.018** 0.6933 0.4037 2.233** Operating income/ 36 15 0.7647 0.3152 1.996** net worth 0.7559 0.3729 1.736* Mean tariff 26 6 1.7088 1.0469 2.032** 1.4213 1.0296 1.977** Operating efficiency Cost per unit 36 15 0.5101 0.6649 1.242 0.4162 0.5036 1.137 Log(sales/PPE) 36 15 1.0055 1.0488 0.828 0.9663 0.9989 1.116 Log(sales/employees) 36 15 1.0296 0.9910 1.394 1.0324 0.9955 1.220 Operating income/ 36 15 1.2877 1.1073 1.142 employees 1.2240 1.0026 1.199 Labor 309 Log(employees) 36 15 1.0013 0.9974 0.128 1.0477 0.9974 0.475 (Table continues on the following page.) 310 Table 6.10 (continued) N N Mean before Mean after t-statistic Variable before after Median before Median after Z-statistic Assets and investment Log(PPE) 36 15 1.0133 0.9763 1.826** 1.0031 0.9593 2.150** Investment/salesa 36 6 1.1585 1.0014 0.371 0.9723 0.9074 0.539 Investment/employeesa 36 6 1.6667 0.6618 1.350 0.8622 0.3741 1.546 Investment/PPEa 36 6 1.2120 1.0372 0.411 1.0115 1.4047 0.108 Log(PPE/employees) 36 15 1.0733 0.9021 1.386 1.2772 0.9461 1.530 Output Log(sales) 36 15 1.0115 0.9937 1.225 1.0361 1.0081 1.199 *Significant at the 10 percent level. **Significant at the 5 percent level. ***Significant at the 1 percent level. Note: N sample size and PPE property, plant, and equipment. This table presents the industry-adjusted results for three privatized power holdings. Performance proxies are adjusted relative to Public Enterprises of Medellín. The data set is a balanced panel by construction, and the sample size refers to firm-year observations for two time periods: 1983­94, the years before the regulatory reform, and 1995­98, the postreform years. The maximum number of firm-year observations before the reform is 36; afterwards it is 15. See table 6.9 for an explanation of the significance tests; see appendix table 6C.1 for definitions of the variables. a. Postreform data are for 1995­96 due to unavailability of appropriate data for later years. Source: Authors' calculations. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 311 whose objective is to generate a hedge for the system. In fact, thermal units are spread across electricity holdings, and some of them became inde- pendent companies after privatization as mentioned earlier (see table 6.4). Overall, 63 thermal plants with an effective capacity of 3,800 megawatts were operating in 1998; that represented a 32 percent share of all gener- ated electricity. Among those thermal units, 21 started commercial opera- tions after 1993, and 16 are privately owned. This is not a coincidence since the government had already undertaken an emergency plan to ex- pand thermal generators to overcome the 1992 power generation crisis.26 Figure 6.5 illustrates the evolution of the country's effective and avail- able capacity from 1991 to 1999. The available capacity rose from 1,000 gigawatt hours (GWh) in 1990 to around 2,500 GWh in 1998.27 In sum, fixed investment in thermal generation has played a central role in im- proving system reliability as well as promoting market entry in power generation. The behavior of profitability indicators, however, did not mirror the ef- ficiency gains. Notice the striking result that all profitability indicators, adjusted and unadjusted, dropped after the regulatory reform. The mean (median) ratio of operating income to sales was 32.1 percent (35.8 per- cent) before the reform for the study sample (see table 6.9). The indicator fell to 18.9 percent (22.6 percent) during the postreform years. The ratios of operating income to PPE and to net worth, indicators of firms' profit rates on gross and net fixed assets, respectively, were reduced by close to one-half. These changes are significant at the 5 percent level. The adjusted indicators show the same behavior, that is, the privatized holdings lost relative profitability compared with their control group. As in the case of manufacturing, these results are the opposite of the expected effects of privatization on firm profitability. The conventional wisdom would say that any gains in input productivity must have a direct impact on firm profitability rates if and only if there are no drastic changes in market competition. The 1994 regulatory reform implied more market competition for both power generation and distribution. First, ownership composition changed drastically within the first five years after the regu- latory reform, which led to a balanced distribution of power-generating capacity between public and private utilities. By 1998 public utilities accounted for 42 percent of the power-generating capacity, while private and mixed-capital utilities held a 58 percent share. The largest generator had a 21 percent market share.28 This outcome contrasts with the initial divestiture in the United Kingdom where nonnuclear generation was split into a duopoly, and in Chile, where the three largest power generators control 85 percent of the market. On the power distribution side, privatized utilities dropped their rates for final users after 1995. Moreover, they have converged to EPM's final- user rate. The industry-adjusted rate for residential users dropped from 1.70 to 1.04 after the reform. If one takes into account the unregulated 312 POMBO AND RAMÍREZ Figure 6.5 Thermal Capacity versus Thermal Generation, 1991­99 Jan. 1991 Feb. 1992 March 1993 April 1994 May 1995 June 1996 July 1997 Aug. 1998 Sept. 1999 0 500 1,000 1,500 2,000 2,500 3,000 3,500 Gigawatt Available capacity Efective capacity Source: ISA 1998, 1999. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 313 electricity market (industrial users), the drop must be even greater. Table 6.11 summarizes the main variables of the wholesale electricity market. Two facts are noteworthy. First, the evolution of electricity spot prices suggests that buyers--power distributors--have effectively hedged against pool price volatility. Real contract prices dropped 42 percent from 1996 to 2000. That outcome is important since contracts have a 75 percent market share in bulk electricity. Another important outcome is that mar- ket deregulation has sharply increased the number of unregulated users, most of which are large industrial and commercial clients. A sharp increase in financial costs during the first half of the 1990s also contributed to narrowing gross and net utility profits. The four regional markets under study had, on average, a 90 percent real increase in their financial costs relative to the average of the 1980s. The Bogotá Power Com- pany faced most of the indebtedness burden because of the overcosts gener- ated by the five-year delay in the start-up of the Guavio hydroelectric plant. An Analysis of the Productive Efficiency of Thermal Plants The previous section shows that the 1994 regulatory reform and resulting market competition encouraged power firms to achieve improvements in efficiency, to try to stop theft and other nontechnical losses of power, and to undertake new investments in power-generating capacity. This section presents the measurement of productive efficiency at the plant level for a sample of thermal plants that belong to public, private, and mixed-capital utilities. Of the 63 plants that belong to the interconnected system, only 32 units have been active, having a permanent or temporary production within a specific year. Because of changes in the statistical sources, the data set was divided into two samples. The first sample has records for 33 thermal plants, on average, for the 1988­94 period, that is, the years be- fore reform. The second one has records for 32 thermal units for the postreform years (1995­2000). The measurement of plant technical efficiency is based on Data Envel- opment Analysis techniques and requires information about inputs and output for each thermal unit. Plant inputs are capital, or capacity in megawatts (MW); labor (number of employees); and fuel consumption (coal, gas, fuel oil, and diesel oil). All fuels must have a common meas- ure unit, such as British thermal units (Btus) or thermal calories.29 Output is given in gigawatt hours (millions of kilowatt hours). Informa- tion for power generation, consumption by type of fuel, and capacity at the plant level is available by crossing the different data sets before and after 1994. Labor input is not directly observable for most units. There are two reasons for that problem. One is that before privatization thermal units were vertically integrated with power utilities; thus, payroll series were recorded following accounting criteria. Power companies kept labor 314 Table 6.11 Annual Averages for Wholesale Electricty Market Efficiency Variables, 1996­2000 Mean Mean Spot PPA spot PPA price price Commercial Unregulated Regulated Unregulated Unregulated price price index index demand demand demand demand users Year (US$/kWh) (US$/kWh) (Dic98 100) (Dic98 100) (GWh) (GWh) (GWh) share (percent) (number) 1996 0.0084 0.0348 52.8 125.0 3,329.6 454.5 2,875.0 0.1365 11.2 1997 0.0548 0.0321 342.7 115.2 3,410.1 453.9 2,956.2 0.1336 95.3 1998 0.0374 0.0288 233.7 103.4 3,452.5 659.5 2,793.0 0.1910 678.8 1999 0.0159 0.0220 99.3 79.1 3,316.5 676.1 2,640.4 0.2038 891.6 2000 0.0204 0.0203 127.7 72.9 3,387.3 843.7 2,543.6 0.2489 2,377.0 Note: GWh gigawatt hour, kWh kilowatt hour, and PPA power purchase agreement. This table shows the main variables of the wholesale electricity market. The spot price indicates the pool daily prices, and the PPAs are forward contracts of electricity prices and dispatched quantities. Both are market prices. Final residential and small commercial users, whose price formula is set by the regulatory commission, form the regulated demand. Unregulated users are large clients that underwrite purchase contracts with power generators and distributors. Commercial demand is equal to the sum of the regulated and unregulated demand. The last column reports the average number of large clients that are registered in the electricity market for a given year. Prices per kilowatt hour are in US$ at 1998 prices. Value series were deflated by the U.S. consumer price index. Source: ISA 1998, 1999; Mercado de Energía Mayorista (MEM) requested files. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 315 records to fulfill the requirements of financial reporting. Second, there was no regulator requesting information by power activity. Labor sta- tistics after 1996 have improved sharply since the regulator (Superin- tendent of Domiciliary Public Services) has been in charge of the SIVICO database. Labor series by power company are broken down by occupational categories, sectoral activities (that is, generation, trans- mission, and distribution), and by type of power generation. In addi- tion, after privatization the plants that were sold became new utilities. This allowed for making direct inferences about labor input (number of employees) by thermal substations. Fixed coefficients of labor to capac- ity were assumed based on the information sent by power generators to complete labor series before 1995. The results of the efficiency frontier measurement exercise based on 42 thermal plants that were active as marginal producers before and after the 1995 regulatory reform are dis- played in appendix table 6E.1. The results show that the most efficient plants before the reform were not the most efficient afterward. Entrants indeed pushed efficiency up and became benchmark technologies. Thus, the 1994 reform that sought to promote plant entry also spurred gains in technical efficiency through the new plants and the overhaul of oth- ers. Such was the case of Termo-Barranquilla, which is the country's largest thermal plant. Econometric Analysis of IFI Manufacturing Plants This section analyzes the role that plant characteristics, foreign trade vari- ables, and privatization played in determining privatization outcomes for the sample of IFI firms. The econometric analysis focuses on two key per- formance variables: plant profitability rates (Lerner indexes) and the translog indexes of total factor productivity as a proxy for technological change. This econometric exercise hopes to shed light on plant efficiency and markup determinants as well as to evaluate the significance of priva- tization within the model. The data set is an unbalanced panel of 28 IFI firms that records information for the 1974­98 period. The estimating equation follows the baseline pooled regression model: performanceit ( 0 i ) Xit B Zit it (6.1) where i equals 1, ..., n is the number of individual firms; t 1, 2,..., T is the number of observations in each panel; X is equal to firm characteris- tics variables; and Z is equal to specific industrial classification variables. Equation 6.1 allows the running of several types of regression models ac- cording to specific assumptions on the residual variance covariance matri- ces and individual effects i. In particular, the estimations relax the as- sumptions of constant variance across panels, the nonexistence of 316 POMBO AND RAMÍREZ individual effects, and instruments for endogeneity on right-hand-side variables. Plant characteristic variables are related to technology structures, la- bor composition, and the firm's market positioning actions. One ex- pected result is that technology-related variables have a positive impact on profitability gains. In that sense, plant size, operative scale, quality of raw materials, capital intensity, and relative labor productivity result in lower average costs that represent productivity gains due to new economies of scale. Plant payroll composition reflects quality in labor in- put. Thus, technicians should lead overall plant labor productivity be- cause skilled workers are more dynamic and generate productivity spillovers. Administrative employees, in turn, may generate inflexibilities that end up hurting profitability. Market positioning variables are those actions that strengthen a firm's market share. The firm's signals are in- vestment rates, the use of technological licenses, and product differentia- tion tactics such as advertising. These actions may persuade rivals to soften competition and adopt collusive prices, but because a competitor's best response might include hardening competition and setting dumping prices, there is no expected sign. Industry-specific variables are mostly related to foreign trade. Three main variables are used in the estimating equations: nominal tariffs, effec- tive protection rates, and Grubel and Lloyd (1975) indexes. The last is a proxy for trade in differentiated goods.30 Protectionism increases domestic profitability by deterring entry. Intraindustry trade, in contrast, implies trade in similar goods that makes entry credible because of the monopo- listic competition market structure, which drives sale prices to second-best prices.31 Hence, profitability decreases. Table 6.12 displays the main results for the markup determinants; these results call for several comments. First, in all cases, the firm's market share is the robust determinant. This is consistent with the observation that eco- nomic openness reduces the firm's market power and therefore decreases markup rates. Estimations show that a 10 percent decrease in market share reduces profitability by 9 percent. Second, the foreign trade variables are robust regressors and show the expected sign. On average, an increase of 10 percent in the effective pro- tection rate increases markups by 4 percent. In contrast, if intraindustry trade indexes rise 10 percent, markups will decrease on average by 3 per- cent. This finding is important from the perspective of strategic trade pol- icy. Competition through similar goods forces firms to undertake further specialization strategies to promote efficiency gains in order to compen- sate for the reduction in markup rates. Third, plant size and productive efficiency are important sources of profitability gains. IFI firms are on av- erage seven times larger than their competitors. As a result the observed gains in TFP partially offset the falling trend in firm markup. On average, if TFP indexes rise 10 points, they will increase markup rates by between Table 6.12 Markup Determinants for IFI Firms Equation 1 Equation 2 Equation 3 Equation 4 Equation 5 Pooled Panel Panel Within Pooled Equation 6 Independent variable OLSa FGLS FGLS FE 2SLSa FE IV Market share 0.8751*** 0.9318*** 0.9264*** 0.8931*** 0.9641*** 0.9459*** (0.0470) (0.0263) (0.0255) (0.0305) (0.0311) (0.0341) Labor productivity 0.0195*** 0.0085*** 0.0089*** 0.0108*** -- -- industry-adjusted (0.0039) (0.0017) (0.0016) (0.0024) -- -- Translog index TFP 0.0001* 0.0001*** 0.0001*** 0.0001* 0.0003*** 0.0002** (0.00003) (0.00001) (0.00001) (0.00004) (0.00011) ( 0.00006) Size 0.0012** 0.0008*** 0.0009*** -- -- -- (0.0006) (0.0002) (0.0002) -- -- -- Human capital 0.0226*** -- -- -- -- -- (0.0054) -- -- -- -- -- Grubel and Lloyd index 0.0521*** 0.0280*** 0.0385*** 0.0170* 0.0400** 0.0182* (0.0160) (0.0079) (0.0084) (0.0094) (0.0156) (0.0095) Effective protection -- 0.0381*** 0.0433*** 0.0337*** 0.0878*** 0.0328*** -- (0.0055) (0.0059) (0.0079) (0.0136) (0.0080) Dummy privatization -- -- 0.0125*** -- -- -- -- -- (0.0029) -- -- -- Dummy foreign investment -- -- 0.0144*** -- -- -- -- -- (0.0031) -- -- -- Constant 0.0084 0.0228*** 0.0214*** 0.0039 0.0470** 0.0016 (0.0076) (0.0035) (0.0041) (0.0069) (0.0206) (0.0089) 317 (Table continues on the following page.) 318 Table 6.12 (continued) Equation 1 Equation 2 Equation 3 Equation 4 Equation 5 Pooled Panel Panel Within Pooled Equation 6 Independent variable OLSa FGLS FGLS FE 2SLSa FE IV Regression statistics R2 0.6315 0.6699 0.6249 0.6245 Number of groups 28 28 28 28 Number of observations 613 621 621 621 620 564 Observations per group: Minimum 13 13 13 13 Maximum 25 25 25 25 F-test 116.03 238.69 145.15 235.16 [0.0000] [0.0000] [0.0000] [0.0000] Wald-Chi2(k 1) 2741 2932 [0.0000] [0.0000] F-test for all i 0 93.6 [0.0000] Heteroscedasticity tests Cook-Weisberg 267.79 [0.0000] White 293.58 [0.0000] Variance matrix residuals Homoscedastic panels Yes No No Yes Yes Yes Instrumental variable No No No No Yes Yes RHS endogenous variables TFP TFP Other equations in system TFP F(partial labor productivity, industry-adjusted; scale; capital/labor ratio, industry-adjusted). * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: FE fixed effects, FGLS feasible generalized least squares, IFI Instituto de Fomento Industrial, OLS ordinary least squares, IV instrumental variables, 2SLS two-stage least squares, and TFP total factor productivity. The table reports results from ordinary least squares, feasible generalized least squares, fixed effects, two-stage ordinary least squares, and fixed effects with instrumental variables for an unbalanced panel of 28 IFI firms for the 1974­98 period that were included in the 1988 privatization program. The maximum number of observations is 621. The dependent variable in all equations is the Lerner index, or markup rate. Standard errors appear in parentheses and p-values in brackets. Definitions of each variable and its methodology can be found in appendix table 6C.1. a. White-Hubert robust heteroskedastic standard errors. Source: Authors' calculations. 319 320 POMBO AND RAMÍREZ 0.005 and 0.02 points in markup rates. Fourth, privatization shows a consistent sign. Privatization induced a 1.2 percent increase in profit rates (column 3 in table 6.12). Finally, the foreign investment dummy has the opposite sign. In the context of the IFI sample, this result is not surpris- ing since some firms are located in industries that were once highly pro- tected and that kept lower efficiency levels with respect to parent firms and international standards. The econometric results on productive efficiency are displayed in table 6.13. Five comments are worth making. First, plant characteristics are rel- evant for TFP indexes. All the regression equations show that plant labor productivity, licensing, and number of technicians have positive effects. On average, an increase of 10 percent in partial labor productivity relative to the specific control group raises TFP by 3.2 percent. The effect of li- censing is the largest. If plants expand their spending on technological li- censing relative to their value added by 1 percent, they boost productivity by between 5.5 and 8.2 times. This finding is consistent with previous re- sults for total manufacturing and calls attention to the short-run effective- ness of using patented licenses for improving productivity rather than en- gaging in direct research and development spending.32 The number of technicians is a proxy for labor input quality. A 10 percent increase in this variable improves productivity by 1.12 percent. Second, the equation includes two variables to capture demand ef- fects on TFP measured by either the growth in value added or the log value of a firm's specific control group. The sign matches with the ex- pected one, which is consistent with the traditional hypothesis derived from the Verdoom law by which growth and productivity are con- strained by effective demand. The impact of aggregate demand is two- fold: domestic demand and export demand induce growth and improve productivity by learning. This in turn leads to improvements in price competitiveness, which induces higher levels of effective demand (Dixon and Thirlwall 1975).33 Third, privatization has a positive effect on productivity, causing an increase ranging from 0.27 to 0.53 points on TFP indexes. Fourth, the scale and the adjusted capital partial pro- ductivity coefficients have a negative impact on productivity. The inter- pretation of this result is not straightforward. The losses in capital pro- ductivity due to overinvestment suggest that IFI firms adjusted capital spending to close gaps with industry benchmarks. Fifth, markups are inversely related with TFP, which is not consistent with the self-investment-financing hypothesis of endogenous growth mod- els (Romer 1990; Barro and Sala-i-Martin 1995). In particular, one should expect a positive impact since larger profitability rates ease the self- financing of capital equipment and spending on research and develop- ment. After controlling for fixed effects, however, the expected sign is recovered. The within-regression coefficients show that a 10 percent in- crease in markups improves TFP by 13 percent. Table 6.13 Total Factor Productivity Determinants for IFI Firms Equation 1 Equation 2 Equation 3 Equation 4 Equation 5 Pooled Panel Panel Within Pooled Equation 6 Independent variable OLSa FGLS FGLS FE 2SLSa FE IV Labor productivity, industry- 0.3803*** 0.3633*** 0.3295*** 0.2219*** 0.4811*** 0.2384*** adjusted (0.0598) (0.0321) (0.0318) (0.0248) (0.0769) (0.0223) Capital productivity, 0.0094*** 0.0049*** 0.0050*** -- 0.0103*** -- industry-adjusted (0.0017) (0.001) (0.0009) -- (0.0019) -- Demand growthb 0.2011*** 0.0812** 0.0977** -- 0.1915** -- (0.0782) (0.0457) (0.0444) -- (0.0783) -- Scale 0.0316*** 0.0190*** 0.0175*** 0.0136*** 0.0262*** 0.0185*** (0.0043) (0.0023) (0.0023) (0.0038) (0.0045) (0.0035) Licensing 7.9312*** 7.9246*** 8.1922*** -- 5.3740** -- (1.9463) (1.3614) (1.4664) -- (2.1096) -- Compensation per worker, 0.3283*** 0.1031*** 0.1135*** -- 0.2392** -- industry-adjusted (0.0943) (0.0424) (0.0432) -- (0.1125) -- Advertising coefficient 2.2255*** 1.6442*** 1.4812*** -- 2.2842*** -- (0.5537) (0.4407) (0.4422) -- (0.5543) -- Log technicians 0.1152*** 0.1019*** 0.0778*** 0.1418*** 0.1875*** 0.1071*** (0.0324) (0.0176) (0.0163) (0.0262) (0.0353) (0.0215) Privatization dummy 0.3626*** 0.0525 -- -- 0.2700 -- (0.1073) (0.0438) -- -- (0.1118) -- Lerner index -- 0.4968*** 0.4130*** 1.3185*** 2.1912*** 1.3116*** -- (0.1570) (0.1502) (0.2899) (0.4225) (0.3626) 321 (Table continues on the following page.) 322 Table 6.13 (continued) Equation 1 Equation 2 Equation 3 Equation 4 Equation 5 Pooled Panel Panel Within Pooled Equation 6 Independent variable OLSa FGLS FGLS FE 2SLSa FE IV Log value addedb -- -- -- 0.3308*** -- 0.2702*** -- -- -- (0.0462) -- (0.0412) Relative capital/labor ratio -- -- -- -- -- 0.0339*** -- -- -- -- -- (0.0040) Constant 0.9934*** 0.6140*** 0.7227*** 5.7045*** 0.8739*** 4.3417*** (0.1497) (0.0762) (0.0691) (0.8684) (0.1591) (0.7796) Regression statistics R2 0.1954 0.2708 0.1663 0.4047 Number of groups 28 28 28 28 Number of observations 554 554 554 575 554 476 Observations per group: Minimum 4 4 5 5 Maximum 24 24 25 25 F-test 22.9 40.25 20.73 54.93 [0.0000] [0.0000] [0.0000] [0.0000] Wald-chi2(k 1) 289.6 296.4 [0.0000] [0.0000] F-test for all i 0 142.5 [0.0000] Heteroscedasticity tests Cook-Weisberg 73.51 [0.0000] Breuch-pagan LM stat 158.9 [0.0000] Variance matrix residuals Homoscedastic panels Yes No No Yes Yes Yes Instrumental variables No No No No Yes Yes RHS endogenous variables LERNER LERNER Other equations in system Lerner F( market share, Grubel & Lloyd index, effective protection) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: FE fixed effects, FGLS feasible generalized least squares, IFI Instituto de Fomento Industrial, OLS ordinary least squares, IV instrumental variables, and 2SLS two-stage least squares. The table reports results from ordinary least squares, feasible generalized least squares, fixed effects, two-stage ordinary least squares, and fixed effects with instrumental variables for an unbalanced panel of 28 IFI firms for the 1974­98 period that were included in the 1988 privatization program. The maximum number of observations is 621. The dependent variable in all equations is the translog index of total factor productivity (1974 100). For each year and firm, we compute industry-adjusted indicators by taking the ratio of the value of the indicator for the IFI firm to its industry control group. Industry control group is the four-digit standard industrial classification group to which a specific firm belongs. Standard errors appear in parentheses and p values in brackets. Definitions of each variable and its methodology can be found in appendix table 6C.1. a. White-Hubert robust heteroskedastic standard errors. b. By specific industrial classification. Source: Authors' calculations. 323 324 POMBO AND RAMÍREZ Power Plants: Statistical Analysis of Efficiency Scores This section reports the results of an econometric analysis of thermal power plant DEA efficiency scores. The exercise follows a limited dependent variable model because the dependent variable under analysis is censored by construction. It takes positive values and is bounded at 1.00; thus, the efficient plants record an efficiency score, yit, of 1. Other- wise, 0 yit 1. The sample might also be truncated because there is knowledge of independent variables only if yit is observed. This is partic- ularly important for marginal power producers when the thermal plants are shut down by maintenance, transmission, and generation constraints because there is no power dispatch. The baseline-censored model follows a linear specification: 0 1 yit c0 x B eit it yit (6.2) otherwise and the residuals follow a normal distribution with zero mean and con- stant variance. Equation 6.2 models efficiency scores as a function of plant characteris- tics, ownership structure, and regulatory policy dummies. Plant character- istics include plant age, capital-to-labor ratio, technology type, and load factor. Controlling for the load factor indicates how marginal a given pro- ducer is. A dummy that takes the value of 1 for all private plants captures ownership. The regulatory dummy tries to capture the effect of large cus- tomer definition. Thus, for each plant that dummy takes a value of 1 after 1998 (when the minimum usage limit was lowered, thus increasing the number of unregulated users from 100 to 900). The data set includes all ob- served records from all active thermal plants during the 1995­2000 period. Therefore, the data set is an unbalanced panel with 166 observations. Table 6.14 displays the Tobit regressions. The dependent variable in the first two equations is Score1, which represents plant efficiency scores measured under the assumption of constant returns to scale, and capital input is adjusted by its effective utilization. This adjustment normalizes plant capacity by load factor, which means that all producers are treated as if they were off-peak generators. The dependent variable in the third equation is Score2, in which the measure of plant efficiency relaxes the as- sumption of constant returns to scale. The reading of those results is as fol- lows. First, the equations exhibit high quality of fit, reported by the R2 of the OLS regressions.34 In particular, the overall effect of the plant charac- teristics, ownership structure, and regulatory policy dummy explains 90 percent of the efficiency scores once capital input is adjusted by capacity utilization; that number falls to 78 percent when the assumption of con- stant returns to scale is relaxed. Second, dummy variables for technology PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 325 Table 6.14 Determinants of Thermal Plants' Efficiency Scores Equation 1 Equation 2 Equation 3 Pooled Pooled Pooled Tobit Tobit Tobit Independent variable Score1 Score1 Score2 Adjusted capacity 0.0004* (0.0002) Age 0.0155*** 0.0175*** 0.0183*** (0.0018) (0.0018) (0.0029) Age2 0.0004*** 0.0005*** 0.0005*** (6E-04) (6E-05) (9.5E-05) Load factor 0.4169*** 0.3700*** 0.1128*** (0.0445) (0.031) (0.0428) Load factor2 5.1005*** 4.5298*** (1.207) (1.125) Capital/labor ratio 0.0010 (0.0006) Gas dummy 0.3653*** 0.3704*** 0.3960*** (0.0118) (0.0122) (0.0196) Combined cycle 0.1431 dummy (0.0923) Private ownership 0.0323*** dummy (0.0116) Public ownership 0.0423*** dummy (0.0117) Regulatory policy 0.0201* 0.0229** 0.0432** dummy (0.0108) (0.0112) (0.1762) Constant 0.4098*** 0.4593*** 0.5020*** (0.0206) (0.0208) (0.0315) Sigma 0.0660 0.0691 0.1122 Regression statistics R2-OLS 0.9104 0.9074 0.775 Uncensored 155156 152 observations Censored 710 14 observations LR~Chi(k 1) 377.3 379.5 228.9 [0.0000] [0.0000] [0.0000] Test residuals Cook-Weisberg (OLS) 0.00 0.04 2.46 [0.9924] [0.8445] [0.1168] Breuch-Pagan (OLS) 6.87 [0.4416] (Table continues on the following page.) 326 POMBO AND RAMÍREZ Table 6.14 (continued) Equation 1 Equation 2 Equation 3 Pooled Pooled Pooled Tobit Tobit Tobit Independent variable Score1 Score1 Score2 Ramsey-RESET (OLS) 1.83 0.59 0.28 [0.1439] [0.6225] [0.8391] Swilk (OLS) 4.99 4.67 3.35 [0.0000] [0.0000] [0.0004] * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: OLS ordinary least squares. This table reports results from Tobit regressions for an unbalanced panel of 33 thermal plants for the 1995­2000 period. The total number of observations is 166. The dependent variables are Score1 and Score2. Score1 is equal to plant efficiency scores measured under the assumptions of constant return to scale and convex technology. Capital input is multiplied by its effective utilization rate. Score2 is equal to plant efficiency scores measured under the assumptions of variable returns to scale and convex technology. Capital input is multiplied by its effective capacity utilization rate. Gas technology is a dummy that takes the value of 1 if plant technology is gas based, and 0 otherwise; combined cycle is a dummy that takes the value of 1 if plant technology, whether a thermal plant, has a combined cycle technology, and 0 otherwise; private ownership is a dummy that takes the value of 1 if the plant is privately owned, and 0 otherwise; public ownership is a dummy that takes the value of 1 if the plant is publicly owned, and 0 otherwise; and regulatory policy is a dummy that takes the value of 1 for the years that the definition of a large client was set to a minimum consumption of 0.5 megawatt per month. Standard errors appear in parentheses and p-values in brackets. Source: Authors' calculations. are robust and statistically significant in all equations. This implies that new gas-based technologies improve system efficiency, because they save on fuel consumption. Entrants played a central role in this particular issue. Third, the load factor is positively related, meaning that power losses associated with the frequent and costly plant start-ups are effectively reduced. The square of the variable is negatively related, however, showing that there are decreasing returns to scale at full plant capacity. Fourth, plant age is negatively related, meaning that older plants lose relative efficiency. Nonetheless, the behavior of the square of the age vari- able shows the presence of positive learning effects that partially offset plant aging. For example, the accumulated efficiency loss after 10 years is 17 percent, but the learning effect represents a 4.5 percent efficiency gain. Fifth, regulatory policy has had positive effects. The regression coefficients indicate that, on average, the three equations show an overall efficiency gain of 2.8 percent. Sixth, the exercise is inconclusive about whether PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 327 ownership leads to structural differences in productive efficiency. The pri- vate ownership dummy was not significant once capital input was cor- rected for capacity utilization and the assumption of constant returns to scale was relaxed (see table 6.14, column 3). This result is in line with other studies. Pollit (1995) reports statistically insignificant regression co- efficients for his ownership dummy based on a cross-sectional data set of 768 thermal power plants for 14 countries. Concluding Remarks This chapter has given an overview of the privatization program in Colombia, gathering detailed information in a comprehensive way that sets this process in context within the global economic deregulation and market promotion competition strategy. In that sense, the chapter offers for the first time a complete description of Colombia's privatization expe- rience in the 1990s and is also the first to provide empirical evidence based on an ex post performance evaluation for the privatized plants. The chap- ter has explored in-depth the cases of IFI manufacturing enterprises and power holdings. These sectors account for 95 percent of the privatization sales up to 1998, which undoubtedly make the results comprehensive in terms of the overall effects of privatization on firm performance. The study yielded several interesting results. For the case of manufac- turing, we found that IFI firms followed the cycles and trends of their pri- vate competitors across the manufacturing industry. This was proved through the study's measurement of 25 indicators of economic perfor- mance. The evolution of firm market power and profitability rates indicates that privatized firms are pricing more competitively and still adjusting to global economic deregulation and foreign competition. In particular, the decreasing trends in firms' markup after 1990 is a counter- intuitive result. Most studies on privatization show an opposite result be- cause when they were state-owned, the privatized firms were unprofitable or experiencing permanent operative deficits. The IFI firms were different because the government did not exercise control of the companies' man- agement policies. As a result, the IFI companies behaved more like their private competitors than like typical state-owned enterprises. Thus, the lower markup rates after privatization show that the markets are more competitive because of deregulation on entry and the exposure to interna- tional competition brought about by lower tariffs. The study also shows that privatization had a positive effect on firm efficiency. The analysis of the changes in performance of the efficiency indicators shows a sharp increase in labor productivity as well as a re- duction in the cost per unit. In contrast, IFI firms experienced a drastic reduction in the median of the ratios of sales and value added to capital stock. This was a direct consequence of the overinvestment problem of 328 POMBO AND RAMÍREZ the late 1970s and 1980s that hit the IFI companies harder because of their capital-intensive technologies. Privatization in this case did not boost investment. Instead, IFI companies rationalized capital spending to overcome their fall in sales per unit of capital. Nonetheless, the measurements of total factor productivity had a net positive change for these firms--the industry-adjusted indicators even showed that their overall efficiency gain was greater than their private competitors. Thus, privatization was important as a complementary mechanism that facil- itated and sped up industrial restructuring in these plants. This obser- vation is supported in econometric results in which the privatization dummy turned out to be a robust determinant of the indexes of total factor productivity. The analysis of the power sector also yielded important results. The general trends of electricity contract prices, the evolution of plant entry in thermal generation, and the increasing share of unregulated users in commercial demand suggest that the regulatory reform has been effective in promoting market competition and system efficiency. As was true for the IFI manufacturing firms, the performance analysis shows that the 1994 regulatory reform and the privatization program had a positive impact on the efficiency of electric utilities but a negative impact on their profitability. The sources of efficiency gains are explained by market competition in power generation; the reduction of nontechnical power losses, such as theft, in distribution; and the new investment in gas-based thermal technologies. The measurement of efficiency scores by thermal units re- inforces the evidence in favor of the existence of an overall gain in sys- tem efficiency and reliability. Regulatory policy has had positive effects on plant efficiency based on the econometric results. The increasing number of unregulated users has led generators to offer more competi- tive prices to secure generating contracts and thus a steady flow of busi- ness. Consistent with other studies, we found no evidence that private ownership of electricity utilities had a positive impact on efficiency. In- stead, regulation and market reform pin down the positive change in technical efficiency across thermal units. The results of the reform on utilities' profitability do not go in the ex- pected direction; instead of rising, they have fallen. This outcome is par- tially explained by the introduction of market competition in electricity generation. Competition implies that utilities have less market power, which limits their capacity to obtain extra profits through power genera- tion. In fact, the wholesale electricity market in Colombia is one of the most competitive given the number of power generators relative to the size of the market. Two additional facts are worth noting. First, power hold- ings faced financial burdens caused by an increase in their external debt service and delays in start-up in some hydro units at the beginning of the 1990s. Second, power distributors have decreased real prices to final PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 329 users. This is a consequence of a deregulated market that has introduced competition in the market of forward contracts. Finally, although privatization in Colombia has been small in size and scope, this study provides elements to conclude that the experiment was successful because it helped to consolidate entry and market competition in previously protected industries and to break up vertically integrated monopolies. Appendix 6A Table 6A.1 Infrastructure Concession Projects with Ongoing Private Investment by 1998 Project name Project name Roads Airports Armenia­Pereira­ Bogota-Second Track Manizale Barranquilla, Aereopuerto Barranquilla­Cienaga del Caribe S.A. Bogota­Facatativa Cartagena, Airport Bogota­Villavicencio Water and sanitation Buga­Tulua­Paila ACUACAR-Cartagena Carreteras Meta Triple A-Bquilla Cartagena­Barranquilla Sta Marta Cortijo­Vino Metro-Agua Espinal Neiva TIBITOC Plant for Medellin­RioNegro water treatment Patios­Guasca Bogota Water Company Sta­Marta­ Paraguanchon Railroads Atlantic Line­cargo Total road kilometers 1,388 under concession Maritime ports Total ports under concession 15 Gas pipelines Sebastopol­Medellin Telecomunicationsa Barranca­B/manga COMCEL (mixed capital) Sur­Huila CELCARIBE (mixed capital) Mariquita­Cali OCCEL (mixed capital) Huila­Tolima CELUMOVIL (private) Total length in kilometers 861 CELUMOVIL COSTA (private) under concession COCELCO (private) Note: This table excludes the power sector. Concession contracts are either build- operate-maintain-transfer or rehabilitate-operate-maintain-transfer. a. Refers to mobile phone company names and their capital structure by 1998. Source: Law 37 of 1993; DNP 1993, 1995, 1997c; Bonilla and others (2000); Alonso and others (2001). 330 Appendix 6B Table 6B.1 List of IFI Enterprises in the Sample Number Name Startup ISIC4 ISIC­name 1 Acerías Paz del Río 1947 3710 Iron and steel basic industries 2 Aicsa 1977 3845 Manufacture of aircraft 3 Alcalis-Betania 1951 3511 Basic industrial chemicals except fertilizers 4 Alcalis-Mamonal 1967 3511 Basic industrial chemicals except fertilizers 5 Astivar 1974 3841 Shipbuilding and repairing 6 Catsa Compañía Colombiana 1978 3116 Grain mill products 7 Automotriz 1974 3843 Manufacture of motor vehicles 8 Cementos Boyacá 1955 3523 Cement, lime, and plaster 9 Cementos Rioclaro 1986 3692 Cement, lime, and plaster 10 Cerromatoso 1979 3722 Recovery and founding of tin and nickel 11 Colclinker 1974 3692 Cement, lime, and plaster 12 Conastil 1969 3841 Shipbuilding and repairing 13 Empaques del Cauca 1965 3211 Spinning, weaving, and finishing textiles 14 Fatextol 1988 3220 Wearing apparel 15 Federaltex 1987 3211 Spinning, weaving, and finishing textiles 16 Ferticol 1966 3511 Fertilizers and pesticides 17 Frigopesca 1978 3114 Canning, processing of fish, shellfish 18 Icollantas 1942 3551 Tire and tube industries 19 Ingenio de Risaralda 1978 3118 Sugar refineries 20 Intelsa 1979 3832 Manufacture of radio, tv, and telecommunications equipment 21 Monomeros 1967 3512 Fertilizers and pesticides 22 Penwalt 1967 3512 Fertilizers and pesticides 23 Propal 1961 3411 Pulp, paper, and paperboard 24 Quibi 1968 3522 Manufacture of drugs and medicines 25 Simesa 1938 3710 Iron and steel basic industries 26 Sofasa 1969 3843 Manufacture of motor vehicles 27 Sucromiles 1973 3511 Basic industrial chemicals except fertilizers 28 Tejidos Unica 1953 3216 Weaving and cotton manufactures 29 Texpinal 1973 3211 Spinning, weaving, and finishing textiles 30 Vikingos de Colombia 1968 3114 Canning, processing of fish, shellfish Note: IFI Instituto de Fomento Industrial. Source: DANE, Industrial Directory; IFI, Investment Department. 331 332 POMBO AND RAMÍREZ Appendix 6C Performance Indicators for IFI Firms: Definitions and Methodology The Annual Manufacturing Survey of Colombia (Encuesta Anual Manu- facturera, or EAM) is in practice a census of medium and large enterprises in manufacturing. The EAM has undergone three methodological changes affecting the following time periods: 1970­91, 1992­93, and 1994 to date. The changes have affected the inclusion or exclusion of variables within chapters; the addition or suppression of new information across chapters; the modification of the format or variable classification criteria; and the rescaling of the sample cohorts. Some specific examples are the changes of the payroll classification, the inclusion of temporary workers after 1987, the exclusion of direct exports as a component of a firm's sales, the elimination of the direct tax variables after 1991, the redefinition of large enterprise according to number of em- ployees, and the addition of new components for fixed investment after 1992, among many others. Despite the format modifications, the survey has kept the basic vari- ables and structure across time. The database cleanup process was a two- step procedure. First, we worked with the basic variables of the 1970­91 survey. Second, all basic series were overlapped and grouped, keeping the original definitions of the older survey.35 The manufacturing survey offers the following five types of variables: · Identification variables--location, specific standard industrial classi- fication, firm's legal capital structure, and size classification. · Labor variables--wages, benefits, permanent and temporary em- ployees, administrative employees, workers, technicians, and gender sta- tistics. · Output-related variables--gross output, value added, intermediate consumption components, industrial expenditures, and inventories of fi- nal products and raw materials. · Finance-related variables--fixed asset investment, accounting depre- ciation, sales, marketing spending, paid royalties, and other general ex- penditure variables. · Consumption, generation, and sales of electricity. The survey recorded data for 133 variables from 1970 to 1991. The survey recorded 380 variables during 1992 and 1993. Since 1994 the survey has worked with 200 variables. The 1992­93 period is problem- atic because the survey included information that was not comparable with previous data. However, the core variables were recorded. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 333 Table 6C.1 The Indicators for IFI Firms in the Sample Variable Description Fixed capital stock kt kt 1(1 ) It, perpetual inventory method, series by type of IB0 depreciable assets and K0 is the initial capital k0 , where g g is the historic growth rate of the fixed assets gross investment series; is the economic depreciation rate; IB0 is gross investment at the initial date. Depreciation rates are taken from Pombo (1999b). Cost per unit The ratio of cost of sales to net sales. Cost of sales is equal to the direct expenses involved in the production of a good, including raw materials expenditures plus compensation paid to blue-collar workers. Sales are equal to the total value of products and services sold nationally and internationally. Log(sales/employees) The log of the ratio of sales to the total number of employees. Sales are equal to the total value of products and services sold nationally and internationally. Employees correspond to the total number of paid workers who depend directly on the company. Log(sales/capital The log of the ratio of sales to capital stock. stock) Sales are equal to the total value of products and services sold nationally and internationally. Capital stock series follow the perpetual inventory method, defined above. Value added/capital Partial capital productivity is the ratio of value added to capital stock. Capital stock series follow the perpetual inventory method, defined above. Value added/labor Partial labor productivity is the ratio of value added to labor. Capital stock series follow the perpetual inventory method, defined above. Labor is the total number of permanent employees. Value added/workers Partial labor productivity. Workers stand for the number of blue-collar workers by firm. Total factor TFP decomposition following the methodology in productivity (TFP) Jorgenson and Grilliches (1967), using a translog technology. Operating Operating income is equal to sales minus operating income/sales expenses, the cost of sales, and depreciation. Sales are equal to the total value of products and services sold nationally and internationally. (Table continues on the following page.) 334 POMBO AND RAMÍREZ Table 6C.1 (continued) Variable Description Net income/sales Net income is equal to operating income minus interest expenses and net taxes paid. Sales are equal to the total value of products and services sold nationally and internationally. Operating income/ The ratio of operating income to capital capital stock stock. Operating income is equal to sales minus operating expenses, the cost of sales, and depreciation. Capital stock series follow perpetual inventory method. VAi Wi Gross margin rate The ratio of value added minus VAi wages over value added. Lerner index i The ratio of firm's market share over demand elasticity. Capital intensity ratio 1 Ratio of capital over permanent employment. Capital intensity ratio 2 Ratio of capital over blue-collar workers. Plant size Ratio of total workers of firm i in sector j over plant average workers in industry j. Permanent Ratio of permanent blue-collar workers of workers/average firm i in sector j over average permanent permanent workers blue-collar workers in industry j. Yij Output scale Ratio of output of firm i in sector indicator Yj j over average plant output in sector j. Ii Gross investment rate Ratio of investment of firm i over Yi output in firm i. Machinery and The ratio of new investment of firm i in equipment investment machinery and equipment to the total rate/output output of firm i. Machinery and The ratio of new investment of firm i in equipment investment machinery and equipment to the total rate/total investment output of investment of firm i. Real wages per Real average wages paid per worker by worker firm. The consumer price index was used as deflator to calculate real wages at constant pesos. The monthly average PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 335 Table 6C.1 (continued) Variable Description official exchange rate of 1995 is used to convert series to constant prices in US$. A similar procedure is used for the calculation of real wages per blue-collar and per white-collar workers. Industry-adjusted real Industry control group is the four-digit wages per worker standard industrial classification to which a specific firm belongs. For each year and firm we compute the industry-adjusted indicator by taking the ratio of the value of the indicator for the IFI firm to its industry control group. A similar procedure is used for blue-collar and white-collar workers. Total wages/sales The ratio of total wages paid by the firm to the sales of the firm. Total wages is equal to the total wage bill paid to all workers in payroll. Sales are equal to the total value of products and services sold nationally and internationally. Wage rate The ratio of total wages paid by the firm to the firm's value added. Total wages is equal to the total wage bill paid to all workers in payroll. Value added is equal to the firm's gross output minus intermediate consumption. Compensation rate The ratio of total compensation paid by the firm to the firm's value added. Total compensation is equal to wages plus social benefits paid to workers in payroll. Value added is equal to the firm's gross output minus intermediate consumption. CR4 entropy index The output of the four largest plants of industry j over industry j output. n Herfindal index a s2ij Summation of market shares (s) of i 1 firm i in industry j. Imported raw A measure of quality in intermediate material/domestic consumption, where imported raw materials raw materials stand for the raw materials used in a production process whose country of origin is not Colombia. Domestic raw materials are raw materials produced in Colombia. (Table continues on the following page.) 336 POMBO AND RAMÍREZ Table 6C.1 (continued) Variable Description Human capital Ratio of technicians over workers. Advertising rate Ratio of advertising expenditures over value added. Licensing indicator Ratio of paid royalties on value added. Intraindustry trade Grubel and Lloyd index of intraindustry trade coefficient by industry. Gross output For the manufacturing industry, gross output is equal to the total of product sales plus sales of all self-produced electricity plus income for industrial services plus sales of all nontrans- formed goods. Intermediate For manufacturing, intermediate consumption is consumption the combination of raw materials plus industrial expenses plus purchases of electricity. Value added Gross output minus intermediate consumption. Note: IFI Instituto de Fomento Industrial. Appendix 6D The Power Sector Indicators and Data Sets The performance indicators for the power sector are based on the finan- cial statements. These indicators follow the definition and methodology of the privatization studies of Megginson, Nash, and van Randenborgh (1994) and La Porta and López-de-Silanes (1999). Most of the variables are defined in appendix table 6C.1. The cost per unit indicator is expressed in pesos per kilowatt at 1995 prices. Output is in gigawatts. The Data Sets Currently, the power sector statistics in Colombia are split among the the National Grid Company (Interconexión Eléctrica S.A, or ISA); the Mining and Energy Planning Unit; the Electricity and Gas Regulatory Commission; the National Planning Department; and the Superinten- dent of Domiciliary Public Services (SSPD). As a result, each source has a different format and contents. The information is sorted by plant, utilities, regional electricity mar- kets, and geographical regions, or at a countrywide aggregate level. Table 6D.1 describes the contents of the collected data sets. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 337 Table 6D.1 Colombia: Power Sector Statistics and Description of the Data Sets Data sources Contents ISA Reports Operative reports of the National (1995­99) Interconnected System Hydrology Grid constraints Generation Demand Available effective capacity The Electricity Spot Market Report Pool's prices and contacts Total traded amount (gigawatts) Pool's marginal supply prices by type of generation SIVICO The following data are available by utility 1997­99 level: Source: SSPD Financial statements Income statement Balance sheet Labor statistics Number of employees by sector's activity Number of employees by occupational category Number of employees by type of generation Market composition by type of user Consumption Invoicing Number of subscribers Average tariffs by users Results and performance control process indicators Quality service indicators Spending and indebtedness indicators SIEE The Energy and Economic Information 1970­98 System (SIEE) is a data set covering the Source: Organización Latin American economies' energy-related Latinoamericana de Energía statistics. The SIEE sections are Prices Demand and supply Energy-related equipment Environmental impact (Table continues on the following page.) 338 POMBO AND RAMÍREZ Table 6D.1 (continued) Data sources Contents Economic energy indicators Worldwide energy statistics FEN The power sector historical financial data 1983­94 compiled by the Financiera Electrica Source: FEN Nacional (FEN). The database offers a summary by power company of Income statements Balance sheets Other variables: purchase sales of bulk electricity; available capacity; power losses SINSE The power sector national system is a 1970­94 comprehensive database. The data are Source: MME available by utility and regional market. The SINSE chapters are Energy balances Generation and electricity demand Number and type of subscribers Average tariffs by users In addition to the data sets, we made direct requests to ISA for the monthly indicators of the Mercado de Energia Mayorista, starting in July 1995, and the Thermal Park Data Set. The crossing of information among ISA's thermal park data set, SIVICO, and SINSE allowed us to collect the input-output variables by thermal unit that are depicted in appendix table 6D.2 To make direct inferences of labor input by plant after 1996, a sur- vey was carried out among the members of the Colombian Generators Association. The collected information allowed for distinguishing bench- marks of capacity-to-labor ratios; under normal assumptions of Leontief technology of noninput substitution, that coefficient turns out a constant parameter. The data provided by the power utilities along with SIVICO allowed us to identify the number of employees by thermal plants for the period 1996­99, given the reported capacity per unit. The estimated benchmark labor-to-capacity ratios by occupational category for a base- technology thermal plant were 0.036597 (directors), 0.151852 (adminis- trative), and 0.527731 (operative). For the 1988­94 period, the FEN books recorded some physical vari- ables per power utility, among them the permanent employment series. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 339 Table 6D.2 Thermal Plants: Input and Output Variables Sample Variables 1988­1994 Generation (gigawatt hours) Gross capacity (megawatts) Net capacity (megawatts) Coal (tons) Fuel oil (gallons) Diesel oil (gallons) Gas (cubic feet) 1995­1999 Generation (gigawatt hours) Effective capacity (megawatts) Labor (number of employees)a Heat rate Note: Labor information is recorded by power utility and industry activity: generation, transmission, and distribution (SIVICO). a. Since 1996. Source: SINSE, ISA, SIVICO. Thus, the inference of labor series by the thermal units followed a constant distributing capacity assumption, that is: Thermal Unit Labor (L1) (Max Theoretical Thermal Plant Unit Capac- ity in GWh/Utility Available Capacity in GWh) * Total Permanent Utility Employees Other formulas were used to generate alternative labor series by thermal plants. One was based on power generation: Thermal Unit Labor (L2) (Thermal Plant Generation in GWh/Utility Available Capacity in GWh) * Total Permanent Utility Employees Then an adjusted L2 series was generated under the assumption: L L MgPhydro a b avg ; MW thermal a MW b hydro / (1 x); where: x MgPthermal Rationing Price: MgPhydro MgPthermal 1.8; Without Rationing: MgPhydro MgPthermal 0.6, where MgP marginal price, MW megawatts of installed capacity. The above coefficients are observed parameters. L1 and L2 were used as the labor input series in the estimation of plant efficiency scores. 340 Appendix 6E Table 6E.1 DEA Efficiency Scores in Thermal Generation before and after the Regulatory Reform Decision- making Startup Capacity Owner- Score Score Score1 Score1 Relative Relative unit Plant name year megawatts ship before after before after effic. effic.1 1 Barranca1 1982 13 Public 0.7859 0.5932 0.7859 0.7939 2 Barranca2 1982 13 Public 0.7203 0.5932 0.7448 0.7702 3 Barranca3 1972 66 Public 0.8798 0.6404 0.8798 0.8211 4 Barranca4 1983 32 Public 0.6625 0.6118 0.6625 0.8110 5 Barranca5 1983 21 Public 0.7023 0.6176 0.7023 0.8217 6 Bquilla1 1980 58 Public 0.9211 . 0.9139 . 7 Bquilla3 1980 66 Private 1.0000 0.6624 1.0000 0.7156 8 Bquilla4 1980 69 Private 0.9699 0.7439 0.9803 1.0000 9 Cartagena1 1980 66 Private 0.8677 0.6428 1.0000 0.7447 10 Cartagena2 1980 54 Private 0.7932 0.6515 0.7437 0.8274 11 Cartagena3 1980 67 Private 0.8712 0.6815 0.8603 0.8245 12 Chinu4 1982 14 Public 0.4242 . 0.7097 . 13 Cospique1 1960 4 Public 0.9086 . 1.0000 . 14 Cospique2 1960 4 Public 0.7277 . 1.0000 . 15 Cospique3 1967 8 Public 1.0000 . 0.9722 . 16 Cospique4 1966 9 Public 1.0000 . 0.7791 . 17 Cospique5 1965 12 Public 0.4487 . 0.8584 . 18 Flores1 1993 152 Private 0.9881 1.0000 0.9881 1.0000 19 Guajira1 1987 160 Public 1.0000 0.8563 1.0000 0.7743 20 Guajira2 1987 160 Public 1.0000 0.8374 1.0000 0.8915 21 Paipa 1 1963 31 Public 0.4048 0.4977 0.3208 0.8859 22 Paipa 2 1975 74 Public 0.7307 0.3794 0.4755 0.7891 23 Paipa3 1978 74 Public 0.6331 0.4154 0.3874 0.7735 24 Palenque 3-4 1972 15 Public 0.8780 0.4586 1.0000 0.8011 25 Palenque5 1985 21 Public 0.6706 . 0.6706 . . . 26 Proeléctrica1 1993 46 Private 0.9993 0.9695 0.9993 0.8857 27 Proeléctrica2 1993 46 Private 1.0000 0.9695 1.0000 0.9654 28 Tasajero 1985 163 Private 1.0000 0.6755 1.0000 0.8241 29 Tibú1 1965 6 Public 0.1669 . 0.3157 30 Tibú2 1965 6 Public 0.1632 . 0.8026 31 Zipa2-3 1976 104 Mixed 0.4904 0.8888 0.4213 0.6721 32 ZIPA3 1976 66 Mixed . 0.2235 . 0.8021 . . 33 Zipa4 1981 66 Mixed 0.4626 0.1879 0.4601 0.6797 34 Zipa5 1985 66 Mixed 0.2692 0.3213 0.3042 0.8655 35 Flores2 1996 100 Private . 0.9199 . 0.9205 36 Flores3 1998 152 Private . 1.0000 . 1.0000 37 Merilectrica 1998 157 Private . 0.7887 . 0.9273 38 TebsaB1 1998 768 Private . 1.0000 . 0.9141 39 Termocentro1 1997 99 Public . 0.9160 . 1.0000 40 Dorada1 1997 52 Public . 0.2554 . 0.8010 41 Sierra1 1998 150 Public . 0.1442 . 0.8564 (Table continues on the following page.) 341 342 Table 6E.1 (continued) Decision- making Startup Capacity Owner- Score Score Score1 Score1 Relative Relative unit Plant name year megawatts ship before after before after effic. effic.1 42 Termovalle1 1998 214 Private . 0.8237 . 0.8858 Total decrease 19 12 (plants) Share capacity 35.3 24.6 (percent) . No data are available. Note: This table shows the results of the efficiency frontier measurement on 42 thermal plants that were active as marginal producers before and after the 1995 regulatory reform. The first three columns depict the plant name, startup year, and plant capacity in megawatts. All thermal units before the reform belonged to one of the five electric holdings in the country as described in text. The fourth column indicates the ownership status by the year 2000, that is, whether the thermal unit belonged to a public, private, or a mixed-capital electric utility. The next four columns describe the efficiency scores before and after the 1994 reform. The table presents two types of scores. The first one is the variable Score, which uses megawatts of capacity as total capital input. The second one is the variable Score1, in which capacity is adjusted by its effective utilization, and this is the definition of capital input used in these estimations. The reason to make such an adjustment is that most thermal units are marginal producers. In Colombia the base system is hydro. The efficiency measures assume constant return to scale assumption, that is, the value of total output is equal to total input spending and therefore the sum of inputs weights is equal to 1. Source: Authors' estimations based on EMS 1.3 software written by Scheel (2000). PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 343 Notes The authors wish to give special thanks to the following people and institutions: Eduardo Granados (DANE), Francisco Ochoa (ACOLGEN), Adriana Calderon (IFI), Gerson Castaneda (SSPD), Alberto Jose Uribe (Banco de la República), Ferney Niño (Banco de la República), and Diana Espinoza (ECOPETROL). The authors are grateful for the comments of Florencio López-de-Silanes, Alberto Chong, and other IDB seminar participants; Claude Crampes and the electricity workshop participants at the Institut d'Economie Industrialle, Toulouse; and Luis Eduardo Fajardo at the Universidad del Rosario. Rodrigo Taborda provided su- perb research assistance. Financial support from the IDB's Latin American Re- search Network is gratefully acknowledged. 1. The general objectives and the scope of the economic openness program are in the 1990­94 Development Plan (DNP 1991a). The main institutional reforms are embodied in the following laws and CONPES (National Council for Economic and Social Policy) documents: foreign control regime (Law 9/1991), foreign trade reform (Law 7/1991), financial reform (Law 45/1990), new statute of foreign in- vestment (CONPES document, January 22/1991), labor reform (Law 50/1990), and privatization of maritime ports (Law 1/1991). See DNP (1991b). 2. One example is the study by Spiller and Guasch (1998) on the regulatory process in Latin America, in which they skip over the Colombian experience de- spite the country's advances in public utilities regulation. Furthermore, in the col- lective studies of privatization in Latin America, such as those by Glade (1996), Baer and Conroy (1994), and Baer and Birch (1994), references to Colombia are usually limited, in contrast to other Latin American countries. 3. The CONPES documents are 2648 (DNP 1993); 2775 (DNP 1995); and 2929 (DNP 1997c). Law 37 of 1993 deals with concessions contracts for telecom- munications; Law 142 of 1994, the residential public services reform; Law 143 of 1994, the power sector reform; and Law 226 of 1995, the privatization transfers. 4. For details, see Alonso and others (2001) and DNP 1993, 1995, 1997b, and 1997c. 5. The schedule for public divestiture of public and mixed-capital enterprises and public financial institutions was laid out in CONPES documents 2378 (DNP 1988) and 2648 (DNP 1993). 6. All dollar values are in U.S. dollars unless otherwise indicated. 7. For details, see Megginson, Nash, and van Randenborgh (1994). 8. It is important to highlight that the IFI made major transfers of assets to the private sector before the privatization program of the 1990s. One example was Icollantas, in which the IFI sold its equity shares in 1980 and 1985. In 1994 the IFI participated in a 20 percent share of the company's capitalization of $60 million. 9. "Special legal procedures" were applied to companies with property shares from two or more public institutions, companies with direct investments from for- eign government agencies, and companies with ongoing settlement processes with their lenders. For details, see DNP (1988). 10. For instance, by March 1993 ECOPETROL had equity shares in three do- mestic investment banks (Corficaldas, Corfinorte, Corfinanza), one power utility (ESSA), one fertilizer plant (FERTICOL), and one craft enterprise (Artesanias de Colombia). For details, see DNP (1993). 11. For details, see ECOPETROL's press release of June 6, 1997. According to that bulletin there was tight competition among the winner (Gas Natural of Spain) and British Petroleum, Amoco, Empresas Públicas de Medellín, and France Gas. 12. These cases were TERPEL CENTRO and TERPEL SUR, which sold 60 per- cent and 38 percent of their shares, respectively. 344 POMBO AND RAMÍREZ 13. The national grid company Interconexión Eléctrica S.A. (ISA) was founded in 1967. By that time, the sectoral development view was to consolidate ISA as the largest nationwide power generator and transporter of bulk electricity following the vertically integrated natural monopoly model. For more details, see World Bank (1991). A complete description of the regulatory reform in Colombia's power sector is in Pombo (2001b) and ISA reports. Historically, Colombia's power sector has been divided into five regional markets: Bogotá Power Company (EEB), the At- lantic Coast Regional Electric Corporation (CORELCA), Public Enterprises of Medellín (EPM), Public Enterprises of Cali and the Cauca Valley Corporation (EMCALI and CVC), and the Colombian Power Institute (ICEL). Only three of the five regional power distribution networks have been privatized, leaving 70 percent of the distribution network in the National Interconnected System in public own- ership. Hence, privatization and entry competition remain a pending and unfin- ished task for local distribution. 14. The power market in Colombia parallels the British pool of the early 1990s. For more details on electricity markets and the British experience, see Armstrong, Cowan, and Vickers (1994) and Newbery (2000). For Latin America, good reviews are found in Spiller and Guasch (1998) and the Inter-American Development Bank's 2001 annual report. 15. Although the survey is conducted annually, dissemination of the informa- tion at the plant level is restricted to protect the sources' identities. We were able to access the plant level data thanks to a technical cooperation agreement between the Universidad del Rosario in Bogotá and DANE. All the information was processed at DANE's headquarters. 16. See appendix table 6B.2 for a list of the companies in the date set and ap- pendix 6C for a description of Colombia's Annual Manufacturing Survey. 17. Again we want to point out that although the use of nontraditional indica- tors is not that common within the privatization or financial economics literature, these indicators are widely used in industrial economics. 18. Ocampo (1994) presents a comprehensive analysis of the trade policy and industrialization in Colombia for the 1967­91 period. 19. Firm i's gross margin is equal to GMi (VAi Wi) /Wi, where VA is the firm's value added and W denotes wages. 20. For details of the reforms of the 1990s, see Montenegro (1995). 21. See La Porta and López-de-Silanes (1999) for more details about the Mex- ican privatization program. For more details on Chile's privatization process, see, for example, Maloney (1994); Hachette, Lüders, and Tagle (1993); and the study on Chile in this volume. 22. Partial productivities are expressed in U.S. dollars at 1995 prices. Thus, av- erage labor productivity per worker before privatization was $30,487, and average productivity per unit of capital stock was $6.02. 23. Garay's (1998) study presents a demand-side growth decomposition exer- cise for Colombian manufacturing. The main result is that the contribution of import-substituting industrialization to manufacturing growth has not been posi- tive since the mid-1970s. For the case of Mexico, La Porta and López-de-Silanes (1999) found that the mean (median) ratio of investment to sales of privatized firms is 3.58 (5.80) percentage points lower than the control group levels after privati- zation. In the study of Mexico, the adjusted indicators are differences rather than ratios, which is the methodology used in this study. 24. Unfortunately, we do not have detailed discrete information across IFI firms regarding plant restructuring such as firms' labor training programs, adjust- ments in plant automation, administrative adjustments, and reengineering in processes and products. What we do have, from some internal IFI documents and special publications sponsored by IFI (1987) and MDE (1995), is partial informa- tion regarding the technological history for some companies. PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 345 25. See Pombo (2001b) for more details. The point here is that there are two sources of power losses. One is the technical loss due to the power losses in transmis- sion necessary to maintain the system's stability. The nontechnical loss is the difference between real consumption and invoicing. Cities such as Bogotá used to have power stealing, illegal connections, and adulterated meters, among other irregularities. 26. See Pombo (2001b) for an analysis of the 1992 blackout. The official ver- sion of the blackout causes and policy measures is in the 1993 Ministry of Mining report to the Congress. 27. Effective capacity refers to all plants generating at full capacity. Available capacity does not take into account units that are shut down for maintenance and the fraction of power that is used to stabilize the electric system, such as the load factor and voltage. 28. See Pombo (2001b) for more details. 29. One kilowatt hour (kWh) 3,412.1 Btus; 1 gigawatt hour 0.86 thermal calories; 1 megawatt of capacity 1,000 kWh. The literature of DEA as well as its applications is extensive. Good introductions can be found in Coelli, Prasada Rao, and Battesse (1998) and Thanassoulis (2002). 30. See Pombo (2001a) for a specific study on intraindustry trade and technol- ogy applied to the case of Colombia. 31. This idea is similar to the competition behind contestability in which firms apply the hit-and-run strategy in order to capture profits. However, in this case there are significant sunk costs. For theoretical details, see Baumol (1982) and Baumol, Panzar, and Willig (1988). 32. For details, see Pombo (1999b). 33. Notice that the possible simultaneity bias that arises from running TFP against value added growth is partially avoided here because value added growth refers to the overall industry-specific group. 34. In general, the variables included in the Tobit regressions are robust. Resid- uals are homoskedastic according to the reported OLS tests. The residuals are not normal, which is associated with the distribution Kurtosis. The distribution of the residuals is symmetric. 35. The main problem with the methodological changes was the modification in the basic plant identification (ID) variable from 1991 to 1992 and 1993. This is troublesome if one wants to track the information at plant level. We ran a cross- matching program throughout plant commercial names, recorded at the industrial directories, and generated an identification key for the ID variables in the 1991­92 and 1992­93 surveys. References Alonso, Juan, Juan Benavides, Israel Fainboim, and Carlos Rodríguez. 2001. "Par- ticipación privada en proyectos de infraestructura y determinantes de los es- quemas contractuales: El caso colombiano." Latin American Research Network Working Paper R-412. Inter-American Development Bank, Research Depart- ment, Washington, D.C. Armstrong Mark, Steven Cowan, and John Vickers. 1994. Regulatory Reform: Economic Analysis and British Experience. Cambridge, Mass.: MIT Press. Baer, Werner, and Michelle Birch, eds. 1994. Privatization in Latin America. West- port, Conn.: Praeger. Baer, Werner, and Michael Conroy. 1994. "Latin America: Privatization, Property Rights, and Deregulation." Quarterly Review of Economics and Finance Spe- cial Issue 34. December. 346 POMBO AND RAMÍREZ Barro, Robert, and Xavier Sala-i-Martin. 1995. Economic Growth. New York: McGraw Hill. Baumol, William. 1982. "Contestable Markets: An Uprising in the Theory of In- dustry Structure." American Economic Review 72(1): 1­15. Baumol, William, John Panzar, and Robert D. Willig. 1988. Contestable Markets and the Theory of Industrial Structure. Rev. ed. San Diego: Harcourt Brace Jovanovich. Bonilla, J., V. Cantillo, A. Escudero, C. González, and A. Herrera. 2000. "Los ser- vicios públicos en las ciudades puerto del Caribe colombiano." Universidad del Norte, Barranquilla, Colombia. Coelli, Tim, D. S. Prasada Rao, and George Battesse. 1998. An Introduction to Ef- ficiency and Productivity Analysis. London: Kluwer Academic Publishers. Dager, Vanessa. 1999. "Proceso de venta de empresa del IFI." Departamento In- geniería Industrial, Universidad de los Andes, Bogotá. DANE (Departamento Administrativo Nacional de Estadística). Various years. "Annual Manufacturing Survey." Bogotá. DNP (Departamento Nacional de Planeación). 1988. Venta de las empresas del IFI. Documento CONPES 2378, Bogotá, Colombia: Departamento Nacional de Planeación. ------. 1991a. La revolución pacífica: Plan Nacional de Desarrollo 1990­1994. Bogotá. ------. 1991b. Modernización y apertura de la economía, vol. 1. Bogotá. ------. 1993. Nuevos espacios para la inversión privada en Colombia. Documento CONPES 2648. Bogotá. ------. 1995. Participación del sector privado en infraestructura física. Docu- mento CONPES 2775. Bogotá. -------1997a. Asignación de utilidades de la nación y saneamiento de las empre- sas industriales y comerciales del sector eléctrico. Documento CONPES 2923. Bogotá. ------. 1997b. Balance de los procesos de vinculación de capital privado: Las pri- vatizaciones. Documento CONPES 2929. Bogotá. ------. 1997c. La participación privada en infraestructura: Seguimiento. Docu- mento CONPES 2928. Bogotá. Dixon, R., and P. Thirlwall. 1975. "A Model of Regional Growth Rate: Differ- ences on Kaldorian Lines." Oxford Economic Papers 27 (2): 201­14. Garay, Luis Jorge. 1998. "Colombia: estructura industrial e industrialización: 1967­1996." Departamento Nacional de Planeación, Colciencias, Bogotá. Glade, William., with Rosanna Corona, eds. 1996. Bigger Economies, Smaller Governments: Privatization in Latin America. Boulder, Colo.: Westview Press in cooperation with Institute of the Americas. Grubel, Herbert, and Peter Lloyd. 1975. Intraindustry Trade: The Theory and Measurement of International Trade in Differentiated Goods. London: Macmillan. Gutierrez, Luis, and Sanford Berg. 1999. "Telecommunications Liberalization and Regulatory Governance: Lessons from Latin America." Telecommunications Policy 24: 865­84. Hachette, Dominique, Rolf Lüders, and Guillermo Tagle. 1993. "Five Cases of Pri- vatization in Chile." In Rosanna Corona and Manuel Sánchez, eds., Privatization PRIVATIZATION IN COLOMBIA: A PLANT PERFORMANCE ANALYSIS 347 in Latin America. Washington, D.C.: Johns Hopkins University Press for the Inter-American Development Bank. IFI (Instituto de Fomento Industrial). 1987. Aporte al progreso de una nación. Bogotá. ISA (Interconexión Eléctrica S.A.). 1998, 1999. Informes de gestión. On CD. Medellín. Jorgenson, Dale W., and Zvi Grilliches. 1967. "The Explanation of Productivity Change." Review of Economic Studies 34(3): 249­80. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." Quarterly Journal of Economics 114: 1193­242. Maloney, William. 1994. "Privatization with Share Diffusion: Popular Capitalism in Chile." In Werner Baer and Michelle Birch, eds., Privatization in Latin Amer- ica: New Roles for the Public and Private Sectors. Westport, Conn.: Praeger. Mann, H. B., and D. R. Whitney. 1947. "On a Test of Whether One of Two Ran- dom Variables Is Stochastically Larger than the Other." Annals of Mathemati- cal Statistics 18: 50­60. MDE (Ministerio de Desarrollo Económico). 1995. Instituto de Fomento Indus- trial: 1940­1995. Bogotá. Megginson, William, and Jeffry Netter. 2001. "From State to Market: A Survey of Empirical Studies on Privatization." Journal of Economic Literature 39(2): 321­89. Megginson, William, Robert Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An Interna- tional Analysis." The Journal of Finance 49(2): 403­52. MME (Ministerio de Minas y Energía). 1996, 1998. Memorias al Congreso Na- cional. Bogotá. Montenegro, Armando. 1994. "El sector privado y la reforma del estado." Revista Planeación y Desarrollo 24(2): 153­66. ------. 1995. "Economic Reforms in Colombia: Regulation and Deregulation 1990­1994." EDI Working Paper 95-04. World Bank, Economic Development Institute, Washington, D.C. Newbery, David. 2000. Privatization, Restructuring, and Regulation of Network Utilities. Cambridge, Mass.: MIT Press. Ocampo, Jose Antonio. 1994. "Trade Policy and Industrialization in Colombia, 1967­1991." In G. Helleiner, ed., Trade Policy and Industrialization in Turbu- lent Times. London: Routledge. Pollit, Michael. 1995. Ownership and Performance in Electric Utilities. Oxford, U.K.: Oxford University Press. Pombo, Carlos. 1999a. "Economías de escala, markups, y determinantes del cam- bio técnico en la industria en Colombia." Coyuntura Económica 29(4): 107­32. ------. 1999b "Productividad industrial en Colombia: una aplicación de números índices," Revista de Economía del Rosario 2(1): 107­39. ------. 2001a. "Intraindustry Trade and Innovation: An Empirical Study of the Colombian Manufacturing Industry." International Review of Applied Eco- nomics 15(1): 77­106. ------. 2001b. "Regulatory Reform in Colombia's Electric Utilities." Quarterly Review of Economics and Finance 41(5): 683­711. 348 POMBO AND RAMÍREZ Romer, Paul. 1990. "Endogenous Technical Change." Journal of Political Econ- omy 98: 71­102. Scheel, Holger. 2000. "EMS: Efficiency Measurement System User's Manual." Available at http://www.wiso.uni-dortmund.de/lsfg/or/scheel/ems/. Spiller, Pablo, and José Guasch. 1998. "Managing the Regulatory Process: Design, Concepts, Issues and the Latin American and the Caribbean Story." World Bank, Washington, D.C. Sprent, Peter, and Nigel Smeeton. 2001. Applied Non-Parametric Statistical Meth- ods, 3d ed. New York: Chapman and Hall. Thanassoulis, Emmanuel. 2002. Introduction to the Theory on Application of Data Envelope Analysis. London: Kluwer Academic Publishers. World Bank. 1991. "Colombia: The Power Sector and the World Bank 1970­1987." In Comision Nacional de Energía, Evaluación del Sector Eléctrico Colombiano. Bogotá: Ministerio de Minas. Zuleta, L. H., L. Jaramillo, E. Ballen, and A. M. Gomez. 1993. "Privatization in Colombia: Experiences and Prospects." In Manuel Sanchez and Rosanna Corona, eds., Privatization in Latin America. Washington, D.C.: Johns Hop- kins University Press for the Inter-American Development Bank. 7 Privatization in Mexico Alberto Chong and Florencio López-de-Silanes WHAT SHOULD PRIVATIZATION ATTEMPT TO achieve? Why does it work? How does it affect firm performance? What are the determinants of privatization prices? What are the effects in terms of fiscal policy? How should a privati- zation program be structured? What role should deregulation or reregula- tion have in a privatization policy, and why? In this chapter, we provide empirical answers to these questions through a micro- and macroeconomic analysis of the Mexican privatization program. The Mexican program, car- ried out since the mid-1980s, has been one of the largest in the world in its scale and scope. It implied the reversal of 40 years of state interventionism and profoundly transformed the economic landscape of Mexico. Recently, politicians and the media have harshly questioned the effect of privatization, claiming that benefits are greatly exaggerated and that privatization invariably leads to welfare losses for society. This chapter's key finding is that privatization leads to dramatic improvements in firm performance and that those improvements result from efficiency gains, not from transfers from workers or exploitation of consumers. We find that the operating-income-to-sales ratio for the median firm increased 24 per- centage points, and by assuming that quality is unchanged and that fired workers have zero productivity, we estimate that price increases and trans- fers from workers can account for at most 5 percent and 31 percent of the increase in profitability, respectively. Similar improvements are observed for operating efficiency and output. Furthermore, we calculate industry- adjusted indicators to ensure that the results obtained are directly attrib- utable to the transfer of control to the private sector and not a reflection of general macroeconomic trends or a sector-specific phenomenon. Even when this initial objection is overcome, many argue that privatiza- tion should be opposed because benefits accrue to the new owners only, while consumers and the poor are left behind. It is claimed that the sale of state-owned enterprises (SOEs) amounts to a privatization of benefits while 349 350 CHONG AND LÓPEZ-DE-SILANES the government is left to foot the bill if things go wrong. Contrary to this belief, we find that there are significant social benefits to privatization, par- ticularly in greater access to goods and services. It is precisely the poor who are usually left out of the telephone, electricity, and water networks when they are public and therefore it is the poor who have the most to gain from increased coverage. Moreover, although it is true that the government has sometimes footed the bill for botched privatizations, an often-overlooked aspect is its beneficial fiscal effect. Lower subsidies, increased taxes, and of course, the direct revenues from the sale provided Mexico with the rev- enues needed to dramatically reduce its stock of external debt and increase spending on education and poverty alleviation programs. Among those who accept that privatization is a positive force, there is a widespread belief that SOEs need extensive restructuring and nurturing before they are fit for sale. With this in mind, governments often pour vast sums of money and spend valuable time implementing investment and efficiency plans which are not valued by bidders. In the case of Mexico, the presence of an efficiency restructuring program or an investment plan reduced the net privatization price received by the government by 56 percent and 95 percent, respectively. Furthermore, each additional month taken to complete the privatization decreased prices by 2.2 percent. This evidence clearly suggests that governments should think twice before engaging in these types of restructuring programs and that often the best solution is simply to concentrate on the sale of SOEs and leave the restructuring to the market. If politicians wish to help displaced workers, there is sure to be a more cost-effective way to do it than by trying to fix public firms before they are sold. An additional lesson that can be drawn from the Mexican experience is that simplicity and transparency are para- mount to a successful program. Special requirements such as cash-only sales reduced net prices by 30 percent, while allowing foreign participation boosted prices by 32 percent, and the presence of an additional bidder in the final round increased them by 17 percent. Finally, this chapter attempts to make sense of the cases of failed pri- vatization by analyzing the importance of a set of complementary policies such as the design of the privatization contracts, deregulation and reregu- lation of privatized firms, and the corporate governance framework. We find that the main instances of botched privatization can be traced back to mistakes in one or more of these complementary policies. Although more attention is clearly needed in these areas, the good news is that the failures seen so far seem to have a readily available solution. The Mexican SOEs and Privatization To understand the opportunities presented by the Mexican privatization program, it is helpful to analyze first the role of the government in the PRIVATIZATION IN MEXICO 351 economy and the motives behind this role. The 1917 Mexican constitution established the general jurisdictional framework under which the role of the state in the economy was defined. From this foundation, the Mexican government launched a gradual takeover of the economy, and by 1982, the year in which banks were nationalized, the government controlled more than 1,100 firms in all sectors of the economy, including having monopolistic control of strategic areas such as energy and infrastructure. Since then, government ownership of enterprises has declined precipi- tously and is now significant only in some entrenched sectors, most notably oil and electricity, that have successfully used their political clout to resist divestiture. This section analyzes the rise of the SOE sector and the subsequent privatization program in more detail. The Growth of the SOE Sector Table 7.1 shows the main focus of state expansion for different periods during 1917­2003, as well as the number of SOEs at the end of each Table 7.1 State-Owned Enterprises in Mexico, 1917­2003 Number of state- owned enterprises Main focus of state activity Period (end of period) Public administration, creation of 1917­40 36 infrastructure, administration of natural resources, and provision of basic services Import-substitution-oriented investments 1941­54 144 (capital-intensive and long-maturity areas; industry input suppliers); transportation and communications; and social security institutions Stable development, unplanned expansion: 1955­70 272 regional development, production expansion, and creation of employment Planned expansion: oil bonanza, government 1971­75 504 as an industrial investment engine Planned expansion: bank nationalization, 1976­82 1,155 government investment in strategic areas, and takeover of firms in distress Main program of liberalization of the 1983­93 258 economy and divestiture of the state- owned sector Consolidation of the privatization program: 1994­2003 210 public utilities and pension system Source: Aspe 1993; Presidencia de la República, 1982­2003. 352 CHONG AND LÓPEZ-DE-SILANES period. During the 1920s government operations mostly reflected attempts to regulate the economy; the central bank (Banco de Mexico) was established to control monetary policy, and investments in infrastructure were used to stimulate the economy.1 The Great Depression in the United States and the worldwide economic crises that accompanied it led the Mexican government to take direct responsibility for the provision of many basic services. In 1934 the "Cardenista" era began with the im- plementation of a six-year plan that emphasized the role of agricultural development and the provision of basic services. Predictably, this approach quickly spawned an array of funding institutions that would outlive their initial mandate and become engines for state expansion.2 Pemex, the national oil company and probably the most prominent SOE in Mexican history, was created during this period and remains in state hands today. By 1940 the government owned 36 enterprises, and the stage was set for a massive expansion of the state-owned sector. The period from 1940 to the mid-1950s witnessed the state's implementation of an import- substitution model and the undertaking of capital-intensive and long- maturity investments such as steel mills, coal mines, paper mills, and oil refineries. The government also took over the social security system dur- ing this period and created two institutions to run it: the National Institute for Social Security (to manage private sector pensions) and the Social Security of Government Employees. Until 1947 each company was almost completely responsible for its own operations, and the whole SOE sector was managed in a decentralized fashion. In that year, however, as the size of government operations increased, centralized control was established under two ministries.3 The late 1950s and 1960s were known as a period of stable develop- ment, during which the economy grew swiftly and the government expanded its influence in a seemingly random way. The number of SOEs more than doubled, and public ownership of firms expanded to new sec- tors including sugar cane mills and the manufacture of textiles, tobacco, and food processing. The state's major investments were directed toward regional development with the aim of increasing employment and production across the nation. Supervision became more complex as the number of SOEs increased, and in 1970 control of these firms was centralized under the control of three ministries: the Ministry of Finance (SHCP), the Secretariat of the Presidency (SP), and the Secretariat of National Patrimony (SENEPAL). Starting in the first half of the 1970s, a fall in private investment and stricter restrictions imposed by the government consolidated the role of the public sector as the main engine of investment. The government bor- rowed heavily and used income from high oil prices to expand the num- ber of SOEs under its control. During the 1970s and early 1980s, the government followed a haphazard strategy of taking over companies that PRIVATIZATION IN MEXICO 353 fell into financial distress or were of particular interest to the politicians in charge. By 1982 the number of SOEs reached 1,155, and their weight in the economy was unprecedented: they accounted for 4.4 percent of the country's labor force and 30 percent of fixed capital formation, and they received subsidies equivalent to almost 13 percent of gross domestic product (GDP).4 The Privatization Program The Mexican privatization program was one of the world's largest, in terms of both the number of companies privatized and their relative size. Between 1982 and 2003 the number of SOEs dropped from 1,155 to 210. Table 7.2 shows the evolution of the SOE sector from 1982 to 2003, while table 7.3 shows the number of privatized firms and the number of privati- zation contracts per year for the same period.5 The scope of the program entailed the privatization of close to 440 companies in 49 four-digit Standard Industrial Classification (SIC) codes. Table 7.2 State-Owned Enterprises, 1982­2003 State-owned enterprises 1982­88 1989­93 1994­2003 Total at the beginning of period 1,155 666 258 Creation 59 39 108 Liquidations/shutdowns 294 193 58 Mergers 72 17 16 Transfers 25 11 26 Privatizations 157 226 56 In processa 37 Total at the end of the period 666 258 210 Note: To obtain the total number of state-owned enterprises (SOEs) at the end of each period, we add the number of created SOEs to the number at the beginning of the period and subtract the liquidations, shutdowns, mergers, transfers, and privatizations. Total at the beginning of period is the number of legal entities forming part of the parastatal sector at the start of each period; creation is the number of created companies by the state in the specified period; liquidations/shutdowns is the number of companies shut down by the state in the specified period; mergers is the number of SOEs that merged with other SOEs in the specified period; transfers is the number of SOEs that were transferred to other levels of government, including firms that ceased to be treated as state-owned by legal mandate (Ley Federal de Entidades Paraestatales); privatizations is the number of SOEs sold in the specified period; in process is the number of SOEs pending privatization in the specified period; this number reflects the number of ongoing privatizations, liquidations, shutdowns, mergers, or transfers in 2003. a. The most important ongoing processes are Banrural System (13 institutions), FIDELIQ, and Nacional Hotelera de Baja California S.A. de C.V. Source: Presidencia de la República 1982­2003. 354 CHONG AND LÓPEZ-DE-SILANES Table 7.3 The Privatization Program in Perspective Number of transactions Year Companies privatized (privatization contracts) 1983 4 2 1984 3 1 1985 32 10 1986 30 16 1987 22 17 1988 66 51 1989 37 29 1990 91 63 1991 65 37 1992 21 10 1993 12 8 1994 1 1 1995 1 7 1996 1 16 1997 2 12 1998 3 13 1999 32 5 2000 16 2 2001 0 0 2002 0 0 2003 0 0 Total 439 300 Note: The difference between the number of companies privatized and the number of privatization contracts stems from the fact that some companies were sold in a bundle with other privatized firms, while others were split up before the sale. Source: Presidencia de la República 1983­2003. The first period, which lasted from 1982 through 1988, began as the result of a restructuring program intended to increase the overall effi- ciency of the public sector. The program involved restructuring measures as well as a general "cleaning up" of the sector through liquidations, mergers, transfers, and privatizations. This period was also marked by constitutional reforms aimed at reducing the economic role of the gov- ernment; a new federal law for SOEs clarified the relationship and obligations between each SOE and the state and led to a large reduction of unviable operations. Nearly 300 SOEs were liquidated or shut down, and 157 were privatized. The privatization program reached its peak in both size and scope during the Salinas administration, from late 1988 to 1993. Firms sold during this period represented over 96 percent of all assets privatized PRIVATIZATION IN MEXICO 355 and employed 311,000 workers, or 35 percent of the total work force of SOEs (López-de-Silanes 1994). To manage this huge task in such a short time, the president created a special unit within the Ministry of Finance, the Office of Privatization of State-Owned Enterprises (OP), to coordinate a decentralized process that encouraged the involvement of commercial banks, foreign businesses, and financial valuators. The third period, from 1994 to 2003, was characterized by the consol- idation of the previous efforts of divestiture of the parastatal sector. The administration in charge undertook the privatization of strategic areas of the economy and public utilities such as telecommunications (satellites), ports, airports, toll roads, railroads, and the distribution of natural gas. A major reform to the private sector pension system took place in 1995. By 2003 the privatization program had lost its appeal, and in fact the gov- ernment marginally reversed the process by expropriating some previously privatized companies.6 All in all, net privatizations of SOEs were negative in 2003. A central objective of divestiture in Mexico was to transfer control of SOEs to privately owned groups that could provide the necessary align- ment in incentives for better financial and investment decisions of the firms. As a general rule, the government sold 100 percent of its ownership in each SOE privatized, retaining a minority share in only eight cases.7 Six of these companies were instances where the companies were already trad- ing in the stock market, and the intention was to sell secondary packages through the stock exchange.8 The Benefits of Privatization Critics often argue that the benefits of privatization come at significant cost to society through higher prices, lower wages, and reduced income for the government (Campbell-White and Bhatia 1998; Bayliss 2002). In this section we analyze the validity of these claims for the case of Mexico and find evidence that points to the contrary. We study unadjusted and industry-adjusted performance ratios to quantify the effects of privatiza- tion on firms and to ensure that these are not explained by macroeconomic factors. We then examine the importance of price increases, market power, and worker exploitation as potential determinants of the observed increase in profitability. Finally, we analyze the effect of privatization on the quality and accessibility of services. Raw Data We rely on seven broad indicators to measure performance: profitability, operating efficiency, employment and wages, capital investment, out- put, taxes, and prices. For each firm, we measure the change in any given 356 CHONG AND LÓPEZ-DE-SILANES indicator by comparing its value in 1993 to the average value during the four years preceding privatization.9 The results, shown in table 7.4, reveal several interesting phenomena. Profitability increased significantly after privatization according to all indicators: the mean (median) change in profitability ranges from a low of 24.1 (12.1) percentage points for operating income to sales to a high Table 7.4 Changes in Performance for the Sample of Privatized Firms Variable Number Mean change Median change Profitability Operating income/sales 170 0.2411*** 0.1208*** Operating income/PPE 170 0.3450*** 0.1347*** Net income/sales 170 0.3996*** 0.1447*** Net income/PPE 170 0.2713*** 0.1567*** Operating efficiency Cost per unit 170 0.2149*** 0.1676*** Log (sales/PPE) 170 0.6464*** 0.2385*** Log (sales/employees) 169 1.0530*** 0.9909*** Labor Log (employees) 169 64.89*** 56.75*** Log (blue-collar workers) 168 53.44*** 60.87*** Log (white-collar workers) 169 53.52*** 46.34*** Assets and investment Investment/sales 170 0.0150* 0.0103*** Investment/PPE 170 0.0474*** 0.0216*** Output Log (sales) 170 0.5428*** 0.6816*** Net taxes Net taxes/sales 170 0.1301*** 0.0763*** Prices Index of real prices (Paasche) 83 1.31 1.27 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: PPE property, plant, and equipment. This table presents raw results for a subsample of 170 privatized firms between 1983 and 1992. The table presents, for each empirical proxy, the number of usable observations, the mean change, and the median change before and after privatization. Before privatization refers to the aver- age value for the four years before privatization, while after privatization refers to the value as of 1993. We report t-statistics and Z-statistics (Wilcoxon rank sum) as our test for significance for the change in mean and median values, respectively. All variables definitions can be found in the appendix. Source: La Porta and López-de-Silanes (1999). PRIVATIZATION IN MEXICO 357 of 40.0 (14.5) percentage points for net income to sales. All t-statistics and Z-statistics are significant at the 1 percent level. This increase in profitability seems to stem from significant gains in efficiency. The mean (median) cost per unit decreased 21.5 (16.8) per- centage points, while the mean (median) ratio of sales to physical assets (property, plant, and equipment, or PPE) increased 64.6 (23.9) percent. Regarding the efficiency of employees, both mean and median sales per employee double. Once again, all changes are significant at the 1 percent level. The higher profitability and especially the higher levels of sales per employees can be partially explained by reduced employment levels, as the mean (median) numbers of employees plummet by 64.9 (56.8) percent. This reduction is shared more or less equally between blue- and white-collar workers: the mean (median) number of white-collar workers falls by 53.5 (46.3) percent, while the mean (median) number of blue-collar workers falls by 53.4 (60.9) percent.10 The data indicate that investment increases only slightly and therefore cannot be responsible for the vast increase in profitability and operating efficiency documented above. The mean (median) investment-to-sales ratio increased 1.5 (1.0) percentage points and the investment-to-PPE ratio increased by a mean (median) of 4.7 (2.2) percentage points. One of the most surprising findings is that privatized firms increased their sales substantially despite a reduced labor force and only marginal increments in their capital stock. The mean (median) real growth in sales is 54.3 (68.2) percent and is statistically significant at the 1 percent level.11 Once we consider that there is no statistically significant increase in prod- uct prices of privatized firms, the increase in production suggests that consumer surplus should also have increased significantly. To assess the social impact of privatization, we must also account for the taxes paid by newly privatized firms and the prices paid by consumers. The ratio of net taxes to sales increased by a mean (median) of 13.0 (7.6) percent, a significant increase when we consider that sales increased sub- stantially. The magnitude of this change is more evident when we consider that the average firm received a small subsidy before privatization but paid approximately $8.55 million in taxes in 1993. Contrary to predictions, we find that the prices of the products of the mean (median) firm increased only 1.3 (1.3) percent in real terms. One way to gauge the contribution of price hikes to the observed change in profitability is to compare the observed percentage-point increase in the ratio of operating income to sales to that which would take place if pri- vatized firms had increased output but left prices unchanged in real terms.12 Our results suggest that price increases account for 1.24 (0.9) per- centage points of the ratio of operating income to sales. Accordingly, price increases explain about 5.1 percent of the observed change in mean 358 CHONG AND LÓPEZ-DE-SILANES operating income to sales and 7.4 percent of the change in the observed median operating income to sales. The evidence above does not support the hypothesis that price increases play a substantial role in the increased profitability of privatized firms. Industry-Adjusted Data During the early 1990s the Mexican economy experienced a significant structural transformation, and growth accelerated. To ensure that the increases in performance documented in the previous section were not driven by macroeconomic factors, we measure the performance of priva- tized firms relative to those of the industry to which they belong. Figure 7.1 illustrates the closing gap between privatized and private firms. After dramatically underperforming their private peers in net- income-to-sales, operating-income-to-sales, and costs-per-unit ratios, pri- vatized firms caught up with and even surpassed them. The most remarkable results are found for the ratio of net income to sales, where private firms were 17.2 percentage points less profitable than private firms before Figure 7.1 Gap between Privatized Firms and Private Firms Percentage points 20 15 income/sales unit 10 income/sales per 5 Net Operating Cost 0 -5 -10 -15 -20 Before privatization After privatization Note: The figure covers a subsample of 170 privatized firms between 1983 and 1992. Before privatization refers to the average value for the four years before privatization, while after privatization refers to the value as of 1993. All variables definitions can be found in the appendix. Source: La Porta and López-de-Silanes 1999. PRIVATIZATION IN MEXICO 359 privatization and 4 points more profitable afterward. Costs per unit show a similar converging trend, albeit from a different starting point. Before privatization SOEs had median costs per unit 14.2 percentage points higher than their private competitors. After privatization they closed this gap completely and even experienced costs per unit 1 percentage point below their control group. Table 7.5 shows that controlling for industry factors explains a non- trivial fraction of the employment cuts. Relative to industry benchmarks, mean (median) employment in privatized firms fell by roughly 35.4 (34.4) percent. Growth in sales remains strong relative to industry as the mean (median) industry-adjusted growth in sales is 43.2 (48.9) percent. Macroeconomic factors can therefore account for only about 21.4 (28.2) percent of the mean (median) growth in sales while firm restructuring accounts for the rest.13 Although some of the improvements in performance are attributable to macroeconomic factors, the bulk of the observed increase in performance results from privatization. The next step is to address a number of issues relating to the robustness of our results. In particular, we explore the hypothesis that the increased profitability of privatized firms comes from exploitation of consumers through the use of market power and of workers through wage cuts. Market Power One of the main criticisms of privatization is that the increase in prof- itability is attributable to transfers from consumers extracted through market power. To test this hypothesis, we compare the profitability of privatized firms in competitive and noncompetitive industries. If newly privatized firms use market power to extract rents, the social view of pri- vatization predicts that noncompetitive firms would experience larger increases in profitability than would competitive firms, as well as lower growth in output, employment, and investment. In table 7.6 we classify firms as operating in competitive or noncompetitive industries based on two objective criteria.14 The table shows, for each ratio, the median change for competitive and noncompetitive firms following privatization as well as the difference between these changes. Changes in the ratios of operating income to sales and operating in- come to PPE are higher in competitive industries than in noncompetitive ones. In contrast, both net-income-to-sales and net-income-to-PPE ratios increase more in the noncompetitive sector than in the competitive one. With one exception (net income to sales in the classification based on the number of firms), all differences between the competitive and noncompetitive sectors are statistically insignificant. By construction, three factors can account for the conflicting behavior of operating income and net income: taxes, extraordinary items, and 360 CHONG AND LÓPEZ-DE-SILANES Table 7.5 Industry-Adjusted Changes in Performance for the Sample of Privatized Firms Variable Number Mean change Median change Profitability Operating income/sales 168 0.3264*** 0.1531*** Operating income/PPE 168 2.4274*** 0.1492*** Net income/sales 168 0.4144*** 0.2121*** Net income/PPE 168 0.2524*** 0.1736*** Operating efficiency Cost per unit 168 0.1848*** 0.1528*** Log (sales/PPE) 168 0.4684*** 0.2014** Log (sales/employees) 168 0.9157*** 0.8834*** Labor Log (employees) 169 35.35*** 34.37*** Log (blue-collar workers) 168 36.78*** 32.61*** Log (white-collar workers) 168 32.21*** 31.59*** Assets and investment Investment/sales 168 0.0474*** 0.0495*** Investment/PPE 168 0.0328*** 0.0503*** Output Log (sales) 170 0.4324*** 0.4891*** Net taxes Net taxes/sales 168 0.1609*** 0.0757*** * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: PPE property, plant, and equipment. This table presents industry-adjusted results for a subsample of 170 privatized firms between 1983 and 1992. The table presents, for each empirical proxy, the number of usable observations, the mean change, and the median change before and after privatization. Before privatization refers to the average value for the four years before privatization, while after privatiza- tion refers to the value as of 1993. We constructed the industry control groups using all private firms trading in the Mexican Stock Market (three-digit SIC code level). For each privatized firm and for each year, we compute industry-adjusted indicators by taking the difference between the value of the indicator for the firm and its industry control group. We use economywide aggregates, if available, for those firms for which we cannot find a matched industry sample. We report t-statistics and Z-statistics (Wilcoxon rank sum) as our test for significance for the change in mean and median values, respectively. All variables definitions can be found in the appendix. Source: La Porta and López-de-Silanes 1999. interest expense. The increase in taxes paid is larger for noncompetitive firms than for competitive firms and therefore its explanatory power is diminished. Changes in leverage provide a more promising explanation. We conjecture that firms in oligopolistic and monopolistic sectors, Table 7.6 Median Performance Changes in Privatized Firms in Competitive versus Noncompetitive Industries Sorted by number of firms Sorted by market share Non- Difference Non- Difference Variable N Competitive N competitive in medians N Competitive N competitive in medians Profitability Operating income/ 124 0.1365 42 0.1334 0.0031 104 0.1799 62 0.0712 0.1087 sales Operating income/ 124 0.1714 42 0.1698 0.0016 103 0.1832 62 0.1222 0.0610 PPE Net income/sales 122 0.1838 41 0.4260 0.2422** 101 0.2263 62 0.2524 0.0261 Net income/PPE 122 0.2102 41 0.2781 0.0679 101 0.2015 62 0.2716 0.0701 Operating efficiency Cost per unit 124 0.1255 42 0.2362 0.1107*** 104 0.1415 62 0.1907 0.0492 Log (sales/ 126 1.0353 43 0.7875 0.2478 107 1.0877 62 0.7474 0.3403** employees) Log (sales/PPE) 124 0.5882 42 0.3313 0.2569 104 0.6429 62 0.2859 0.3570 Operating income/ 125 19.7240 43 20.9320 1.2080 106 21.6320 62 14.9110 6.7210 employees Labor Log (total 126 0.4339 43 0.3592 0.0747 107 0.4338 62 0.3645 0.0693 employment) Log (blue-collar 126 0.4365 43 0.2473 0.1892 107 0.4495 62 0.2518 0.1977 workers) 361 (Table continues on the following page.) 362 Table 7.6 (continued) Sorted by number of firms Sorted by market share Non- Difference Non- Difference Variable N Competitive N competitive in medians N Competitive N competitive in medians Log (white-collar 126 0.4264 43 0.3342 0.0922 107 0.3327 62 0.4443 0.1116 workers) Assets and investment Investment/sales 123 0.0059 42 0.0098 0.0039 104 0.0048 62 0.0096 0.0048 Investment/PPE 123 0.0144 42 0.0180 0.0036 104 0.0144 62 0.0191 0.0047 Output Log (sales) 124 0.6479 42 0.3752 0.2727 105 0.6479 61 0.4419 0.2060 Prices Index of real prices 66 4.60 17 3.64 8.24 53 9.29 30 2.82 12.1148* (Paasche) Net taxes Net taxes/sales 124 0.0717 41 0.0897 0.0180 104 0.0675 61 0.0815 0.0140 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: N number of observations and PPE property, plant, and equipment. This table presents median performance results for a subsample of 170 privatized firms between 1983 and 1992. In the first measure, firms are considered competitive if they are in an industry with more than 10 firms and as noncompetitive otherwise; in the second measure, firms are considered competitive if they have less than 10 percent of market share and as noncompetitive otherwise. For each group of firms, the table presents median change in each indicator following privatization and their dif- ference. We report Z-statistics (Wilcoxon rank sum) as our test for significance for the change in median values between competitive and noncom- petitive firms. All definitions of the variables can be found in the appendix. Source: La Porta and López-de-Silanes 1999. PRIVATIZATION IN MEXICO 363 perhaps because of their greater access to government-backed capital before privatization, exhibit greater reductions in leverage after privati- zation than do firms in competitive sectors. Finally, it is also possible that the net income of SOEs was unduly depressed by restructuring changes before privatization. The bottom line is that operating income, in contrast to net income, is unaffected by changes in leverage and is thus a better gauge for the impact of market power on profits. In any event, differences in profitability changes are not statistically significant. We also analyze the behavior of investment, employment, and out- put and find no statistically significant evidence to suggest that market power plays a significant role in explaining the increased profitability of privatized firms. In any case, the differences that exist point toward a more dynamic restructuring in the noncompetitive sector relative to the competitive one. For example, according to the classification of com- petitiveness based on number of firms, costs per unit decreased 11 per- centage points more in noncompetitive firms, while according to the classification of competitiveness based on market share, sales per employees increased 34 percent more in noncompetitive firms than in competitive ones. Perhaps the most interesting result is the behavior of real prices. Not only does the increase in the noncompetitive sector lag behind that of the competitive sector, but prices in the former actually fell. According to the market-share classification, the growth of prices in the competitive sector was 9.29 percent while that of the noncompetitive sector was 2.82 percent. This difference is statistically significant at the 10 percent level. All in all, we find no evidence that market power or the exploitation of consumers explains the increased profitability of privatized firms. Transfers from Workers In this section we consider the role of labor retrenchment in explaining the large gains in profitability experienced by privatized firms. The redistribu- tion hypothesis links postprivatization gains in profitability to transfers from workers to shareholders, as wages fall from above-market levels induced by income-redistribution goals. Political models of state ownership also imply that SOE workers are overpaid, but unlike redistribution mod- els, these models lack strong predictions regarding the behavior of wages in the postprivatization period. This uncertainty comes from the fact that public sector jobs are attractive for several reasons besides pay; for exam- ple, they may be desirable because they do not require much effort or because they provide the opportunity to collect bribes in exchange for pub- lic services. We try to quantify the contribution of layoffs and wage cuts to the observed changes in profitability. Examining changes in real wages is a natural way to test competing hypotheses regarding the channels through which privatization works. 364 CHONG AND LÓPEZ-DE-SILANES Contrary to the predictions of the redistribution hypothesis, available evidence shows that real wages increased substantially for the mean and median firm (table 7.7). Real wages per worker increased by a mean (me- dian) of almost 80 (125) percent. This is even more striking since overall real wages in Mexico stagnated during the sample period. Furthermore, gains by blue-collar workers are larger than gains by white-collar workers. Even though white-collar wages increased substantially, the mean and median index of industry-adjusted blue-collar wages rose much more-- well over 100 percent. Although both skilled and unskilled workers experience impressive increases in wages, it is the lowest-paid workers who gain the most. To provide an estimate of the cost savings attributable to layoffs, we make the extreme assumption that the marginal product of all fired work- ers is zero and calculate what the increase in profitability would have been if no workers were fired. Even in this extreme case, savings that come from layoffs are not as large as the reductions in employment suggest. This is so for two reasons: first, total wages represent only 23.38 percent of sales in the preprivatization period, and second, labor costs in the postprivatiza- tion period are spread over a much wider base since sales increase signifi- cantly. Our estimate yields a saving from layoffs equivalent to 6.80 (4.45) percent of sales in 1993. During the same period, the mean (median) ratio of operating income to sales increased 22.12 (9.59) percentage points. Therefore, savings from layoffs account for only 31 to 46 percent of the observed gains in profitability. If, instead of assuming that fired workers' value added is equal to zero, we assume they are half as productive as retained workers, savings from layoffs drop to between 15 and 23 percent of the increase in profitability. In conclusion, real wages experience large increases in the postprivati- zation period probably because those workers who are retained are required to work and are paid accordingly.15 Overall, the high level of labor redundancy in the preprivatization period, the observed increases in real wages, and the productivity gains in the postprivatization period are consistent with the political view. Services and Access The Mexican privatization program spread benefits to society beyond its direct effects on prices and firm profitability. These benefits take the form of greater access to, extended coverage of, and enhanced quality of the services provided by the privatized firms. Table 7.8 summarizes some of these benefits. Among the most important improvements are a dramatic increase in freight road transport and in the provision of natural gas, a substantial increase in the capacity and efficiency of the port system, and substantial investments and increased coverage in the provision of running water and sanitation. Table 7.7 The Role of Transfers from Workers Mean Median Before After Before After Variable N privatization privatization privatization privatization Index of real wages per worker 101 100.00 179.64*** 100.00 224.92*** Index of industry-adjusted real wages per worker 101 100.00 209.30 *** 100.00 198.96*** Index of real wages per blue-collar worker 101 100.00 235.43*** 100.00 248.14*** Index of industry-adjusted real blue-collar wages 101 100.00 265.61*** 100.00 222.43*** Index of real wages per white-collar worker 101 100.00 158.26*** 100.00 200.95*** Index of industry-adjusted white-collar wages 101 100.00 177.97*** 100.00 147.92*** Index of total employment 101 100.00 57.92*** 100.00 58.05*** Total wages/sales 101 0.2338 0.1441*** 0.1506 0.1143*** Operating income/sales 101 0.1530 0.0682*** 0.0251 0.0708*** *** Significant at the 1 percent level. Note: N number of observations. The table presents data of total number of workers, blue-collar workers, and white-collar workers for a sub- sample of 101 privatized firms between 1983 and 1992 for which we have full employment data. For each empirical proxy we show mean and me- dian real wages per worker and the index of industry-adjusted real wages per worker for all three groups. Before privatization refers to the average value for the four years before privatization, while after privatization refers to the value as of 1993. We report t-statistics and Z-statistics (Wilcoxon rank sum) as our test for significance for the change in mean and median values, respectively. All variables definitions can be found in the appendix. Source: La Porta and López-de-Silanes 1999. 365 366 CHONG AND LÓPEZ-DE-SILANES Table 7.8 Other Benefits of Privatization Programs Program Impacts Freight road transport Between 1988 and 1993, the number of firms providing freight road transport service nearly tripled--from 4,456 to 12,972. The number of trucks increased from 58,133 to 142,973. Natural gas Between 1996 and 2000, the Comisión Reguladora de Energía (CRE) awarded 21 gas distribution permits under which concessionaires were obligated to serve 2.3 million customers by 2004. This represents a 15-fold increase in the customer base relative to 1995. Passenger road The impact of the reform has been seen in transport increased entry of new firms or regularization and registration of existing firms; increased demand (between 1990 and 1996 the number of total passengers transported increased from 1.97 million to 2.75 million and the number of vehicles rose from 36,593 to 53,133); and in significantly improved quality and reliability of services. Ports There have been huge increases in installed capacity (from 59 million tons in 1993 to 94 million tons in 1998) and capacity utilization (from 41 percent in 1993 to 59 percent in 1998). In 1993 the port of Veracruz was handling 43 containers/hour per ship. In 2003 this figure was 84. Manzanillo moves 65 containers/hour per ship, and Altamira has achieved the international standard of 50 moves per hour. In Veracruz the total capacity for loading and unloading agricultural bulk cargo improved from 2,500 to 9,000 tons per day between 1995 and 1998, with the port of Progreso showing similar improvement. Railroads New operators invested more than P$3 billion on maintenance of infrastructure and the renewal of rolling stock during 1997­98 and another P$3.3 billion during 1999. Between 1997 and 1998 the total volume of freight handled by the rail system in Mexico increased by 21.5 percent. Telecommunications During the 1990s the number of wire-line telephones in service doubled; wireless telephony grew from negligible levels to nearly 40 percent of all telephones; and waiting lists for service virtually disappeared during the first decade after privatization. (Table continues on the following page.) PRIVATIZATION IN MEXICO 367 Table 7.8 (continued) Program Impacts Toll roads The highway concession program doubled the length of existing toll highways (from 4,500 kilometers in 1989 to 9,900 kilometers in 1994). Urban water supply A 1994 amendment to the federal water law and sanitation initiated a program of concessions in water supply and sanitation. By the year 2000, approximately 14 million Mexicans were served by water systems with varying degrees of private participation (including the Federal District service contracts). Private investments totaling $400 million have been committed, and private operators are handling approximately 16 percent of the 43 meters per second of wastewater effluent that is treated. Source: Rogozinsky and Tovar 1998; World Bank 2003. Fiscal Impact This section explores the fiscal impact of the privatization program as well as its effects on other macroeconomic variables. There are four primary components to the fiscal impact of privatization: the direct revenue gener- ated from the sale; the costs incurred by restructuring before the sale; the elimination of the net flow of subsidies and transfers from the government to the SOEs; and the new stream of tax payments generated under private ownership. Direct revenues from the sale of SOEs were a major source of govern- ment revenue. The aggregate value of the program from 1983 to 2003 amounts to slightly over 5.3 percent of 2003 GDP (table 7.9), with the 1989­93 period accounting for the vast majority of these sales (79 per- cent). About one-third of privatization contracts required "cash-only" payment, and the vast majority of the remaining contracts allowed only very short-term debt lasting no more than a few years. Because most firms were sold for cash and no long-term debt was exchanged, the cash-versus- debt component of privatization plays only a marginal role in the case of Mexico. The costs of restructuring were by no means negligible; the government spent substantial resources restructuring firms before privatization, par- ticularly through labor retrenchment programs. Although the first period (1982­88) had fewer prior restructuring measures, their costs amounted to one-half of total direct revenues. During the second period (1989­93), 368 CHONG AND LÓPEZ-DE-SILANES Table 7.9 The Fiscal Impact of Privatization: 1983­2003 (percentage of 2003 GDP) Nominal Net subsidies price of Privatization during the four privatization restructuring years before Period contract costs privatization 1983­88 0.40 0.20 0.18 1989­93 4.20 1.30 0.35 1994­2003 0.73 n.a. n.a. Total 5.32 1.51 0.53 Number of privatization 255 220 95 contracts n.a. Not available. Source: Data collected by the authors from the original privatization sales and prospectuses from the Secretaría de Hacienda y Crédito Público and Secretaría de Comunicaciones y Transportes (2000). firms were restructured more often, but total costs were equivalent to only one-third of total revenues. In any case, it is clear that restructuring costs were significant. Subsidies and transfers accounted for a significant percentage of the overall government budget--totaling almost 13 percent of GDP in 1982.16 Table 7.9 provides a measure of subsidies net of dividends, revenues, and restructuring costs associated with privatization during the four-year period before privatization; it reveals that these operations imposed a large net burden on the public finances. During the first privatization period, net subsidies during the four years before privatization equaled 0.18 percent of 2003 GDP, almost half of the sale price of the firms privatized during the period. During the second period, net subsidies were more than twice their previous value but represented only about 8 percent of the nominal price of privatization contracts. Overall, the evidence suggests that privatization had a positive and significant impact on the fiscal position of the government. The operating balance and total public-sector borrowing requirements show a significant turnaround. In fact, during the second period the government's budget deficit (without considering direct privatization revenues) climbed from 16 percent of GDP to a small budget surplus (figure 7.2). Furthermore, the greater fiscal discipline afforded by privatization contributed to a sharp reduction in inflation and therefore to improved macroeconomic stability (López-de-Silanes 1994). The funds obtained from the sale of SOEs were allocated to three prin- cipal uses: an emergency contingency fund to protect against nonrecurrent external shocks; a fund destined to reduce the stock of external debt; and PRIVATIZATION IN MEXICO 369 Figure 7.2 The Overall Fiscal Impact of Privatization Percentage of annual GDP 15 10 5 0 -5 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 -10 -15 -20 Subsidies and transfers Public sector total Public sector borrowing to the total SOE sector operational balance requirements Note: SOE state-owned enterprise. The figure shows the evolution of the subsidies and transfers to the total SOE sector, the public sector borrowing requirements, and the public sector total operational balance as a percentage of annual gross domestic product. Source: Data collected by the authors from SHCP 1982­2003a, 1982­2003b and Banco de México 1996­99. a permanent increase in the budget for education and social assistance. The relative composition of public and private investment also changed because of privatization: total government investment in the economy shrank from more than 10 percent of GDP in 1982 to less than 3 percent in 2001 (SHCP 1982­2003b). During this period, investments by SOEs fell from more than one-half of private investment to less than one-tenth, reflecting the joint effect of lower government involvement in the economy and a recovery of private investment. The privatization program also encouraged foreign direct investment. Although foreign investors won only 9 percent of all privatization con- tracts auctioned, they engaged in joint ventures with domestic investors for an additional 11 percent of privatization contracts, including several of the largest firms (table 7.10). The net effect of these factors contributed to the radical reduction of government debt experienced by Mexico since the mid-1980s. As shown in figure 7.3, total debt, both internal and external, dropped from a high 370 CHONG AND LÓPEZ-DE-SILANES Table 7.10 Foreign Direct Investment in Privatization Number of privatization Present value of sale Winner's contracts in the sample price as a percentage nationality 1938­88 1989­93 1994­2003 of average FDI flows Foreign 6 118 9 15.58 Foreign­ 4 116 8 67.35 domestic joint venture Domestic 77 110 18 171.96 Total 87 134 35 255.17 Note: FDI foreign direct investment. Source: Data collected by the authors from the original privatization sales and prospectuses from the Secretaría de Hacienda y Crédito Público, Secretaría de Comunicaciones y Transportes, and Banco de México. of 80.7 percent of GDP in 1986 to only 38.7 percent in 1991 and to 27 percent in 2001. Although the stock of external debt increased sharply before the 1995 financial crisis, and the stock of internal debt has increased continuously since 1996, it is clear that Mexico's debt situa- tion is very different from what it once was. The privatization program, through its fiscal windfall and especially the lower financial require- ments of a reduced SOE sector, is one of the main factors explaining this transformation. Allocative Efficiency Allocative efficiency refers to the idea that firms should be placed in the hands of those who value them the most. The guiding principle behind the idea of maximizing allocative efficiency is that any other mechanism would provide an inferior potential welfare outcome because any distri- bution goal can be achieved by giving the asset to whoever values it the most and then taxing them to compensate others. Some privatization programs, particularly in former communist countries, used voucher mechanisms to achieve a direct transfer of resources. These programs gave each individual a share of the former national industry with the objective of achieving a more equitable distribution of wealth and to prevent a few rich investors from pocketing most of the potential gains from privatization. Recent literature on the subject (Galal and others 1994; Maskin 1992) has emphasized that voucher programs are less efficient than fiscal policy in achieving wealth distribution and are subject to more problems. Diffused PRIVATIZATION IN MEXICO 371 Figure 7.3 Total Net Debt of the Public Sector Average balances as percentage of annual GDP 70 60 50 40 30 20 10 0 198219831984198519861987198819891990199119921993199419951996199719981999200020012002 External debt Domestic debt Note: GDP gross domestic product. The figure includes debt held by the federal government, government enterprises and entities, development banks, and official trust funds. Source: Data collected by the authors from SHCP 1982­2003a, 1982­2003b and Banco de México 1996­99. shareholders are often bought out at deep discounts by agents with asym- metric information, and entrenched management can more easily resist reform because there is no organized pressure from the new owners. Because of this, it seems a wiser idea for governments to follow a process similar to Mexico's, which focused its attention on maximizing allocative efficiency and then used the proceeds from privatization to benefit the rest of the population. As of June 1992 the Mexican privatization program involved a total of more than 2,200 interested individuals and companies who formally requested information and 839 actual bidders. This wide participation is one of the main reasons for the high prices paid for privatized firms in Mexico (López-de-Silanes 1997). As a result of the income derived from privatization programs, spend- ing on education and social assistance increased dramatically. The re- sources generated and liberated by privatization account for a significant 372 CHONG AND LÓPEZ-DE-SILANES share of these programs, and have helped the levels of education, health, and regional development regain their pre-1982-crisis levels (around 9 percent of GDP). For example, in 1992 the education budget increased 9.5 percent in real terms, while the health and social services sector received an increase of 7.7 percent, and transfers to rural development programs increased almost 40 percent (López-de-Silanes 1994). Dos and Don'ts in Privatization The Mexican privatization program provides an excellent opportunity to study the effects of different privatization policies and restructuring pro- grams. Privatization requires heavy government involvement and is usu- ally fraught with conflicts of interest as politicians set up the method and run the process through which they end up either "selling their own firms" or "firing themselves or their friends" (Perotti 1995; Biais and Perotti 2002; Bortolotti, Fantini, and Scarpa 2001; Earle and Gehlbach 2003). Table 7.11 shows the most common restructuring measures adopted before privatization and their frequency in the different periods of the Mexican privatization program. The most popular restructuring measures in both periods (1983­88 and 1989­92) are labor retrenchment and debt absorption, which were carried out with the hope that bidders would be willing to pay more for firms that had been restructured in this way. During the second period, other restructuring mechanisms such as changing the management team, firing the chief executive officer (CEO), and implementing investment and efficiency programs became much more common. To assess the impact of different restructuring measures on the price paid for privatized firms, we need to construct a variable that controls for the value of the firm sold and captures the resources that accrue to the government from the sale of the SOE. We propose an approxima- tion of Tobin's Q--the ratio of the firm's market value to the replace- ment cost of its physical assets--which we call privatization Q (PQ) and estimate as: GNPP TD sh PQ TA where GNPP is the price received by the government once all restructur- ing costs have been deducted; sh is the number of shares sold; TD are total liabilities at the time of privatization; and TA are total assets at the time of privatization.17 The PQ standardization considers GNPPi as the proxy for market value of stock while controlling for debt and assets and allows for the calculation of a good proxy for Tobin's Q.18 PRIVATIZATION IN MEXICO 373 Table 7.11 Restructuring Actions before Privatization Number of privatization contracts Actions 1983­88 1989­92 Management Chief executive officer replaced 11 25 Change of management team 15 54 Change or creation of board of directors 1 30 Labor Labor cuts 24 65 Collective contract canceled 4 13 Collective contract renegotiated 3 10 Debt absorption Outsiders debt 23 53 Cross liabilities 10 45 Fiscal debt 11 37 Efficiency programs Performance measures 6 29 Increased management responsibilities 0 27 Investment programs Investment-performance agreements 10 29 Other investment programs 5 1 Deinvestment measures 9 32 Legal measures Legal debt absorption or solution of 2 15 disputes Negotiations with minority shareholders 11 9 Reorganization or changes in legal status 7 11 Assets restructuring Clarify or document assets ownership 7 22 Patent registrations 1 5 Breaking up companies for sale 3 37 Bundling companies for sale 14 4 Assets spin-offs 10 14 Number of privatization contracts in 87 134 the sample Note: The table shows the main groups of restructuring actions before privatiza- tion and their frequency by period for a subsample of 221 privatization contracts in Mexico between 1983 and 1992. Some privatization contracts did not undergo any of the actions listed above, while others underwent several restructuring actions simultaneously. The exact definition of each restructuring group can be found in the appendix. Source: Data collected by the authors from the original privatization sales and prospectuses from the Secretaría de Hacienda y Crédito Público. 374 CHONG AND LÓPEZ-DE-SILANES The estimations shown in this section regarding the effects of the sale process on privatization prices consider PQ as the dependent variable and firm-specific characteristics, industry characteristics, auction require- ments, and restructuring measures undertaken before privatization as the independent variables. Ordinary least square (OLS) results as well as instrumental variable (IV) results that take into account the possible endogeneity of our proposed measures are shown in table 7.12. Firm and Industry Characteristics All firm and industry characteristics show their expected effect on PQ and are statistically significant. An increase of 10 percentage points on the ratio of net income to sales increases PQ by 16 (15) percent using estimates from an ordinary least squares (instrumental variables) regression. A more aggressive labor force, measured by the cost of firing workers or by the number of strikes suffered in the years leading up to privatization, significantly reduces the price received for privatized firms. For example, an additional strike in the five years leading up to privatization lowers PQ by 17 (23) percent. Government involvement in the industry significantly increases the price paid for a privatized firm, but privatizations that do not transfer control of the firms to private hands are sold at a significant discount. A 10 percentage point increase in the market share of public enterprises increases PQ by about 7 per- cent, probably as a result of protection against competition granted to sectors under heavy state control.19 In contrast, the sale of government holdings in private firms is heavily penalized. The average noncontrol privatization fetched PQs 91 (78) percent lower than those privatiza- tions where the government owned the controlling interest, probably as a result of poor corporate governance and deficient enforcement of shareholder's rights, which fueled investors' fears that they would be exploited by the current majority shareholder. Auction Process and Requirements Auction requirements make a substantial difference in the net price received by the government. Bans on foreign participation reduced PQs by 32 (30) percent. Cash-only sales, another common restriction, were popular because they lowered the government's risk to future breaches of contract and provided an instant infusion into the treasury. Never- theless, these advantages must be carefully weighed against the 30 (27) percent discount in PQ they entailed. It is important to point out that these penalties are independent of the fact that they lower auction com- petition, a substantial factor when we consider that an additional bidder in the final round increased PQ by 17 (12) percent for the aver- age firm. PRIVATIZATION IN MEXICO 375 Table 7.12 Prior Restructuring: Dos and Don'ts Percentage change Percentage change in privatization in privatization Independent variables Q (OLS) Q (IV) Firm and industry characteristics Net income/sales 15.83*** 15.33*** Contingent labor liabilities 1.97** 1.38* per worker Number of strikes 16.98** 23.12** Government in industry 7.06* 7.12* Noncontrol package 90.89*** 78.28** dummy Auction process and requirements Total length of sale 2.23*** 3.35*** Additional bidder in final 16.96*** 11.64** round FDI allowed dummy 31.56* 29.63* Cash-sale only dummy 30.41 26.52 Prior restructuring policy CEO change dummy 34.13 54.88** Percentage of labor cuts 2.82 12.28* Debt absorbed/total liabilities 40.89 11.44 Efficiency measures dummy 14.71 56.09 Investment measures dummy 2.40 95.31 Deinvestment measures 16.57 3.96 dummy * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: OLS ordinary least squares, IV instrumental variables, FDI foreign direct investment, and CEO chief executive officer. The table shows the effect of firm and industry characteristics, auction characteristics and requirements, and prior restruc- turing policies on the mean privatization Q for a subsample of 140 firms privatized in Mexico between 1983 and 1992 for which full information is available. We calculate the effect of a 10 percentage point increase in the net-income-to-sales ratio; an addi- tional strike; a 10 percentage point increase in the government's preprivatization market share of the industry of the privatized firms; an additional month in the length of sale; and a 10 percent reduction in the labor force. For all other variables, we show the dis- crete change of the dummy from 0 to 1. The first column considers prior restructuring measures and the rest of the variables as "exogenous" and uses estimates from an OLS regression. The second column shows the second step of the two-step instrumental vari- ables procedure in which prior restructuring measures, total length of sale from rumors to completion, and the number of bidders in the final round are treated as potentially endogenous variables. All variables definitions can be found in the appendix. Source: López-de-Silanes 1997. 376 CHONG AND LÓPEZ-DE-SILANES The speed at which privatization takes place also has important effects on the net price raised. While there are potential costs of rushing a sale, the benefits of a quick process include disposing of money-losing firms and avoiding costly restructuring. Additionally, a lengthy process usually leads to a deterioration of the operating performance of the firm as managers' incentives collapse and workers are disgruntled about the possibility of losing their jobs. Evidence from Mexico suggests this is the case, as each additional month taken to complete the privatization process costs an average of 2.2 (3.4) percent of PQ.20 Prior Restructuring Policies Government restructuring of SOEs before their sale is fraught with political difficulties. Direct costs of restructuring are quite substantial-- equivalent, on average, to about 30 percent of the sale price. As with other policies, restructuring programs can be defended rationally on the grounds that they increase revenues from the sale, or that they minimize layoffs and contribute to the future success of the privatized firm. As a result, there is great ambivalence about the optimal policy approach to- ward restructuring before privatization. Analyzing this question is not a straightforward proposition because restructuring measures are not undertaken randomly but are selectively targeted to firms that need them most.21 To solve this problem, we use an instrumental variable approach to capture the true effect of restructuring measures on privatization prices.22 In this section, we look at six types of prior restructuring: change in management, labor retrenchment, debt absorption, efficiency programs, investment measures, and deinvestment measures. Restructuring measures are almost certainly endogenous, and therefore we focus on the results provided by the instrumental variable regression. On one hand, management shake-ups before privatization can lead to lower privatization prices if the loss of experienced managers leads to deteriorating performance (Bolton and Roland 1992). On the other hand, removing an entrenched team can improve the operating performance of a firm as well as make it easier to tackle mismanagement and corruption. This is especially likely if management was appointed for political rather than technical reasons (Barberis and others 1996). The empirical evidence supports the second hypothesis, as firing the CEO leads to a statistically significant increase of 55 percent in the PQ.23 The argument against restructuring labor contracts or firing workers before privatization rests on the assumption that the new owner will be better suited to choose which workers to retain and which to dismiss and can therefore restructure the labor force to suit the firm's needs at a lower cost than the government could. This may be particularly true if unions have significant power to influence the political process through elections PRIVATIZATION IN MEXICO 377 or collective action (Freeman 1986). Conversely, the public sector may have a comparative advantage in bargaining with unions if it can convince workers that social mechanisms will be put in place to assist them. For our sample of Mexican privatized firms, labor retrenchment had a slight neg- ative effect on PQ under the OLS regression but a significant positive im- pact once we account for its endogenous nature. Under this specification, a 10 percent reduction in the labor force is associated with a statistically significant 12 percent increase in PQ. This evidence should be taken with care, as recent investigations dealing specifically with the effects of labor retrenchment on privatization suggest that these programs are seldom the optimal policy (Chong and López-de-Silanes 2003). Classical finance theory holds that government's absorption of SOEs' debt should have a neutral effect on price because potential acquirers would simply increase their bids by the same amount as the decrease in debt (Donaldson and Wagle 1995). However, one can imag- ine scenarios under which debt absorption could have a positive or neg- ative effect on net prices if the borrowing terms for the private buyers are different from those for the government or if debt absorption reduces the expected cost of financial distress for excessively leveraged companies (Bolton and Roland 1992). Once we control for endogene- ity, the effect of debt absorption on PQ becomes quantitatively and statistically insignificant. Some of the most frequent restructuring policies undertaken include investment programs and efficiency measures. We would expect a pre- mium in PQ for restructured firms if the government is able to improve the operating performance of firms at a lower cost than the private sector could. However, if restructuring decisions are driven by political motiva- tions or if the government is unable to match the know-how of private firms, the restructuring effort will be a waste of resources and therefore lower PQ. The latter seems to be the case in Mexico as, controlling for endogeneity, firms subject to efficiency measures and investment plans were sold for PQs 56 and 95 percent lower, respectively. If bidders value investments by the government at less than their cost, it is worth investigating whether PQ could be increased by cutting the flow of resources and canceling previously approved investment pro- grams. This may not be straightforward, however, as deinvestment may hurt the long-term profitability of firms and lead to lower prices. Our results find that, on average, undertaking deinvestment measures has no significant effect. Available empirical evidence strongly suggests that restructuring policies do not lead to better net prices per dollar of assets sold. We find that the optimal policy seems to be to refrain as much as possible from engaging in SOE restructuring. Some of the most popular measures such as debt absorption do not increase net prices, while others such as investment and efficiency programs actually reduce net prices. These 378 CHONG AND LÓPEZ-DE-SILANES results are actually quite intuitive once we realize that all they imply is that new owners are able to satisfy their own goals better than politi- cians could. Complementary Policies of Privatization Privatization should not be looked at in isolation. Its success or failure is likely to depend on a set of complementary policies that are sometimes neglected. First, the outcome of a privatization process depends crucially on the quality and appropriateness of the contracts written, particularly when dealing with the provision of public services or public-private ven- tures. Policymakers should not rely on the good will of the private sector to act in the best interest of society. If the privatization contracts do not establish the right incentives, it should come as no surprise that newly pri- vatized firms will try to use market power to increase their benefits or that concession holders may manipulate their accounts to extract rents from the government. Second, it is crucial to consider carefully the necessary deregulation or reregulation for sectors with market power or in which government ownership represented a substantial percentage of total assets before privatization. Without adequate deregulation, firms may not restructure as planned, and potential efficiency gains may be lost, as it is easier to extract rents from the government. Conversely, without a pro- gram to reregulate privatized firms, those with market power may use their ability to influence prices to exploit consumers, and those in sectors with protective regulations may benefit from barriers to entry and other measures originally intended to protect SOEs from competition. Finally, it is important to establish a set of institutions that promote good corporate governance and strengthen shareholder and creditor rights in order to facilitate firms' access to capital and allow recently privatized firms to finance their growth without dependence on the state. The Type of Contract Written The type of privatization contract written is of the utmost importance to ensure that no room is left for opportunistic behavior at the hands of politicians or private buyers. The simplest contracts are straightforward sales of assets in which the government disconnects itself completely from the operational future of the privatized firm. Other types of contracts may lead to a perverse relationship between the privatized firm and the state as managers and bureaucrats collude to serve their own interests at the expense of consumers and taxpayers. Most vulnerable to this sort of manipulation are the provision of services, the construction of infrastructure projects, and the establish- ment of joint ventures between private companies and the government. PRIVATIZATION IN MEXICO 379 The common element in these cases is that the umbilical cord between the government and the firm has not been severed, leaving room for rent- seeking behavior. When privatized firms depend on the state for funding or need permission to undertake certain decisions, it is likely that the incentives to restructure will not be as strong and that management will instead focus its efforts on extracting rents from the government. These perverse incentives are fed by politicians who often aspire to transform firms into national champions by subsidizing them and shielding them from competition. This collusion with business is often rewarded directly by economic support from firms and by the political benefits derived from being perceived as standing up for labor and against foreign com- petitors. To minimize the potential losses from these arrangements and to ensure that privatized firms live up to their commitments, clear disclosure and monitoring mechanisms are needed. The importance of writing adequate contracts can be illustrated by one of the major instances of failure in Mexico's privatization program, the concessions to construct and manage highways. As Rogozinsky and Tovar (1998) explain, to encourage private sector participation in highways, the government included clauses in its contracts guaranteeing minimum traf- fic growth rates. If revenues fell below the guaranteed level, the private party would be entitled to claim a revision of the length of the concession. Moreover, under this scheme, the only variable of adjustment was the length of the concession. These incentives created perverse rent-seeking incentives for concession holders to inflate construction costs and to charge excessively high toll charges. With the peso devaluation of 1994, a bad situation was made much worse as lower than expected revenues and higher interest rates pushed many concession holders to the brink of bank- ruptcy. Banks that had supplied much of the debt were themselves hard hit and thus unable or unwilling to provide relief through debt restructuring (World Bank 2003). The World Bank's 2003 report on private solutions for infrastructure in Mexico points out four basic causes for the failure of road privatization. First, the program as a whole was not financially sound; concessions with the highest profit potential were awarded first, but the overall program was too extensive for each concession to be financially self-sufficient. Sec- ond, the decision to award concessions based on the shortest period of private ownership led to excessively high toll levels. The average duration of concessions was 12 years, and in two cases an award was given for only 5 years. This led to inefficiencies as several newly built toll roads were vir- tually empty while parallel public roads remained heavily congested; truck traffic, in particular, fell below forecast levels. Third, poor feasibility stud- ies led to overgenerous contracts. The Secretaría de Comunicaciones y Transportes did not have the necessary experience or resources to carry out appropriate cost and demand forecasts for such an extensive program and, in particular, it seems to have substantially underestimated the 380 CHONG AND LÓPEZ-DE-SILANES demand elasticity of toll roads. Finally, underbidding and overvaluation of contractor contributions were widespread. Contractors overvalued their contributions, especially once financial problems emerged, because they could then press for extensions on the concession. The policy lesson is clear: contracts must be designed to take into account moral hazard incentives and the asymmetries of information be- tween the government and the private sector. The design of the contracts should be based on considerations of economic efficiency rather than on political or macroeconomic ones and, in particular, contracts should help limit rent-seeking behavior. Deregulation An appropriate regulatory framework after privatization is a key com- ponent of the success or failure of the program. A common element across many failed examples of privatization is inadequate deregulation, which may lead to suboptimal levels of competition and allows produc- ers to keep the gains from privatization without sharing them with con- sumers.24 Deregulation is particularly important in sectors where the state owned most of the assets before privatization, as they tend to be protected by a web of regulations originally instituted to cut the losses of state-owned firms and reduce fiscal deficits. Without a thorough review of these regulations, privatized firms will be artificially shielded from competition and able to make extraordinary gains at the cost of consumers. Deregulation complements privatization in two ways. First, product market competition provides a tool for weeding out the least efficient firms. This process may take too long--or not work at all--if regulation inhibits new entry or makes exit costly. Second, deregulation may also complement privatization by raising the cost of political intervention. Whereas an inefficient monopoly can squander its rents without endan- gering its existence, an inefficient firm in a competitive industry would have to receive a subsidy to stay afloat. Figure 7.4 shows the frequency and composition of deregulation meas- ures taken in Mexico from 1983 to 1992 before privatization. These in- clude price and quantity deregulation, simplified entry and exit barriers, lower restrictions on foreign direct investment and ownership, increased international competition, and the elimination or reduction of subsidies. During the De la Madrid administration, 87 privatization contracts were carried out, with trade liberalization and changes in specific regulatory schemes as the most popular measures. During the Salinas administration, 134 privatization contracts were signed, and deregulation was undertaken much more aggressively. In particular, price and quantity quotas were dis- banded and entry barriers were lowered, both directly and by granting a more prominent role to foreign direct investment. Although it is clear that PRIVATIZATION IN MEXICO 381 Figure 7.4 Deregulation Actions Taken before Privatization Number of privatization contracts 57 43 38 35 29 24 19 17 14 9 9 10 2 1 Price Quantity/quotas/ Foreign direct Trade Changes to specific Lower entry Elimination/ deregulation routes deregulation investment liberalization regulation schemes barriers reduction of subsidies 1983­88 1989­92 Note: The figure shows the main groups of deregulation actions taken before privatization for a subsample of 221 privatization contracts between 1983 and 1992. Not every privatization contract underwent some form of deregulation, while others underwent several. Eighty-seven privatization contracts were carried out during the first period (1983­88) and 134 during the second period (1989­92). The exact definition of each restructuring group can be found in the appendix. Source: Data collected by the authors from the original privatization sales and prospectuses from the Secretaría de Hacienda y Crédito Público. the Salinas administration engaged in deregulation with greater alacrity than the previous administration, privatization coupled with deregulation played a key role throughout the period. Reregulation Regulation should be carefully revised in conjunction with privatization for firms in industries characterized as natural monopolies or in oligopo- listic markets. The reasoning behind the first case is that firms with mar- ket power may have ample opportunities to exploit consumers and that the institutions necessary to supervise them either are nonexistent or do 382 CHONG AND LÓPEZ-DE-SILANES not have the necessary experience to do the job appropriately. When firms with market power are in public hands, it is likely that the profit- maximizing incentive is held in check as the government places priorities on other goals such as providing subsidized services to certain sectors or maximizing employment. Once firms are handed over to the private sec- tor, however, it is likely that the new owners will use whatever means they have at their disposal to increase profits. It is therefore necessary to com- plement privatization with adequate reregulation and with the creation of institutions that ensure fair competition and a level playing field to all participants. A clear example of the perils of not reregulating privatized SOEs appropriately is that of the Mexican banking industry. During the years under state management, banks functioned as an annex to the national treasury and served mainly as a tool for financing state deficits and handing out favors to politically influential sectors. As a result, banks did not develop the necessary experience in valuing the risks associated with particular loans or the skills needed to value collateral. Moreover, when banks were privatized, no supervisory mechanism was in place to ensure that bank loans to related parties were in the best interests of the bank or that reserves were proportional to the riskiness of the loans undertaken. The combination of these two shortcomings proved fatal for the banks' future and to the thousands of shareholders who purchased minority stakes in them. The changes made in the regulatory framework of the banking indus- try were structured around three main objectives: First, they aimed to de- crease the role of the government in determining the allocation of bank loans; second, they tried to create a new market structure for financial services as a whole; and third, they attempted to curtail barriers to entry. Unfortunately, the reforms carried out to achieve these objectives were executed in a piecemeal fashion and not within the context of a compre- hensive plan to provide the banking industry with adequate regulation. In the end, the greater flexibility granted to private banks allowed them to engage in risky behavior and to extract benefits from consumers and the government without providing incentives for banks to reform and become more efficient. Following privatization, lending increased rapidly and was accom- panied by improvements in bank profitability. The average bank increased its operating margin by almost 2 percentage points to 10.85 percent, and profit margins increased from 5.37 percent to 6.36 percent (figure 7.5). Interest margins, however, remained steady at about 7 per- cent despite the fact that nominal interest rates fell from 80 percent to only 35 percent.25 Two hypotheses can be used to explain the higher profitability of pri- vatized banks without resorting to charges of collusion and market power: the employee-wealth-transfer hypothesis, and the reduced-agency-cost PRIVATIZATION IN MEXICO 383 Figure 7.5 Mexican Banks' Profitability Indicators before and after Privatization 0.5900 0.2804 0.2656 0.2278 0.1085 0.0745 0.0896 0.0699 0.0537 0.0636 Financial margin/ Interest margin Profit margin Operating margin Operating income Financial revenue Before privatization After privatization Note: This figure shows mean performance indicators before and after privatization for the cross section of banks privatized in Mexico. Financial margin/financial revenue is the ratio of the financial margin to financial revenue; interest margin is the financial margin divided by the sum of loans portfolio plus securities portfolio; profit margin is the ratio of period net income to the period total revenue; operating margin is the ratio of period operating income to the period total revenue; operating income is the real annual growth rate in the operating income. Before privatization refers to the period between January 1989 and the month of privatization of each bank, while after privatization is the period ranging from the month of privatization to the end of 1992, or for the largest period for which information is available. The banks included in the sample are Atlantico, Banpais, Banamex, Banco del Centro, Bancomer, Bancrecer, Banco Oriente, Banoro, Banorte, B.C.H., Cremi, Confia, Comermex, Internacional, Mercantil, Serfin, Somex, and Promex. Source: López-de-Silanes and Zamarripa 1995. hypothesis. The former argues that wealth is transferred to investors by laying off workers and reducing the wages of remaining employees (Shleifer and Summers 1988), while the latter argues that proper incentives reduce agency costs and lead to higher profitability. Figure 7.6 shows that the average annual change in employment after privatization is 1.64 percent, which would seem to lend credence to the employee-wealth-transfer hypothesis. However, the real annual growth rate of personnel expenses per employee more than doubled in 384 CHONG AND LÓPEZ-DE-SILANES Figure 7.6 Mexican Banks' Performance Indicators before and after Privatization 0.5197 0.3508 0.3252 0.3007 0.1489 0.0116 -0.0164 -0.0204 Growth in the number Growth in personnel Growth in loan Growth in securities of employees expenses per employee portfolio portfolio Before privatization After privatization Note: This figure shows mean performance indicators before and after privatization for the cross-section of Mexican privatized banks. Growth in the number of employees is the annual percentage increase in the number of employees; growth in personnel expenses per employee is the real annual growth in the monthly personnel expense per employee; growth in loan portfolio is the real growth rate in the loan portfolio; growth in securities portfolio is the real growth rate in the securities portfolio. Before privatization refers to the period between January 1989 and the month of privatization of each bank, while after privatization is the period ranging from the month of privatization to the end of 1992, or for the longest period for which information is available. The banks included in the sample are Atlantico, Banpais, Banamex, Banco del Centro, Bancomer, Bancrecer, Banco Oriente, Banoro, Banorte, B.C.H., Cremi, Confia, Comermex, Internacional, Mercantil, Serfin, Somex, and Promex. Source: López-de-Silanes and Zamarripa (1995). the period following privatization, jumping from 14.9 percent to 32.5 percent.26 When put together, these two pieces of evidence show that labor costs did not decrease, but that they in fact increased significantly following privatization. The employee-wealth-transfer hypothesis, therefore, cannot explain the increased operating performance of banks. In terms of the reduced-agency-costs hypothesis, two elements point toward possible savings. The first concerns banking operations as such. Under government ownership, banks had deviated from their main lending activity and entered the securities market, possibly in reaction to PRIVATIZATION IN MEXICO 385 increased competition from brokerage houses. This deviation reflected management or political objectives, such as size or growth, which are common for SOEs all over the world. After privatization, the growth rate of the loan portfolio increased from 35.08 percent to 51.97 percent, while the securities portfolio decreased at an annual rate of 2.04 percent (see figure 7.6). Given that the financial margin on loans ranged from 3 to 15 percent, while that of securities was only about 1 to 3 percent (López-de-Silanes and Zamarripa 1995), it is tempting to argue that pri- vatization reduced the agency problem because resources were channeled toward lending and not toward other objectives. A second source of re- duced agency costs can be found in the ownership structure that emerged after privatization. Most bank managers were also shareholders, which helps increase the incentives to maximize shareholder value. In addition, although management had seats on the board of directors, the board was controlled by outsiders, a management structure that is more likely to allow the board of directors to act as an effective mechanism of corporate governance (Fama and Jensen 1983). These two pieces of supporting evidence notwithstanding, the banking crisis of the mid-1990s and the role that related lending played in bringing about the collapse of the banking system seriously undermine the idea that reduced agency costs can explain the increase in profitability experienced by privatized banks. The conclusion we are left with, therefore, is that inadequate competition allowed banks to collude and increase their profitability without really restructuring. In terms of failed supervision, the government turned a blind eye to overdue loans. The supervisory system was not modernized, and the gov- ernment did not require banks to increase their reserves or to tie them to the underlying riskiness of the loans. Meanwhile, banks continued to pay high dividends and reserves were progressively decapitalized, leaving the whole system vulnerable to collapse. Signs of stress in the financial sector first appeared in 1993 as the economy slipped into a recession, and by July 1994 most financial institutions were experiencing serious difficulties. As the financial crisis evolved, the government took over financially distressed banks with the goal of restructuring them and finding a buyer for them in better times. Once the dust settled, however, only 3 out of 18 commercial banks remained independent, and none escaped unscathed. As of the year 2000, 7 banks were under government management, 5 had been acquired by foreign financial institutions, and 3 had been acquired by domestic financial institutions. Interestingly, many, if not most, of the defaulted loans that led to the collapse of the banking industry were related loans, or loans granted to directors of the bank or companies in the same industrial conglomerate as others owned by bank directors. Many economists have argued that related lending is beneficial to banks as it allows them to better assess 386 CHONG AND LÓPEZ-DE-SILANES the risks related to particular projects and to monitor the compliance of debtors. If this is true, banks may be able to get higher ex post returns from preferential loans granted to related parties than from arm's- length transactions at full rates. In the Mexican case, however, there is evidence that related lending was used by controlling shareholders to loot banks. There are four basic results that, when put together, overwhelmingly support the looting view of related lending. First, the borrowing terms offered to related parties were substantially better than those available to unrelated ones (figure 7.7). Controlling for observable financial characteristics, the interest rate for fixed-rate related loans in pesos was a mean of 6.88 percentage points lower than that of unrelated loans. The smallest difference is 2.25 percent for floating-rate dollar-denominated loans, still a significant margin and, like all the others, statistically significant at the 1 percent level. Moreover, loans to related parties posted collateral or personal guarantees less often than unrelated loans, and when they did it was for a lower amount. Finally, the maturity of related loans was an average of three months longer and the grace period seven months longer than those of unrelated loans. These facts clearly establish that related loans were given preferential treatment over unrelated ones. The second point is that the default rate on related loans was 70 percent compared with only 39 percent for unrelated parties and recovery rates were 19 to 40 cents per dollar lower for related borrowers than for unre- lated ones (figure 7.8). This clearly undermines the idea that preferential loans can be justified by higher ex post returns. Third, related lending represented about 20 percent of all loans out- standing, the limit established by law. Moreover, as the economy slipped into recession, the fraction of related lending doubled for banks that sub- sequently went bankrupt, while it increased only slightly for the banks that survived. This suggests that when bankers thought they might lose their in- vestment, they stepped up the rate of looting to extract as much value as possible while they still controlled the bank, and it is consistent with what a theoretical model of looting would predict. Finally, and most interestingly, the worst-performing loans were those made to persons or companies closest to the controllers of the banks. In most cases, a dollar lent to a firm owned by the bank's owners turned out to be a dollar lost. The evidence provided here clearly shows that related lending had a negative effect on the banking sector in Mexico. Loans to related parties were granted not because compliance was easier to supervise or because related parties were more suitable candidates. Instead, there is clear evi- dence that related loans were used to divert funds away from banks and to exploit minority shareholders. Even so, bank owners emerged from the crisis relatively unscathed; after "tunneling" money out of their bank, they lost control of the bank but not of their industrial assets. Figure 7.7 Terms of Related and Unrelated Loans a. Real interest rates 0.0688*** 0.0408*** 0.0281*** 0.0225*** Flexible rate and Flexible rate and Fixed rate and Fixed rate and domestic currency U.S. dollars domestic currency U.S. dollars b. Collateral and guarantees 1.7072*** 0.3108*** 0.1860*** Loans with collateral Collateral value/loan Loans with personal guarantees c. Maturity and grace period Loans with collateral Collateral value/ loan -3.1043*** -7.3768*** *** Significant at the 1 percent level. Note: This table shows the main characteristics of related and unrelated loans in Mexico for a random sample of loans between 1995 and 1999. Panel a shows the mean change in real interest rates for the sample of related versus unrelated loans. Panel b shows the mean difference of the percentage of loans with collateral; percentage points difference in the collateral to value of the loan ratio; and the difference in the percentage of loans backed up with personal guarantees. Panel c shows the mean difference in months to maturity of the loan and the difference in months of grace granted to the loans. For a description of the sample and definitions of all variables, refer to the appendix. Source: La Porta, López-de-Silanes, and Zamarripa 2003. 388 CHONG AND LÓPEZ-DE-SILANES Figure 7.8 Default and Recovery Rates: Related and Unrelated Loans (percentage) 79.20 69.96 46.24 38.69 39 27.20 Default rates Recovery rates of bad loans Overall recovery rates Related Unrelated Note: This table covers a random sample of 300 loans during the 1995­99 period. Source: La Porta, López-de-Silanes, and Zamarripa 2003. Although related lending explains a significant part of the problem in financial institutions, it cannot be blamed completely for the current malaise of the banking sector. The other aspect of lending is collecting, and banks need to have effective collecting mechanisms in place. Effective corporate governance through creditor protection has recently proven to be a key component of the development of financial systems around the world. As the next section shows, the problems brought on by related lending were exacerbated and in part instigated by deficient shareholder and creditor rights. Privatization and Corporate Governance The development and appropriate functioning of stock and credit markets need a solid regulatory framework that promotes investor protection and disclosure. Recent research shows a strong link between firms' access to capital and efficiently enforced laws (La Porta and Florencio López-de- Silanes 1999; La Porta and others 1997, 1998, 2000a, 2000b; La Porta, López-de-Silanes, and Shleifer 2002). In countries where large numbers of PRIVATIZATION IN MEXICO 389 firms have been privatized and deregulation has increased competition and lowered trade barriers, institutions are urgently needed that can efficiently channel resources to the private sector. The old laws and insti- tutions that might have been efficient in covering the needs of state-owned enterprises will probably not suffice for the requirements of private enter- prises and privatized firms. This section focuses on the role of shareholder rights and creditor rights as important determinants of the success of stock markets and credit institutions. Modigliani and Miller (1958) argue that the size of capital markets is determined only by the cash flows that accrue to investors; roughly speak- ing, this implies that the size of capital markets should be proportional to gross national product. To explain the large discrepancies in the size of financial markets across countries with similar gross national products, however, we need to recognize that securities are more than the cash flows they represent. They entitle investors to exercise certain rights and to exercise control over management through the voting process. Similarly, debt not only entitles creditors to receive interest payments, but also to regain their collateral in the event of bankruptcy. Countries differ enormously in the extent to which they afford legal protection to investors. Not only does a shareholder in Mexico have a very different set of rights from a shareholder in the United Kingdom or the United States, but his/her recourse to redress is also likely to be signifi- cantly different--that is, weaker. The legal theory (La Porta and others 1997, 1998) predicts larger capital markets in countries where agency costs are reined in by the law and the institutions built to support their en- forcement. The evidence presented in these studies shows that, all in all, Mexico offers investors a rather unattractive legal environment. Figure 7.9 shows market capitalization as a percentage of GDP and the number of listed firms in the Mexican stock exchange. Not surprisingly, the Mexican market has a low capitalization by international standards. The stock market enjoyed a significant boom during the Salinas adminis- tration, as many privatized firms were subsequently listed and those that already were privatized increased their capitalization considerably. Mar- ket capitalization increased from slightly over 5 percent in 1988 to almost 45 percent in only six years. Following the 1994­95 crisis, capitalization plunged to 25 percent and has dropped further since. Moreover, the num- ber of listed firms itself is declining, which suggests that firms are either going bankrupt or being taken private. In either case, it is a clear sign that corporate governance structures are not functioning efficiently. Some recent attempts have been made to strengthen shareholder rights in Mexico. Most notably, the 1997 and 2001 securities acts moved from a merit to a disclosure system of regulation. They also established re- quirements for a minimum number of independent directors on boards of listed companies and mandated the creation of an independent audit com- mittee. A forthcoming securities act attempts to expand protection further 390 CHONG AND LÓPEZ-DE-SILANES Figure 7.9 Market Capitalization as Percentage of GDP and Number of Companies Listed in the Mexican Stock Exchange Percentage of annual GDP Number of listed firms 50 450 45 400 40 350 35 300 30 250 25 200 20 150 15 100 10 5 50 0 0 1985 1988 1990 1991 1992 1993 1994 1995 199619971998 1999 2000 2001 Listed firms Market capitalization/GDP Note: This table shows the number of firms listed on the Mexican stock exchange and their market capitalization as a percentage of GDP (gross domestic product). Source: For market capitalization/GDP, Standard and Poor's Emerging Market Database (and Emerging Stock Markets Factbook); for the number of listed companies in 1985 and 1988, Banco de México and BMV 1985­92); for the number of listed companies 1990 to 2001, the World Federation of Exchanges. by making the economic group the subject of regulation, establishing the duty of loyalty and care for directors, improving disclosure requirements, improving the enforcement powers of the Securities and Exchange Com- mission, and establishing a special committee to monitor conflicts of interest and related-party transactions. Furthermore, the new securities act seeks to limit the issuing of nonvoting shares and expand the rights of minority shareholders to obtain information about the firm, convene shareholder meetings, designate comptrollers, and even challenge unfair actions through the judicial system. Although these reforms are expected to have a significant effect, their scope is inherently limited as they apply PRIVATIZATION IN MEXICO 391 only to listed companies. The next step will be to expand some of these best practices to a broader set of firms. Another source of finance for privatized firms and other businesses is bank lending. However, if financing for privatized SOEs is expected to come from privatized banks--or from any other private credit institution-- then creditor rights, embedded in bankruptcy laws and the efficiency of courts, must be strengthened and streamlined. Without proper bankruptcy procedures that allow for the expedient re- covery of assets, financial institutions will be reluctant to lend and may end up failing to satisfy the financial needs of the private sector. La Porta and others (1997) empirically confirmed the existence of a strong link be- tween efficient creditor rights and an efficient judiciary with deeper debt markets. In a more recent paper, Djankov and others (2003) studied the costs, in both time and money, involved in collecting a bounced check or evicting a tenant in 109 countries. Their results show that Mexico ranks among the worst countries, which suggests that if there are considerable difficulties in using the court system for simple procedures, foreclosing on defaulted loans would be even more maladroit. The lesson is clear: if banks are to function appropriately, they must have mechanisms in place to recover the costs of loans in the case of default. Otherwise, the possi- bility of having a large number of irrecoverable loans default will drive up the interest rates for all projects, stifling growth and aggravating adverse selection problems. Two types of reforms may be key to building stronger and sounder financial institutions. First, potential conflicts of interest in boards of directors of banks must be reduced. These boards have allowed large- scale, unprofitable related lending to occur, which has increased the fragility of the banks and of the overall financial system. Corporate governance reform must be designed to prevent interested directors from voting on approving related transactions that do not benefit the institution as a whole. Tougher limits on related loans, disclosure of related credits, and laws that effectively punish directors who vote in favor of unprofitable transactions should be among the first group of reforms. Second, in the event that a borrower defaults on its loans, financial institutions need to be able to exercise their rights to collect their debts. An essential condition for improved creditor protection is to strengthen bankruptcy laws and to increase efficiency in the implemen- tation of these rights. Conclusion Empirical evidence shows that the Mexican government was effective in improving its fiscal discipline, increasing the efficient allocation of re- sources and restructuring inefficient SOEs. Success, however, has not been 392 CHONG AND LÓPEZ-DE-SILANES absolute. Although the privatization process was drastic and far-reaching, public utilities and firms in the energy sector remain in state possession. Moreover, there have been important failures in the Mexican privatization process, such as banks and highways. The Mexican experience provides valuable insights into what to do and what not to do regarding large-scale privatization programs. Mexico's program should be considered a success, as the performance of firms in- creased dramatically according to all performance measures and both the government and consumers are better off as a result. Nevertheless, partic- ular care should be taken to provide appropriate regulation to oligopolis- tic sectors or those that have spent considerable time under the control of the state. This study provides three useful lessons that any privatization program should heed to maximize the probability of success. First, the privatization process must be carefully designed. Special requirements such as a ban on foreign direct investment or cash-only payments lead to substantial price discounts for firms sold. Simplicity and transparency, in contrast, increase prices because they expedite the whole process and increase the number of bidders. Speed is important because the operational efficiency of SOEs tends to atrophy once rumors of divestiture arise. The ability to draw on a wide pool of bidders is crucial to maximize the price received for privatized firms and therefore to obtain the most social benefit from the program. Privatization of services and infrastructure provides a valuable opportunity to enhance social welfare, but contracts must be carefully designed. Special attention is needed to ensure that the incentives of future owners are aligned with those of the government. Second, restructuring firms before privatization is usually counterpro- ductive in raising net sale prices. Restructuring programs are usually politically motivated and cost more than the value they add to companies; this is in addition to the fact that they lengthen the privatization process and therefore increase the costs related to the sale. Although the evidence from Mexico shows that labor retrenchment paid off, this is probably due to the abnormally bloated work force of most Mexican SOEs. Labor retrenchment is fraught with difficulties because the most politically palatable option, voluntary retirement programs, is plagued by adverse selection problems. Under these schemes, the most experienced and pro- ductive workers tend to leave and the least productive stay. The govern- ment is thus left with a steep bill from severance payments and bidders are left with a lower-quality work force that may not possess the skills necessary to run the firm successfully. Third, reregulation of sectors with market power, deregulation of in- dustries previously protected by the government, and efficient corporate governance institutions are important complementary measures to pri- vatization. Mounting evidence is underscoring the importance of good PRIVATIZATION IN MEXICO 393 corporate governance institutions in ensuring positive results from pri- vatization and avoiding debacles similar to the related-lending problems suffered by the Mexican banking industry. The need for better regulation and corporate governance is not an argument for slower privatization, but it does point to the need for studying similar privatization programs to anticipate the regulatory framework and best practices codes needed for firms to have a smooth transition into the private sector. Privatization and deregulation have allowed the Mexican government to withdraw from virtually all commercial and industrial activities and to concentrate on providing a stable economic environment. The proceeds from privatization helped deal with a crushing debt burden and allowed the government to invest in education and poverty alleviation. The evidence provided in this chapter leaves but one conclusion: The overall effect of privatization has been positive for Mexico. Appendix 7A Description of Variables This appendix describes all variables used to compute the results pre- sented in the chapter. It contains three sections, each based on a different section of the chapter. The first section describes the restructuring ac- tions undertaken for a cross section of 221 privatized SOEs in the 1983­92 period. We gathered these data from the original privatization sales and prospectuses from the Mexican Ministry of Finance and Public Credit (SHCP). The second section corresponds to the quantitative vari- ables collected for a cross section of 170 nonfinancial firms privatized in Mexico during the 1983­92 period and used to measure the performance of privatized firms. We gather the preprivatization data from the original privatization sales and prospectuses from SHCP. For each firm, we com- pute the preprivatization value of each variable or ratio by averaging its value in the four years before the privatization of the firm. The postpri- vatization data come from the 1994 economic census from the Mexican National Statistics Institute (INEGI) and refer to 1993 data for each firm. When necessary, industry adjustments are made to various variables and ratios with the use of financial data of publicly traded firms in the Mex- ican stock market, INEGI's "Monthly Survey of Industry," and the pro- ducer price index at the sector level. The third section of the table defines the loan characteristic variables used in our related-lending section. Reg- ulation in 1995 required banks to submit a list of the 300 largest loans, as well as their size and detailed information about the borrowers. Our sources are the SAM-300 database (largest 300 loans for each bank), the SENICREB database (complete list of loans made by each of the priva- tized banks), and banks' individual databases as reported to the Mexican Banking Commission. 394 CHONG AND LÓPEZ-DE-SILANES Table 7A.1 Definition of Variables Variable Definition Restructuring action Management These actions include changes of the CEO (chief executive officer) and in the next level of the management team, as well as the creation of or changes in the board of directors in the two years before privatization. Labor These actions include preprivatization labor reductions of union or nonunion workers with their related severance payment costs, as well as varying forms of labor contract restructuring, plans for worker ownership of shares, cancellation of collective contracts in some subsidiaries or for the whole company, and union unification under the same contract. Debt absorption The actions include the government's engagement in partial or total debt absorptions of outsiders' debt (that was owed to private companies or private banks), cross liabilities, and the absorption of the company's past-due fiscal liabilities or taxes. Efficiency This category includes all of those detected actions that programs increased the management responsibility, increased the flexibility of investment policies, or referred to the installation of programs with defined performance targets and internal restructuring of operations. Investment These actions involve only detectable programs of programs rehabilitation and maintenance plans, agreements of financial restructuring tied to inflow of resources for operation improvements, and temporary reopenings of the plant. Deinvestment These actions include partial or total shutdown of measures operation, stopping of major investment programs, declarations of emergency-only expenditure taking place, and stopping of all government funding. Legal measures This group involves the solution or clarification of legal disputes, the reorganization and consolidation of all legal information and demands against the company, renewal or creation of necessary operation concessions and permits, and changes in the legal incorporation or in the company's by-laws. Under this category are also considered negotiations with other shareholders regarding either their "preference status" in the case of sale or their agreement to sell their shares to the new buyer or to put them in the government's hands for privatization of a larger package. PRIVATIZATION IN MEXICO 395 Table 7A.1 (continued) Variable Definition Assets The actions include the government's engagement, restructure before the sale, in clarification, or sometimes documentation, of fixed-assets ownership, especially in the cases of broken-up corporations; patent registrations and changes in assets classification mechanisms; break-ups or splits of the SOE for its sale; and spin-offs of specific assets. Deregulation These measures reflect deregulatory actions undertaken by the government tied to privatization. The range of actions includes price and quantity deregulations, foreign direct investment and ownership restrictions, international trade quotas or tariffs, entry or exit barriers for domestic or international competitors, changes in the regulation scheme of the company or the industry, and elimination or reduction of subsidies. Firm characteristics Privatization Q The value of the government's net privatization price (GNPP) adjusted by the fraction of shares sold plus total liabilities at the time of privatization, divided by total assets of the company at the time of privatization. GNPP is calculated as follows: GNPP B P * R GC Adj where (B ) is the present value of the nominal price of sale as registered in the sale contract that is adjusted in the following ways: subtraction of the cost of the restructuring measures (P*R) undertaken by the government before the sale; subtraction of the costs of the "government commitments" and the "special clauses" promised by the government at the time of the sale (GC), matching them with the actual bills paid later on; and addition and subtraction of the adjustments made to the sale contract (Adj), which includes reimbursements on both sides when the financial statements differ from the ones given to the bidders before the sale. Operating The ratio of operating income to sales. Operating income/sales income is equal to sales minus operating expenses, minus cost of sales and depreciation. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. (Table continues on the following page.) 396 CHONG AND LÓPEZ-DE-SILANES Table 7A.1 (continued) Variable Definitions Operating The ratio of operating income to property, plant, and income/PPE equipment. Operating income is equal to sales minus operating expenses, minus cost of sales and depreciation. PPE is equal to the value of a company's fixed assets adjusted for inflation. Net income/ The ratio of net income to sales. Net income is equal sales to operating income minus interest expenses and net taxes paid, as well as the cost of any extraordinary items. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Net income/PPE The ratio of net income to property, plant, and equipment. Net income is equal to operating income minus interest expenses and net taxes paid, as well as the cost of any extraordinary items. PPE is equal to the value of a company's fixed assets adjusted for inflation. Cost per unit The ratio of cost of sales to net sales. Cost of sales is equal to the direct expense involved in the production of a good (or provision of a service), including raw material expenditure plus total compensation paid to blue-collar workers. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Log (sales/PPE) The log of the ratio of sales to property, plant, and equipment. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. PPE is equal to the value of a company's fixed assets adjusted for inflation. Log (sales/ The log of the ratio of sales to total number of employees) employees. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Employees correspond to the total number of workers (paid and unpaid) who depend directly on the company. Log (total The log of the total number of employees. Employees employment) correspond to the total number of workers (paid and unpaid) who depend directly on the company. They receive, in general, either salary or wage payments on a regular schedule and work at least 15 hours PRIVATIZATION IN MEXICO 397 Table 7A.1 (continued) Variable Definitions a week. A minimal number of these workers do not receive a regular salary. This number includes all workers on strike, as well as workers who still report to officials of the company despite different work locations, and workers on sick-leave or vacation. It does not include individuals who are retired or working on commission. Log (white- The log of the total number of white-collar workers. collar workers) White-collar workers perform skilled labor and administrative tasks for modest to high salaries. They are individuals involved in sales, administration, and management. Log (blue-collar The log of the total number of blue-collar workers. workers) Blue-collar workers perform unskilled or semiskilled labor for modest to low wages. They perform tasks directly related to the (mass) production process or menial services. Typically, they are factory line or maintenance workers. Investment/sales The ratio of investment to sales. Investment is equal to the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Investment/PPE The ratio of investment to property, plant, and equipment. Investment is equal to the value of expenditure to acquire property, equipment, and other capital assets that produce revenue. PPE is equal to the value of a company's fixed assets adjusted for inflation. Log (sales) The log of sales. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Index of real The Paasche price index is the ratio (expressed as a prices percentage) of the total value in the given year of the (Paasche) quantity of each commodity produced in the given year to what would have been the total value of these quantities in a base year. To isolate changes in relative prices, for each firm and for each of the firm's line of business, we found an appropriate control group among the 61 sectors that have official producer price index statistics and report the (Table continues on the following page.) 398 CHONG AND LÓPEZ-DE-SILANES Table 7A.1 (continued) Variable Definitions postprivatization behavior of the firm-level price index relative to its control group. Net taxes/sales The ratio of net taxes to sales. Net taxes are equal to corporate income taxes paid net of direct subsidies received during the fiscal year. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Index of total For each firm, the index takes a value of 100 for the employment preprivatization period. The 1993 value is computed by augmenting the preprivatization value by the difference between the cumulative growth rate of employment of the firm and the cumulative growth rate of employment of the control group in the postprivatization period relative to the average employment in the four years that preceded privatization. Industry control groups are given by three digit SIC code sectors for all manufacturing firms, and an index of economywide total employment for firms in the mining and the service sectors. A similar procedure is used for the calculation of the index of blue-collar workers and the index of white-collar workers. Index of real For each firm, the index takes the value of 100 for the wages per preprivatization period. This refers to the real worker average wages paid per worker in each firm. The consumer price index was used as a deflator to calculate real wages. A similar procedure is used for the calaculation of the index of real wages per blue- collar and per white-collar worker. Index of For each firm, the index takes the value of 100 for the industry- preprivatization period. The 1993 value is computed adjusted real by augmenting the preprivatization value by the wages per difference between the cumulative growth rate of real worker wages per worker of the firm and the cumulative growth rate of real wages per worker of the control group in the postprivatization period relative to the average real wage per worker in the four years before privatization. Industry control groups are given by three-digit SIC code sectors for all manufacturing firms, and an index of economywide real wages per worker for firms in the mining and the service sectors. A similar procedure is used for PRIVATIZATION IN MEXICO 399 Table 7A.1 (continued) Variable Definitions the calculation of the index of industry-adjusted real wages per blue-collar worker and the index of industry-adjusted real wages per white-collar worker. Total The ratio of the total wages paid by the firm to the wages/sales sales of the firm. Total wages are equal to the total wage bill paid by the firm to blue- and white-collar employees. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Loan characteristics Real interest rate The average real interest rate paid during the duration of the loan. The average real interest rate is computed as: 1 T (1 it s) , T a t 1 (1 pt) where i is the reference interest rate assigned to the loan, s is the spread above the interest rate, and the inflation rate. For loans in Mexican pesos, the inflation rate was calculated using the producer price index (INPP) excluding oil products. For loans in U.S. dollars and other foreign currencies, the inflation rate was calculated using the U.S. producer price index of finished products. Collateral Variable that takes a value equal to 1 if the loan is backed by collateral, and 0 otherwise. Definitions from collateral include physical tangible assets, financial documents (such as title documents or securities), intangibles, and business proceeds pledged by the borrower to ensure repayment on the loan. Collateral does not include personal guarantees such as obligations backed only by the signature of the borrower or the submission of wealth statements from guarantors to the bank--a usual practice in Mexico. Collateral value/ The ratio of collateral value to loan value when the loan value loan was first granted. Personal Variable that takes a value equal to 1 if the loan is guarantees secured by a personal guarantee, and 0 otherwise. A personal guarantee is defined as the obligation of repayment by a letter of compromise. Usually the (Table continues on the following page.) 400 CHONG AND LÓPEZ-DE-SILANES Table 7A.1 (continued) Variable Definitions debtor must submit wealth statements from a guarantor who is willing to back his/her loan. Maturity The number of months to maturity of the loan, starting the moment the loan was granted. Grace period The number of months beyond maturity given to a debtor to repay the balance due. A grace period is granted to a debtor on an individual basis. A loan may have no grace period at all, but, if granted, the grace period may vary according to the loan type and terms established in the loan contract. Loan in Variable that takes a value equal to 1 if the loan is domestic denominated in Mexican pesos or in Mexican currency inflation-adjusted currency units (UDIs or Unidad de Inversión), and 0 otherwise. Loan with fixed Variable that takes a value equal to 1 if the loan pays interest rate a fixed interest rate, and 0 otherwise. A fixed interest rate loan pays an annual percentage rate on a fixed rate for the duration of the loan. Source: Authors' calculations. Notes The authors thank Patricio Amador and José Caballero for excellent research as- sistance. The authors would also like to thank Guillermo Babatz, José Antonio González, and Abraham Zamora from Secretaría de Hacienda y Crédito Público (SHCP) and Rodolfo Salgado, Alonso Martínez, and Linda del Barrio from Secre- taría de Comunicaciones y Transportes (SCT) for facilitating access to part of the data. 1. For example, the National Commission of Irrigation and the National Commission of Highways and Roads were in charge of providing investments to rural areas and of encouraging trade by improving the infrastructure of Mexico. 2. Among the most important firms and funding institutions created during this period are the National Commission of Electricity, the National Railway Company, the Exporting and Importing Mexican Company (CEIMSA), the National Bank of Agricultural Credit, and the National Bank of Ejido Credit. 3. The Ministry of Finance and the Ministry of National Goods and Admin- istrative Inspection were the two entities with overseeing powers established by the 1947 Law for the Federal Government's Control of Decentralized Institutions and Enterprises of State Participation. 4. According to Aspe (1993), even this number does not capture the true magnitude of the drain on the government from the SOE sector. If we include the banking sector operations that led to the nationalization of commercial banking in September 1982, the corrected figure would increase to a staggering 18.5 percent of GDP. PRIVATIZATION IN MEXICO 401 5. We group our observations in three periods, and for each, the privatization strategy often involved splitting companies into smaller units or making multifirm conglomerates For instance, Ferrocarriles Nacionales was split into 8 railroads, whereas 35 local airports were sold in packages of about 10 each. 6. For example, in 2001 the government expropriated several sugar mills for questionable public interest reasons. In addition, between 2001 and 2003 the government created the following enterprises: Consejo Nacional para Prevenir la Discriminacion (National Council to Prevent Discrimination); Servicio de Ad- ministracion y Enajenacion de Bienes (Service of Administration and Alienation of Goods); Consejo Nacional de la Cultura Física y el Deporte (National Coun- cil of Physical Culture and Sport); and Instituto Nacional de Lenguas Indígenas (National Institute of Indigenous Languages). 7. For a detailed description of the auction process followed for SOE divesti- tures, see López-de-Silanes (1994, 1997). 8. The clearest case is Telmex (the telephone communications monopoly), where the government originally kept 31 percent of shares and now owns only 4.75 percent. The other five cases include three banks (Bancomer, Banca Serfín, and Banco International) and two companies in transportation (Cia. Mexicana de Aviación and Transportación Marítima Mexicana). 9. We use industry-matched producer price indexes (PPI) to adjust all product prices and the aggregate consumer price index (CPI) to adjust the values of all other nominal values. The choice of deflator is not irrelevant since the CPI series shows higher inflation than the PPI series over the sample period. Therefore, the use of the CPI index imparts a conservative bias against finding significant increases in sales, earnings, fixed assets, wages, and taxes in the postprivatization period. 10. Even though these figures seem large, they probably underestimate the true level of labor retrenchment experienced by privatized firms. In fact, looking at the subsample of firms for which we have complete employment data for all four preprivatization years (117), we find that the mean (median) number of workers falls 16.5 (4.2) percent between t-4 and t-1 and a further 62.4 (64.7) percent between t-1 and 1993, where t is the year of privatization. 11. It is possible that this increase in output overestimates the true impact of privatization as some of the observed gains may simply reflect redistribution away from customers. Some SOEs priced their output below market levels or failed to charge for goods and services produced because of corruption, political meddling, or sheer incompetence. Although there is no way to directly quantify the impor- tance of these factors in our sample, available evidence regarding the evolution of prices for the products of privatized firms suggests this is not the driving force behind increased sales. 12. Specifically, we define Sales1993 and Cost1993, respectively, as sales and op- erating costs in the postprivatization period and as the increase in real prices, and compute the following measure for the contribution of price increases to higher profitability: Price Contribution [(Sales1993 Cost1993)/Sales1993] {[(Sales1993/1 ) Cost1993]/[Sales1993/1 ]} 13. We can calculate the macroeconomic contribution to the growth in sales by measuring the difference in raw and industry-adjusted indicators (0.1164 and 0.1925, respectively) as a proportion of the raw increase (0.5428 and 0.6816, respectively). 14. Under the first classification, firms are considered competitive if they are in an industry with more than 10 firms and as noncompetitive otherwise. Under the second classification, firms are considered competitive if they have less than 10 per- cent of market share and as noncompetitive otherwise. 402 CHONG AND LÓPEZ-DE-SILANES 15. To confirm the robustness of our interpretations, we carried out a poll to determine the principal perceived reasons for increased wages. The essential element of the respondents' explanation was that SOE jobs were desirable not because they paid well, but because they required little effort. After privatization, employers quickly moved to dismiss workers who did not increase their produc- tivity and hired new workers from a different pool than those who were hired under state management. Firms retained the most productive workers and offered them conditions similar to both those prevailing in the private sector and those offered to new hires. We therefore believe that higher wages are explained by improvements in worker productivity. 16. Based on data from SHCP 1982­2003a, 1982­2003b. 17. For the calculation of GNNP, see the definition of privatization Q in the appendix. 18. As a robustness check, an additional standardization was calculated in the same fashion but using total shareholders' equity (defined as the sum of total assets and total liabilities at the time of privatization) to normalize the privatization Q. The results obtained from this estimate are very similar and are thus omitted. 19. An alternative hypothesis is that sectors under control of the state tend to be those with the greatest untapped capacity for efficiency gains. In this case the premium is explained because of a higher perceived future growth rate and not because of the belief that protective regulations will be maintained. 20. This hypothesis is supported by López-de-Silanes (1997), who finds that the discount in privatization Q is explained by "internal speed," the length of time between the moment when the first rumors of privatization arise to the time of the public announcement, and not by "public length," the time elapsed from the formal announcement to the conclusion of the sale. 21. We would expect the government to absorb debt of highly indebted SOEs, to fire workers when firms face serious overemployment, or to invest in new ma- chinery when production processes are outdated. If the endogenous nature of these measures is not considered, we run the risk of reaching the wrong conclusions as regression coefficients would capture not only the effect of the restructuring meas- ure, but also the negative effects of being in distress or having a bloated work force. 22. We apply a two-step instrumental variables approach by estimating a nonlinear reduced-form equation that describes the probability that a particular labor-restructuring policy will be implemented. The instruments used are classi- fied in two groups: firm-level and macroeconomic-level determinants. The firm- level variables include the presence of a leading agent bank, involvement of a government ministry before privatization, the political affiliation of unions, and sectoral dummies. The macroeconomic variables include the average GDP growth rate, the degree of openness in the three years before privatization, and the legal origin of the country. None of these variables is statistically significant when included in the price equation. The F statistic for the excluded instruments is statistically significant at 1 percent in all cases. 23. We also investigated the effects of changing the management team inde- pendently of the CEO and found no significant effect. We can therefore tentatively conclude that the main benefit stems from changing the head and not the body of the management team. 24. Megginson and Netter 2001; Boubakri and Cosset 1998. 25. For this section, we consider the period before privatization as the time elapsed from the beginning of 1989 to the time of privatization of each bank; after privatization corresponds to the period from the time banks were privatized until the end of 1992, or for the longest period for which we have data. This division bi- ases our findings against finding dramatic improvements in bank performance, which had been increasing since 1986, first because of a partial flotation of bank PRIVATIZATION IN MEXICO 403 shares through the Mexican stock market and, second, because of the deregula- tion measures adopted between 1987 and 1988. For example, although the ratio of financial margin to financial revenue increased almost 6 percentage points between 1988 and 1994, it increased a further 12 percentage points between 1987 and 1989. For a detailed analysis of bank performance during this period, see López-de-Silanes and Zamarripa (1995). 26. The most likely explanation for this increase in personnel expenses despite the reduction in employment is that wages increased as private financial institu- tions competed for talented employees. This conjecture is reinforced by the fact that the cuts in employment show a higher concentration in unskilled labor. References Aspe, Pedro. 1993. Economic Transformation The Mexican Way. Cambridge, Mass: MIT Press. Banco de México. 1996­99. "The Mexican Economy." Annual publication of the Dirección de Organismos y Acuerdos Internacionales of the Bank of Mexico, Mexico D.F. Barberis, Nicholas, Maxim Boycko, Andrei Shleifer, and Natalia Tukanova. 1996. "How Does Privatization Work? Evidence from the Russian Shops." Journal of Political Economy 104: 764­90. Bayliss, Kate. 2002 "Privatization and Poverty: The Distributional Impact of Utility Privatization." Annals of Public and Cooperative Economics 73: 603­25. Biais, Bruno, and Enrico Perotti. 2002. "Machiavellian Privatization." American Economic Review 92: 240­58. BMV (Bolsa Mexicana de Valores). Various issues for 1985 through 1992. Indi- cadores Bursatiles. Mexico City. Bolton, Patrick, and Gerald Roland. 1992. "Privatization Policies in Central and Eastern Europe." Economic Policy 15: 275­303. Bortolotti, Bernardo, Marcela Fantini, and Carlo Scarpa. 2001. "Privatisation: Pol- itics, Institutions and Financial Markets." Emerging Markets Review 2: 109­36. Boubakri, Narjess, and Jean Claude Cosset. 1998. "The Financial and Operating Performance of Newly Privatized Firms: Evidence from Developing Countries." Journal of Finance 53(3): 1081­110. Campbell-White, Oliver, and Anita Bhatia. 1998. "Privatization in Africa." Direc- tions in Development series. World Bank, Washington, D.C. Chong, Alberto, and Florencio López-de-Silanes. 2003. "Privatization and Labor Restructuring Around the World." Yale University, New Haven, Conn. Djankov, Simeon, Rafael La Porta, Florencio López-de-Silanes, and Andrei Shleifer. 2003. "Courts." Quarterly Journal of Economics 118: 453­517. Donaldson, David J., and Dileep M. Wagle. 1995. "Privatization: Principles and Practice." Lessons of Experience Series, 1. World Bank, International Finance Corporation, Washington, D.C. Earle, John, and Scott Gehlbach. 2003. "A Spoonful of Sugar: Privatization and Popular Support for Reform in the Czech Republic." Economics and Politics 15 (1): 1­32. 404 CHONG AND LÓPEZ-DE-SILANES Fama, Eugene, and Michael Jensen. 1983. "Separation of Ownership and Control." Journal of Law and Economics 26 (2) (June): 301­25. Freeman, Richard. 1986. "Unionism Comes to the Public Sector." Journal of Economic Literature 24 (March): 41. Galal, Ahmed, Leroy Jones, Pankay Tandon, and Ongo Vogelsang. 1994. Welfare Consequences of Selling Public Enterprises. Oxford, U.K.: Oxford University Press. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "The Benefits of Privati- zation: Evidence from Mexico." Quarterly Journal of Economics 4: 1193­242. La Porta, Rafael, Florencio López-de-Silanes, and Andrei Shleifer. 2002. "Govern- ment Ownership of Banks." Journal of Finance 57: 265­302. La Porta, Rafael, Florencio López-de-Silanes, and Guillermo Zamarripa. 2003. "Related Lending." Quarterly Journal of Economics 118 (1): 231­68. La Porta, Rafael, Florencio López-de-Silanes, Andrei Shleifer, and Robert Vishny. 1997. "Legal Determinants of External Finance." Journal of Finance 52 (3): 1131­50. ------. 1998. "Law and Finance." Journal of Political Economy 106:1113­55. ------. 2000a. "Agency Problems and Dividend Policies Around the World." Journal of Finance 55: 1­33. ------. 2000b. "Investor Protection and Corporate Governance." Journal of Financial Economics 58: 1­25. López-de-Silanes, Florencio. 1994. "A Macro Perspective on Privatization; The Mexican Program." In S. Levy and L. Svensson, eds., Macroeconomic Aspects of Privatization. Washington, D.C.: World Bank. ------. 1997. "Determinants of Privatization Prices." Quarterly Journal of Economics 112 (4): 965­1025. López-de-Silanes, Florencio, and Guillermo Zamarripa. 1995. "De-regulation and Privatization of Commercial Banking." Revista de Análisis Económico ILADES/Georgetown University 10: 113­64. Maskin, Eric. 1992. "Auctions and Privatization." In Horst Siebert, ed., Privatiza- tion, pp.115­36. Tübingen, Germany: J. C. B. Mohr Publisher. Megginson, William, and Jeffry Netter. 2001. "From State to Market: A Survey of Empirical Studies on Privatization." Journal of Economic Literature 39 (2): 321­89. Modigliani, Franco, and Merton Miller. 1958. "The Cost of Capital, Corporation Finance, and the Theory of Investment." American Economic Review 48 (June): 261­97. Perotti, Enrico. 1995. "Credible Privatization." American Economic Review 85: 847­59. Presidencia de la República. 1982­2003. "Informe de Gobierno." Mexico City: Presidencia de la Republica Mexicana, Direccion General de Comunicacion Social. Rogozinsky, Jacques, and Ramiro Tovar. 1998. "Private Infrastructure Conces- sions: The 1989­1994 National Highway Program in Mexico." In High Price for Change: Privatization in Mexico. Washington, D.C.: Inter-American Development Bank. Secretaría de Comunicaciones y Transportes. 2000. El Sector Comunicaciones y Transportes 1994­2000. Mexico City. PRIVATIZATION IN MEXICO 405 SHCP (Secretaria de Hacienda y Crédito Público). 1982­2003a. Cuenta Pública. Mexico City. ------. 1982­2003b. Situación Económica de México. Mexico City. Shleifer, Andrei, and Lawrence Summers. 1988. "Breach of Trust in Hostile Takeovers." In Alan J. Auerbach, ed., Corporate Takeovers: Causes and Consequences. Chicago: University of Chicago Press. Reprinted in R. Romano, ed., Foundations of Corporate Law, Oxford University Press, 1993. World Bank. 2003. Private Solutions for Infrastructure in Mexico. Washington, D.C.: World Bank. 8 Peruvian Privatization: Impacts on Firm Performance Máximo Torero SINCE THE EARLY 1980S, COUNTRIES AROUND the world have embarked on ma- jor privatization programs. Yet many remain reluctant to privatize, while still more have had to halt ongoing processes of privatization. This is particularly true in developing countries, where state-owned enterprises (SOEs) still ac- count for more than 10 percent of gross domestic product, 20 percent of in- vestment, and about 5 percent of formal employment (Kikeri, Nellis, and Shirley 1994). The aversion to privatization appears to be associated with public distrust of the privatization process. Unions and other traditional op- ponents of privatization have argued that it results in layoffs and poorer serv- ices. Political leaders, meanwhile, fear that the higher profitability of private companies comes at the expense of the rest of society, especially during the difficult transition period from state ownership to private ownership. The transfer from the public to the private sector (Vickers and Yarrow 1988) necessarily implies a change in the relationships between those respon- sible for the firm's decisions and the beneficiaries of the profit flows (the so- cial view and the agency view). In theory, the transfer of property rights leads to a different structure of management incentives, causing changes in mana- gerial behavior, company performance, and quality of service in terms of ac- cess and use. To prove these theories, however, empirical work is crucial. Yet little empirical knowledge is available about how well privatization has worked. There are difficult methodological problems as well as special problems with data availability and consistency. Furthermore, the possibility of bias in the sample selection can arise from several sources, including a gov- ernment's desire to sugarcoat the process by privatizing the healthiest firms first. Megginson and Netter (2001) carried out a detailed review of 22 stud- ies on privatization in nontransitional economies and concluded that Galal and others (1994), La Porta and López-de-Silanes (1999), and the studies 407 408 TORERO summarized in D'Souza and Megginson (1999) are the most solid and per- suasive supporting the proposition that privatization improves the operating and financial performance of firms. The author of this chapter considers La Porta and López-de-Silanes (1999) the finest study of an individual country, since it examines nearly the entire universe of Mexican privatizations. These studies, especially La Porta and López-de-Silanes (1999), investigate whether companies increase profits after privatization and whether privati- zation inflicts significant social losses, and, if so, through which channels. They conclude that the improved performance of privatized firms is the re- sult of significant restructuring efforts, not of market power exploitation nor massive layoffs and lower wages. In other words, firms undergo a harsh re- structuring process following privatization and do not simply mark up prices and lower wages, as many economists have predicted. Deregulation, particu- larly the removal of price or quantity controls and trade protections, is asso- ciated with faster convergence to industry benchmarks. The authors suggest that the sale profits and increased tax revenues the government receives from privatizations are probably enough to offset the cost of job losses to society. According to these studies, newly privatized firms cut employment, usually reducing the roll of white- and blue-collar workers by nearly half. These numbers may actually underestimate the effects of privatization, since in the preceding years most companies have already trimmed pay- rolls to prepare for divestiture. These findings suggest that transfers from workers to shareholders play a role in the success of privatization. Even so, productivity gains resulted in large real wage increases of 114 percent in the postprivatization period. La Porta and López-de-Silanes showed, for example, that privatized firms increased sales 54.3 percent, despite work force reductions and only modest increases in capital. Surprisingly, prices rose only 2.9 percent relative to the producer price index. La Porta and López-de-Silanes also decomposed re- ported increases in profitability. Approximately 10 percent of the gain in profits was attributable to higher prices and 33 percent to worker layoffs, while productivity gains accounted for the remaining 57 percent. Some of the social effects of higher prices and layoffs were offset by corporate taxes, which absorbed slightly more than half of the gains in operating income. In this study, the author follows an approach similar to that of La Porta and López-de-Silanes (1999) by collecting information on nearly the entire population of privatized firms in Peru to evaluate the impact of privatiza- tion there. The author then compares the performance of those firms with the remaining SOEs and, when possible, with industry-matched private firms. Through this method, the impact of privatization on profitability ratios, operating efficiency ratios, labor indicators, and capital-deepening indicators is analyzed. Even though the ultimate effect of changes in man- agement incentives depends on the competitive and regulatory environment in which a given firm operates, it is argued that the degree of market com- petition and the effectiveness of regulatory policy have more important PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 409 effects on performance than does change of ownership (Vickers and Yarrow 1988). This is extremely important in the case of the Peruvian pri- vatization process because it was accompanied by large-scale sectoral reforms in which competitive structures and independent regulatory agen- cies were established to monitor and promote competition in each sector. Therefore, variables needed to identify the roles played by the regulatory agencies and the competitive forces that determine firm performance (exis- tence of a regulatory framework, autonomy of the regulatory agency, and so forth) are taken into account in the analysis. Peru's privatization experience was rated one of the early success sto- ries in Latin America. The privatization process, begun by then-president Alberto Fujimori, was launched as part of a rigorous process of stabiliza- tion and structural reform initiated in response to a crisis in the Peruvian economy. At the time, inflation had reached an annualized rate of 36,000 percent, and per capita income had dropped to its lowest level in 30 years. Although privatization was not part of the initial set of reforms, it soon became a central plank of the overall reform program. By 2001 the privatization process had involved 252 transactions, in- cluding 42 SOEs; had brought $9.2 billion in revenue (including capital- ization) to the national treasury; and had mobilized an additional $11.4 billion in new investments.1 Nevertheless, Peru's considerable success in at- tracting private participation and capital was focused on a few sectors such as telecommunications, electricity, banking, hydrocarbons, and mining. Unlike some countries, such as Argentina and Bolivia, there has been vir- tually no private participation in the transportation, water, or sanitation sectors. Furthermore, as is the case in other countries, public support for privatization has been declining steadily--from 65 percent in May 1992 to less than 25 percent by 2000. This decline has brought the privatization process nearly to a halt. This report, written during a period of antipriva- tization sentiment, is of special importance because it aims to analyze em- pirically the impact on performance of privatized SOEs. This study reviews the privatization process and its principal results. It then summarizes the empirical methodology followed by the author, de- tails the database developed, and presents the results of the calculations of the differences in pre- and postprivatization performance, difference-in- difference comparisons for which control groups were developed, and a panel data regression analysis of the static and dynamic performance of privatized firms relative to SOEs. The Privatization Process At the beginning of 1990, Peru faced its worst macroeconomic situation ever.2 The country had never experienced such large and prolonged periods of inflation and recession. The economic model implemented in response 410 TORERO to the crisis assigned the state a central role in economic policymaking. The policies adopted by the government were not up to the challenge at hand: public expenditure and public internal credit rose impressively, price con- trols and subsidies were established, tariffs on public services were fixed, and exchange rate controls were set. These policies translated into a per- sistent fiscal imbalance, a considerable drop in tax revenues and a striking decline in financial intermediation.3 The macroeconomic crisis was hardest on Peru's poorest citizens, around 43 percent of the 1990 population. The situation worsened as pub- lic services, such as education and health, deteriorated. In addition, by the end of the 1980s, informal economic activity, delinquency, drug traffick- ing, and terrorism had all increased. In 1990 Peru reached record under- employment (86.4 percent), while unemployment was around 8.3 percent, and formal employment barely reached 5.3 percent.4 In this context, public enterprises were characterized by inefficient pro- vision of goods and services, ambiguous objectives, extensive intervention by politicians, decapitalization of investment resources, and a lack of fresh investment. Not surprisingly, public firms registered accumulated losses of more than $4 billion in 1989­90.5 In an effort to reverse this situation, the Peruvian government decided to design an attractive normative and insti- tutional framework to promote private investment as the main vehicle of economic growth. One of the key aspects of this new framework was a program to privatize public-sector companies in 1991. In February 1991 the privatization process was launched with the enact- ment of Supreme Decree 041, which regulated and restructured the manage- rial activity of the state, even though the state was limited to managing no more than 23 companies. In November 1991 the government extended more active and decisive support for the privatization process by enacting Legisla- tive Decree (LD) 674, also known as the Promotion of Private Investment in State Companies. LD 674 introduced the Commission for the Promotion of Private Investment (COPRI) and the Special Privatization Committees (CEPRIs), as well as private investment promotion schemes, which included sales of stocks and assets, service provision, concessions, and other items. One of the most important laws enacted was LD 662, or the "Law of For- eign Investment Promotion," which mandated equal treatment of national and foreign capital. This law permitted foreign investment in all economic sectors and its execution through any legal administrative means. To give more dynamic and political support to the process, President Fujimori appointed five state ministers to lead COPRI. These ministers were in charge of the general management of the privatization process; they had to establish the policies and objectives of the process, appoint the CEPRIs that were responsible for the planning and execution of individ- ual privatizations, and approve the most important decisions. Diverse reforms were instituted in 1992 to facilitate the privatiza- tion process. The government was authorized to grant the safeguards and PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 411 guarantees necessary to protect foreign acquisitions and investments. For- eign investors were also granted facilities for the payment of taxes and debts owed by SOEs in the privatization process. In some cases, these commit- ments were suspended until the end of the process. In 1993 all of these reforms were codified into law with the approval of the new Political Constitution. The new constitution included the promotion of free private initiative, the establishment of equality between national and foreign investors, the encouragement of competition and equal treatment for all economic activities, and a guarantee of the possibility of the signing of sta- bility agreements between private investors and the state. In addition, the state subscribed to many international agreements for the protection of for- eign investment and conflict resolution through international arbiters. Together with the launching of the privatization program, the govern- ment undertook another set of structural reforms. Through these reforms, the government promoted market-based competition and free interna- tional trade, installed policies to create a more flexible labor market, lib- eralized the financial system, eliminated price controls, and implemented sector reforms for the deregulation of markets. All the reforms carried out were complementary and necessary to the privatization program. In so do- ing, the government recognized that adequate regulatory and institutional frameworks and a competitive market, not just ownership, were deter- mining factors in the success of the privatization process. Peru's privatization scheme began in earnest between November 1991 and February 1992. Its main objective was simple: privatize as many pub- lic companies as quickly as possible. The initial tasks were defining priva- tization methods, prioritizing the public enterprises to be privatized (which depended on their importance and the ease with which they could be privatized), and creating the CEPRIs.6 Because of its transparent and competitive scheme, the public auction was the most common practice adopted for privatization. In the following years, the design of an appropriate juridical-legal framework for the development of private investment continued. One par- ticularly important law provided for the regulation of immigration appli- cations and facilitated the nationalization of foreign citizens who wanted to provide capital and invest in Peru. The results of the privatization process were outstanding. Beyond the simple transfer of assets, companies were purchased and significant amounts of investment were committed (see table 8.1 for details). In 1991 two public companies were privatized (Sogewiese Leasing and Buenaven- tura Mine). In 1992, under an operational COPRI and various CEPRIs, 10 SOEs were privatized, drawing in revenues of $208 million and another $706 million in projected investment. In 1993 the process gathered mo- mentum, with 13 companies privatized for a total of $317 million and projected investment of $589 million. The next year the government sold its natural monopolies in the telecommunications and electricity sectors, 412 Table 8.1 Privatization Revenues and Investment, 1991­2001 (US$ million) Transactions Option Sale of shares rights, small Year and assets Concessions assets, and others Capitalizations Total Projected investment 1991 2.6 2.6 0.0 1992 207.5 1.4 208.9 706.0 1993 316.7 20.7 6.5 343.9 589.3 1994 2,579.2 4.7 610.8 3,194.7 2,050.0 1995 1,089.0 6.6 9.1 120.1 1,224.8 70.1 1996 2,281.8 344.2 2.7 40.0 2,668.7 2,695.0 1997 447.1 99.0 8.8 126.4 681.3 706.2 1998 251.6 35.1 5.2 291.9 220.6 1999 285.8 10.9 3.1 299.8 323.3 2000 95.1a 209.9a 317.0b 4,480.0 2001 19.5a 65.4a 261.0b 98.0 Total 7,575.9 791.8 41.5 897.3 9,494.6 11,938.5 Note: Sale of shares and assets includes resources obtained through private purchases of public firms; concessions are resources acquired through concession contracts with private agents; option rights, small assets, and others are resources obtained through sales of minor assets like machines or equipment from companies; and capitalizations are resources assigned to raise a firm's equity. a. The information for these items is incomplete because of unavailability of data, particularly for the year 2001. b. These numbers are gross amounts for annual totals, so they do not equal the sum of the first four columns. Source: PROINVERSION. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 413 which resulted in $2,579 million in revenues collected and a total amount of $2,050 million of projected investments. During 1995 and 1996 the privatization process accelerated and deep- ened. Sixty-four companies were privatized, producing revenue of $3.4 billion and investment commitments of $2.8 billion. This continued in 1997 when 25 more companies were transferred for $447 million and pro- jected investments of $706 million. In 1998 the privatization process made way for the concessionary process of transportation infrastructure.7 CEPRIs were created to handle the concessions of airports, ports, road networks, and mobile telephone bands, among other facilities. Between 1991 and 2001, Peru's privatization and concessionary processes generated revenues totaling $9.5 billion (including capitalizations) and in- vestment commitments of approximately $11.45 billion. Figure 8.1 shows the evolution of the privatization process and the timeline for the transfer of nearly all of the public enterprises since 1991. Figure 8.2 shows the revenues and projected investment resulting from privatization outlined by sector.8 Al- together 203 privatization operations were carried out, representing $7.85 billion in revenues and $6.4 billion of investment commitments. Concessions raised $726 million in revenue and $4.60 billion in investment commitments. Most of the privatization process occurred in the telecommunica- tions, electricity, finance, mining, and hydrocarbons sectors. Figure 8.3 shows the percentage privatized in various sectors. Telecommunications and finance are already entirely privatized. Those two sectors, along with electricity and mining, represent more than 80 percent of privatiza- tion revenues collected by the Peruvian government. Yet, to date, there has been virtually no private participation in the transportation, water, or sanitation sectors, and there are still sectors, such as agriculture, where much remains under public ownership. Despite the increase in government revenues and investment commit- ments, public approval of the privatization process has decreased steadily, as shown in figure 8.4. Therefore, to develop a complete picture of the im- pact of privatization on other fundamental areas of the Peruvian economy, the results of previous privatization studies must be complemented by a detailed analysis of the impact on firm performance. This chapter evaluates the privatization process by analyzing the per- formance of all privatized firms in Peru. The report studies a sample of firms representing 63 percent of the privatized SOEs and 91 percent of the transactions involved in the privatization process. In addition, this study analyzes in detail the three sectors where most of the privatization took place: telecommunications, electricity, and financial services. In the telecommunications sector, the Peruvian government sold both Compañía Peruana de Teléfonos (CPT) and Empresa Nacional de Tele- comunicaciones (ENTEL). CPT provided basic telecommunications services in the Lima area, and ENTEL provided national and interna- tional long-distance services, as well as local service for the rest of Peru. 414 TORERO Figure 8.1 Evolution of the Privatization Process, 1991­2000 Percent 40 35 30 25 20 15 10 5 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000a Number of projects Sales of shares or assets Projected investment Note: The bars represent the number of projects completed in one year as a percentage of projects privatized between 1991 and 2000, the amount of sales in one year as a percentage of total sales of shares or assets made between 1991 and 2000, and the amount of projected investments in one year as a percentage of total investments made between 1991 and 2000. a Through June 30. Source: PROINVERSION. Divestiture took place in 1994 after an auction to the highest bidder. Us- ing a first-price sealed bid mechanism, approximately 35 percent of CPT and ENTEL common shares (the minimum required to give the buyer control of the merger) were sold to Spain's Telefónica de España.9 The results of the auction were impressive: Telefónica paid a little more than $2 billion, far larger than the second highest bid of $800 million--a bid that was closer to the base price set by the government. Soon after buy- ing both companies, Telefónica de España merged them and created Tele- fónica del Perú S.A. (TdP). Initially, TdP was granted a five-year national monopoly for the provision of lines, local calls, national long-distance, and international long-distance throughout the country.10 Simultaneously, the government created the Supervisory Agency for Private Investment in Telecommunications (OSIPTEL). PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 415 Figure 8.2 Privatization Revenues by Sector, 1991­99 (US$ million) Telecommunications Electricity Mining Hydrocarbon Industry Financing Fishery Transport Tourism Agriculture Other 0 0 0 0 50 1,000 1,50 2,000 2,500 3,000 3,50 4,000 4,500 5,000 Projected investment Transactions Source: PROINVERSION. The privatization of TdP continued over the following years. In 1996, 65 percent of the company's shares were divided between minor share- holders (36.3 percent) and the Peruvian government (28.7 percent). The latter decided to sell 26.6 percent of its shares to small individual investors 416 TORERO Figure 8.3 Privatization Process Progress, 1991­2000 Telecommunications Electricity Mining Hydrocarbon Industry Financing Agriculture Other 0 10 20 30 40 50 60 70 80 90 100 Percent Source: PROINVERSION. through a process known as Sistema de Participación Ciudadana (Citizen Participation System). In total, privatization in the telecommunications sector raised $3.6 billion in revenues and $1.56 billion in investment com- mitments for the government. For the electricity sector, the government approved in 1992 the Law of Electric Concessions (LD 25844), which split power generation from elec- tricity distribution and transmission. Power generation is a market open to competition, whereas transmission and distribution are usually considered natural monopolies. Between 1994 and 1997, the government privatized 10 electricity SOEs (5 in distribution and 5 in generation) for a total of $1.43 billion. There was also a significant investment commitment to in- crease the total capacity of the privatized generation companies by 560 megawatts (MW). At present, the privatized companies represent 64 per- cent of the total power generation capacity of the national electric system and 79 percent of the distribution service. The government also created two regulatory bodies for the electricity sector: the Supervisory Agency for Private Investment in Energy (OSIN- ERG) and the Commission of Energy Tariffs, which was absorbed later by OSINERG. The privatization process in this sector is not yet concluded because one of the major generating enterprises in the south of Peru, Cen- tral Hidroelectrica del Mantaro, and all of the region's distribution enter- prises are not yet privatized. Even though the privatization is incomplete, PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 417 Figure 8.4 Public Approval of Privatization, 1991­2000 Percent 70 60 50 40 30 22% 20 10 1991 1992 1993 1994 1995 1996 1997 1998 1999 Rate of approval of privatization Note: Values refer to December of the corresponding year. Source: PROINVERSION. the electric sector has already become the second largest generator of pri- vatization revenues and investment commitments for the state: $2.3 bil- lion in revenue and $716 million in investment commitments. Water and sanitation is the only public utility where no privatization has occurred, although a concession was given to the Italian company Impregilo to operate wells and a water treatment plant in the Chillón river basin to sell water to the Lima water company. Instead of priva- tizing, the government sought to improve the organization and man- agement of the water and sanitation system by decentralizing it. This new reform gave municipalities control over water services. The only exception was the most important municipal water service, Empresa de Servicio de Agua Potable y Alcantarillado de Lima (SEDAPAL), which remains a state company. SEDAPAL was the only water-service provider included in the privatization program, but it has yet to be pri- vatized. Despite this, the government has tried to improve SEDAPAL's services and coverage. Additionally, in 1992 the government created the Superintendencia Nacional de Servicios de Saneamiento (SUNASS), the National Office for Services of Sanitation, as the regulatory body for this sector. SUNASS is responsible for controlling the quality of the 418 TORERO service provided, the pricing system, and regulation, as well as intersec- tor coordination, establishment of norms for the execution of invest- ment plans, and supervision of those plans. With respect to the financial sector, on July 20, 1994, 99.86 percent of the government's shares in Interbanc were auctioned. The winner was a consortium formed by International Financial Holding (Grand Cayman) and IFH Peru S.A., with the advice of Banco Osorno and La Union (Chile), for $51 million (workers paid $4.83 million of this for a total of 9.46 percent of the bank's shares). Interbanc branches Financiera Peruana (Interfip), Internacional de Inmuebles, and Empre- sas de Servicios Internacionales (Interserv) were also included in the auction. And on April 18, 1995, 60 percent of the shares of Banco Continen- tal were awarded to the consortium formed by Banco Bilbao Vizcaya (Spain) and the companies Inversiones Breca, Inversiones San Borja, Ur- banizadora Jardin, and Minsur (all belonging to the Brescia Group of Peru). In August 1995, in agreement with the share purchase sale con- tract of Banco Continental, 15,325,388 shares belonging to the state were transferred to Holding Continental S.A for $32 million.11 By July 21, 1998, the Peruvian government had managed to sell 19 percent of Banco Continental's remaining shares on the international and local markets. Methodology Following Boubakri and Cosset (1998), the analysis conducted in this chapter tries to determine whether firms improve their performance after privatization. Firm performance is measured by profitability, operating ef- ficiency, capital expenditures, output, employment, and leverage. Table 8.2, taken from Megginson, Nash, and van Randenborgh (1994), shows details on the proxies for these performance measures as well as the pre- dicted relationships. Based on these performance measures, the empirical approach consists of two stages. In the first stage, a simple statistical analysis is conducted to study the postprivatization changes in firms' performance. In the second stage, a regression analysis is performed, controlling most of the differ- ences between firms and variables, other than privatization, that could ex- plain the performance of the firm. The statistical analysis consists of computing the performance variables for each company in the sample for a 15-year period (1986­2000). Then the means for each performance variable (Y) for the preprivatization and postprivatization periods are computed. To avoid any bias resulting from a preprivatization restructuring of the firm, all years in which re- structuring took place before the divestiture are excluded.12 After the Table 8.2 Firm Performance Measures Performance measure Proxies Predicted relationship Profitability Return on sales (ROS) net income/sales ROSA ROSB Return on assets (ROA) net income/total assets ROAA ROAB Return on equity (ROE) net income/equity ROEA ROEB Operating Sales efficiency (SALEFF) sales/number of employees SALEFFA SALEFFB Efficiency Net income efficiency (NIEFF) net income/number of employees NIEFFA NIEFFB Capital Capital expenditure to sales (CESA) capital expenditures/sales CESAA CESAB Investment Capital expenditure to assets (CETA) capital expenditures/total assets CETAA CETAB Output Real sales (SAL) nominal sales/consumer price index SALA SALB Employment Total employment (EMPL) total number of employees EMPLA EMPLB Leverage Debt to assets (LEV) total debt/total assets LEVA LEVB Long-term debt to equity (LEV2) long-term debt/equity LEV2A LEV2B Payout Dividends to sales (DIVSAL) cash dividends/sales DIVSALA DIVSALB Dividends payout (PAYOUT) cash dividends/net income PAYOUTA PAYOUTB Note: This table presents some of the most common firm performance measures found in the literature (see appendix table 8B.1 for variable def- initions). Each type of measure is associated with one or more proxy variables (column 2). For each proxy there is a predicted relationship between two firms (column 3). Firm A (the postprivatization firm) would have a better performance than firm B (the preprivatization firm) if the predicted relationship stated is satisfied. Source: Megginson, Nash, and van Randenborgh 1994. 419 420 TORERO means are calculated, using differences from the sample counterpart of the privatization effect and the performance variables, the following is obtained: ¢Y [Ypostprivatization Ypreprivatization] The two-tailed Wilcoxon signed-rank test and the Hotelling test are then used to test for significant changes in performance variables after pri- vatization. Both tests are based on the assumption that the distributions are normal. If the sample size is small and the true distribution of differ- ences is far from normal, the stated probability levels may be significantly in error. Specifically, when looking at each individual firm, the central limit theorem cannot be applied since the sample of years for each is small. For that reason, it is necessary to verify the normality of the series. There- fore, the Shapiro-Francia test for normality is used. When the Shapiro- Francia test rejects the null hypothesis of normality, a nonparametric test, the Kolmogorov-Smirnov (K-S) statistic, is used to formally test the equal- ity of the empirical hazards functions of the different pre- and postpriva- tization performance indicators.13 The above methodology is equivalent to considering the simplest pos- sible model for capturing the effect on performance with no regressors; it can easily be derived so that performance depends only on the date of the privatization dummy, Yi,t Privatizationi,t ui,t E(ui,t/Privati,t) 0 (8.1) Nevertheless, the above result is likely to be biased for two reasons. First, the two groups of years may have different characteristics and thus different performance behavior. Second, the two groups of years may be subject to different shocks. Part of the differences in pre- and postprivati- zation performance patterns may simply be a result of these differences. An alternative way to solve this problem is to develop a benchmark to control for these different characteristics and shocks. In this sense, a dif- ference-in-difference measure is calculated for each economic sector in which privatization is important: ¢2Y [Ypost privyear Ypre privyear priv. firms ] [Ypost privyear Ypre privyear notpriv. firms ] (8.2) The main caveat of the difference-in-difference measure is the lack of an appropriate control group with which to compare the difference in per- formance of the privatized firms. It is not possible to use an optimal matching methodology such as propensity scores, as detailed in Rubin (1974, 1977, 1979), Heckman and Smith (1995), Heckman and others PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 421 (1996), Heckman, Ichimura, and Todd (1997), and Heckman, Lalonde, and Smith (1999), because in all sectors under analysis, except banking, there are not enough cases to find the appropriate control group. For this reason, the author tries to reduce this problem by complementing the above equations with a regression analysis. The regression analysis added to equations 8.1 and 8.2 incorporates into the model regressors that control for observable characteristics at the firm level. It also includes sectoral and macroeconomic variables. The lat- ter variables try to capture different shocks, thereby isolating the impact of privatization. The regression analysis consists primarily of an attempt to model each of the performance measures (P) as a function of the following variables: Yi,t f(Xi,t, Ti,, Pi,t, Pi,r, Sj, Rj, Zt) where Yi,t are the different performance measures previously detailed for firm i in period of time t; Xi,t are firm characteristics; Ti are the characteristics of the privatization process of the specific firm; Pi,t is the date in which the firm was privatized or given in concession; Pi,r is a dummy indicating whether the firm is privatized or not; Sj are variables at the sector level of the firm; Rj are characteristics of the regulatory agency (for details, see Guasch and Spiller 1999), and Zt are other controls such as macroeconomic variables. Additionally, the author explored interaction effects of the privatiza- tion dummy, and carried out panel estimations using differences to drop out all firm-observed and -unobserved time-invariant fixed effects. There- fore, the three econometric specifications regressed are: Pi,t 0 0 i,t P 1t 2 i,t X 3 j S 4 i T 5 j R 6 1 Z i,t (8.3) Pi,t 0 0 i,t P 0 i,r P 1t 2 i,t X 3 j S 4 i T 5 j R 6 1 Z i,t (8.4) Pi,t 0 0 i,t P 0 i,r P 1 t 1 tPi,r 2 i,r P t 2 i,t X 3 j S 4 i T 5 j R 6 1 Z i,t (8.5) Equation 8.3 is the same as equation 8.1 but includes firm, sector, and macroeconomic variables and, when available, some variables for the characteristics of the regulatory agency. Equation 8.4 includes a privati- zation dummy and a control group in the sample to be able to carry out a difference-in-difference estimation as in equation 8.2, but again with the controls previously specified. Finally, equation 8.5 includes additional in- teractions of the year-privatized dummy (Pi,t) and the dummy of whether the firm is privatized or not (Pi,r), and a time trend to capture trend and convergence over time of newly privatized firms with firms in the control group (public firms or already private firms). 422 TORERO Equations 8.3, 8.4, and 8.5 are estimated using simple ordinary least squares (OLS) panel data of firms, as detailed below. In addition, it is nec- essary to account for unmeasured industry and industry/year effects. Mak- ing establishment fixed effects allows all firm-observed and -unobserved time-invariant fixed effects to drop out. However, since these performance models could suffer from endogeneity problems, simultaneous determina- tion and reverse causality of the explanatory variables are used, following what is now standard procedure in the literature of instrumental vari- ables.14 Because the privatization process directly affects most of the ex- planatory variables for many of the performance indicators, reverse causality or simultaneous determination is a latent problem. Additionally, the GMM-IV (generalized method of moments instrumental variables) es- timation allows for heteroskedasticity of unknown form.15 In order to have appropriate instruments, the lags of the instrumentalized variables as well as the privatization variables are used. Also, a test of overidentifying restrictions--Hansen's J statistic (1982)--is provided to check whether the equation is overidentified by an abundance of instruments. If this sta- tistic rejects the null hypothesis, the validity of the model is called into question.16 Although this estimator is restricted to models linear in the parame- ters, it is relatively more efficient than an OLS with instrumental vari- ables, even with correction for heteroskedasticity with the White proce- dure. The efficiency gain is derived from the GMM-IV estimator's use of an optimal weighting matrix (rather than the identity weighting matrix implicit in any least squares estimator) to define the appropriate combi- nation of moment conditions.17 In this context, the moment conditions are the orthogonality conditions of each instrument with the error process. A discussion of the development of the estimator is given in Greene (2000). The Data The construction of the database required several sources of statistical in- formation. For the preprivatization period, the primary sources of infor- mation were the White Books (a collection of all information available for firms to be privatized) and the published histories of the respective firms. This information was complemented by sources such as the fiscal statisti- cal summary of the Central Reserve Bank of Peru, the National Institute of Statistics and Computing, annual economic surveys, and others (for fur- ther details on data sources, see appendix 8A). For the postprivatization period, the author collected information from various sources. Again, company histories were an excellent source of in- formation. Data on the characteristics of the firms were complemented by PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 423 statistics from the Supervising Committee of Companies and Securities, annual economic surveys, and monthly financial reports of the Bank and Insurance Superintendent. The author also collected data for sectoral in- dicators from statistics published by the regulatory agencies. To provide enough preprivatization sectoral data, the data collected include infor- mation for the period 1986­2000. The preprivatization data allowed the author to control for the period of restructuring that many enterprises ex- perienced before they could be privatized. One major problem was the merger, absorption, or division of many companies or business units during the privatization process. This sort of restructuring made it difficult to follow companies as a single unit through the privatization process. Two alternative methods were adopted to resolve the issue. In the first method, the author aggregated prepriva- tization accounting information provided in the White Books; the second alternative relied on the fact that in most of the privatization agreements, such as the merger of CPT and ENTEL into TdP, the privatized entities were required to keep separate accounting books. Thus, the author could either aggregate a company's important data or follow the merged unit over time. A second problem of data collection was that some privatization strate- gies required that SOEs be divided and each individual unit offered sepa- rately. The registers kept before the privatization processes were based on aggregated data, since all the different business units operated as a single enterprise. However, from the day of the decision to privatize, the regis- ters were kept separately for each unit and then consolidated into one record for the company. Although the companies were considered single units after the privati- zation process, in some cases there existed a combination of private busi- nesses and SOEs that were only partially privatized. Mixed ownership in a firm's record complicated the measurement of the impact of privatiza- tion. To control partially for this problem, the author generated a variable based on the percentage of the firm still owned by the government to measure the intensity of the privatization process and added a discrete dummy of the period in which the privatization started. The final problem was that parts of the SOE portfolios had been liqui- dated. Those companies usually represented inefficient units of SOEs that had not been absorbed by the private system. In these cases, when possi- ble, the unit of the company liquidated was excluded, or it was assumed that the new private owner decided to shut down that unit for efficiency reasons. Between the years 1992 and 2000, 185 transactions took place. This process included 42 privatized SOEs. However, for several important rea- sons, the sample of companies in this chapter does not include all of the privatized SOEs: 424 TORERO · Some state companies were divided horizontally or vertically into small units and privatized separately. In most cases, it was possible to join all the parts in which the company was divided and to assume that it remained a single operating unit. In the case of Telefónica del Perú, in- formation from CPT and ENTEL Perú has been added. This was not possible in other cases due to lack of information about some of the units into which the company was divided. · Most of the concessions and projects were not included due to the lack of financial information from the preprivatization period. · Several firms were liquidated or had their operations stopped. · Some firms were absorbed, or another firm acquired some of their business units.18 · In some cases, information was unavailable. Despite these limitations, the sample includes 86 percent of the total value of transactions undertaken and 47 percent of the privatized SOEs involved in the process. These percentages increase to 91 percent and 63 percent, respectively, when liquidated or extinct companies are not con- sidered. Table 8.3 presents the set of nonfinancial SOEs included in the study; nonfinancial companies not included in the study are listed in table 8.4. A separate database was constructed for the financial sector. It consists of annual data and considers private banks during the period as the con- trol group and the Banco de la Nación as a state-owned enterprise. The privatized banks are detailed in table 8.5. The evolution of privatization in the financial sector is given in appendix table 8C.2. Appendix table 8B.1 gives a detailed explanation of the variables constructed and the manner in which they were calculated. Figure 8.5 plots all of the performance indicators for the entire database of priva- tized firms using a nonparametrical approximation (kernel densities) for the distribution of the values of the pre- and postprivatization per- formance indicators.19 A clear increase (larger for the privatized firms than for the SOEs) of the performance indicators can be seen since 1994, when the process of privatization accelerated. For some indica- tors, such as return on assets, the difference between SOEs and priva- tized firms is not clear. This could occur because privatized firms significantly increased their possession of assets, which, in turn, reduces the impact of an increase in sales. When looking at employment, in- come, sales, and asset efficiency, the positive impact of privatization on the efficiency of firms is even more apparent, even though the reduction in total number of workers is similar for both SOEs and privatized firms. Furthermore, after analyzing the performance indicators for each in- dividual firm, it becomes apparent that the distributions of privatized firms shifted to the right for practically all of the performance indica- Table 8.3 Nonfinancial Companies Included in the Study Firm data availability (years) Under public Under private Name of state-owned enterprise Name of private firm ownership ownership Electrolima Edelnor 1986­93 1994­99 Luz del Sur Edegel Ede ­ Cañete Ede ­ Chancay Electroperú Electroperú 1986­94 1995­99 Egenor Egesur Cahua Empresa Eléctrica de Piura Empresa Eléctrica de Piura -- 1997­99 -- Electro Andes -- 1997­99 Electro Centro Electro Centro 1986­98 1999 Electro Noroeste Electro Noroeste 1986­98 1999 Electro Norte Electro Norte 1986­98 1999 Electro Norte Medio Electro Norte Medio 1986­98 1999 Electro Oriente -- 1986­99 -- Electro Sur -- 1986­99 -- Electro Sur Este -- 1986­99 -- Electro Sur Medio Electro Sur Medio 1986­96 1997­99 Etevensa Etevensa 1994­95 1996­99 425 Seal -- 1986­99 -- Cemento Sur Cemento Sur 1986­89, 1994 1996­98 (Table continues on the following page.) Table 8.3 (continued) 426 Firm data availability (years) Under public Under private Name of state-owned enterprise Name of private firm ownership ownership Cementos Lima Cementos Lima 1987­93 1994­2000 Cementos Norte Pacasmayo Cementos Norte Pacasmayo 1992­93 1994­2000 Cemento Yura Yura 1986­90 1994­95 Centromín -- 1986­90 -- Sociedad Minera Cerro Verde Sociedad Minera Cerro Verde 1993 1994­96, 1999­2000 Compañía Minera Condestable Compañía Minera Condestable 1987­90 1992­2000 Hierro Perú Shougan Hierro Perú 1986­90 1998­99 Minero Perú -- 1986­90 -- Empresa Minera Especial Tintaya -- 1986­89 -- Empresa de la Sal Empresa de la Sal 1991­94 1995­2000 Petroperú Petroperú 1986­91 1992­98 Petroperú - Refinería la Pampilla Refinería la Pampilla -- 1996­98 Química del Pacífico Química del Pacífico 1988­92 1993­2000 Certificaciones del Perú Certificaciones del Perú 1991­93 -- Reactivos Nacionales Reactivos Nacionales 1987­89, 1991­92 1993­2000 Industrias Navales Industrias Navales 1991­92 1993­96 Sudamericana de Fibras Sudamericana de Fibras 1991­92 1993­96 Siderperú Siderperú 1986­90, 1993­95 1996­97 Solgas -- 1986­90 -- Compañía Peruana de Teléfonos Entel Perú Telefónica 1986­93 1994­2000 SEDAPAL -- 1986­99 -- -- Not applicable. Source: Author's data. Table 8.4 Nonfinancial Companies Not Included in the Study Absorbed by Ceased or liquidated Divided Land privatizations another firm Information not found Minpeco USA Sociedad Paramonga Proyecto Especial Lar Carbón Petrolera Transoceánica Aeroperú Epsep Chavimochic Sia Refineria Cajamarquilla Petromar Tierras del Proyecto Nisa Pesca Perú Ecasa Especial Pastogrande Planta de Cemento Enafer Flopesca Tierras del Proyecto Rioja Empresa Minera Yauliyacu Pesquera Grau Especial Chinecas Petrolube Empresa Radio Fertisa Tierras del Proyecto Enata Panamericana Epersur Especial Majes-Siguas Empresa Minera Empresa Difusora Radio Plesulsa Tierras del Proyecto Especial Mahr Túnel Tele Metaloroya Jequetepeque-Zaña Empresa Minera Pletasa Amfa Tierras Eriazas Paragsha Planta Lechera de Iquitos Talleres de Tierras del Proyecto Especial Moyopampa Chira-Piura Empresa Minera Cobriza Cedega T Enatru Perú Ertur Arequipa Eretru Ertsa Puno Entur Perú Emturín Kuélap Complejo Pesquero de Samanco Ergusa 427 Incasa Ertur Note: See text for an explanation of why firms were excluded from the study. Source: Author's data. 428 TORERO Figure 8.5 Evolution of Performance Indicators Return on sales Return on assets Index Index .224043 60.6412 -1300.09 -332.535 1985 1990 1995 2000 1985 1990 1995 2000 Year Year Nonprivatized Nonprivatized Privatized Privatized Return on equity Index 63,1283 -934.179 1985 1990 1995 2000 Year Nonprivatized Privatized Total number of workers Sales efficiency Index Index 122.627 199.616 72.3938 99.6661 1985 1990 1995 2000 1985 1990 1995 2000 Year Year Nonprivatized Nonprivatized Privatized Privatized tors. This signifies that the mean value of the specific performance indi- cator is bigger than it was when the firms were SOEs. The profitability ratios and the operating efficiency ratios increased after the privatiza- tion process. It must be mentioned, however, that the positive tendency PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 429 Figure 8.5 (continued) Net income efficiency Assets to employment Index Index 180.942 221.182 -632.605 130.759 1985 1990 1995 2000 1985 1990 1995 2000 Year Year Nonprivatized Nonprivatized Privatized Privatized Debt to assets Debt to equity Index Index 176.941 5305.84 55.5072 41.6735 1985 1990 1995 2000 1985 1990 1995 2000 Year Year Nonprivatized Nonprivatized Privatized Privatized Note: The individual figures show the trend of performance variables between privatized and nonprivatized companies during 1985­2000 (see appendix 8B.1 for variable definitions). The group of privatized companies includes all the firms that were privatized during the period 1985­2000, while the nonprivatized group comprises firms that have remained under public ownership. Values have been estimated using lowess smoothing--KSM (kernel densities). Banks have been excluded from the estimations. in profitability ratios emerged a few years before the actual process began, since many of the privatized enterprises had to undertake a re- structuring process instituted through the implementation of reforms in all these sectors. Additionally, it should be noted that the profitability indicators for the banks under examination showed an important decline a few years after 430 TORERO Table 8.5 Privatized Banks Included in the Study Number of yearly observations Bank Privatization date Preprivatization Postprivatization Continental April 1995 9 6 Interbank July 1994 8 7 Comercio June 1992 6 9 Popular November 1993 6 -- -- Not applicable. Source: Author's data. the privatization process. This result can be explained by the severe global financial crisis at the end of the 1990s. Improvement in the operating efficiency ratios demonstrated not only a recovery in sales and income of companies across sectors, but also the strong decline in postprivatization total employment across sectors. These indicators had a positive, but weak, evolution in the years before the privatization process, but only after privatization occurred did their pace accelerate. The capital-deepening indicator (the ratio of assets to employment) showed a very important increase after the privatization process. In all sectors, this indicator was more or less stagnant before the process oc- curred, but afterward, it started rising very rapidly. Furthermore, lever- age indicators, which had a very negative and unstable tendency before privatization, began to improve, although not immediately. In all sectors the negative tendency was reversed after the privatization process began, but many sectors faced a relapse during the global financial crisis. The means and variances for these ratios during the period of analysis, as well as for the values of the most important indicators for all sectors, can be found in appendix 8D. Even though businesses in the sanitation sector (specifically SEDA- PAL) have not been privatized, they are included in this project as a con- trol group for the privatized firms in the utilities sector. SEDAPAL is used as a control group because it is a utility, like telephones or electricity, and because it had a similar evolution in performance indicators for the preprivatization period. That occurred because the government initially prepared SEDAPAL for privatization. This similarity presents an oppor- tunity to compare an "untreated group" (a firm that was prepared for pri- vatization but then not privatized) with a "treated group" (TdP). In the case of electricity, as detailed in appendix table 8C.1, state-owned firms still exist. Thus, there are enough controls to evaluate the impact of pri- vatization in the utilities sector under the difference-in-difference methodology. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 431 Empirical Results In this section, the methodologies outlined earlier are used to analyze the impact of the privatization process on firm performance. First, a detailed analysis of performance indicators is carried out for all the privatized firms in the sample; then the three major privatized sectors are analyzed: telecommunications, electricity, and the financial sector. Each of the tables consists of two tests comparing pre- and postpri- vatization. The first test is a first-difference analysis using firm and year fixed effects to analyze the difference between pre- and postprivatiza- tion information for all firms under study. The second test is a differ- ence-in-difference test, as detailed in the methodological section. The difference-in-difference statistic both tests for the change in firm per- formance compared with the privatization period and takes into ac- count relative firm performance when compared with a control group of SOEs that did not undergo the privatization process. In the all-firm panel, the control groups are all the SOEs present for every year for which information was collected. The gross national product per capita for each sector is also included to control for the size of the sector to which the specific firms belong. When analyzing the two principal sectors where privatization took place, the control firms were those identified as most similar to the ones under analysis. In the case of telecommunications, the control group is SEDAPAL, the main firm in the water and sanitation sector. This firm was not privatized but underwent a preprivatization reform process similar to that of the telecommunications firm.20 For the electricity sector, the con- trol group is a group of nonprivatized electric companies (Electro Oriente, Electro Sur, Electro Sur Este, and Electro Sur Medio). Finally, for the case of the two privatized banks, Banco Continental and Interbanc, two differ- ent control groups were used. The first group consisted of all private banks in operation between 1986 and 2000; the comparison was between priva- tized banks and private banks. The second control was the state-owned bank, Banco de la Nación. Since there were enough private banks to develop the control group, a propensity score, in which the probability of belonging to the treated group, given observable characteristics (interbank funds, assets, total lia- bilities, and equity),21 was used as a summary of those characteristics in order to measure the average treatment effect on the treated variables in comparison with the performance variables. Finally, a regression analysis and the estimation of equation 8.5 were carried out to find a possible convergence of performance indicators. The regression analysis also allowed controlling for different variables men- tioned in the literature that could explain the impact of the privatization of the SOEs. In addition to characteristics of the firm such as size, sector, 432 TORERO gross domestic product, and assets over employment, controls that helped measure the size of market failure were included. As noted above, and as mentioned by Megginson and Netter (2001), welfare theory argues that privatization tends to have the greatest positive impact in cases where the role of the government in minimizing market failure is the weakest, that is, for SOEs in competitive markets or markets that can readily become competitive. In contrast, Shleifer (1998) and oth- ers have argued that both in natural monopolies and in the markets for public goods, where competitive considerations are weaker, government- owned firms are rarely the appropriate solution. Consistent with this literature, the regression analysis includes a set of variables that control for the degree of competition approximated by concentration indexes, as well as variables that measure the type of reg- ulatory processes that accompanied the privatization process. In addi- tion, and of utmost importance, the inclusion of such variables allows for the separation of the effects of market power from those of privati- zation. It is possible that some companies may have had market power before their sale but did not exploit it because they were under state con- trol. If this is the case, then part of the improvement in firm performance could be a result of exploitation of consumers through market power and not a consequence of efficiency and incentive improvements result- ing from privatization. Results for All Privatized Firms Table 8.6 presents the results for all of the privatized firms in the sample. In the table, first and second differences in performance changes are pre- sented using both the mean and the median. The second differences are presented using as a control group all the firms not privatized in the re- spective periods of analysis. In all performance indicators and as men- tioned in the methodological section, a simple regression was carried out (equation 8.1) in which fixed effects were included at the level of the firm (figure 8.5 plots each of the indicators). Additionally, a test for normality was carried out as well as the Kolmogorov-Smirnov nonparametric test to determine if the difference in the distribution of the performance indica- tors was significant. In all the performance indicators, with the exception of leverage and assets to employee, the test showed significant differences. Even accounting for the wide range of firms included in the study, the performance indicators show a significant improvement after privatiza- tion. Specifically, when analyzing three basic indicators, sales, cost per unit, and labor, the results obtained are as expected: privatized firms sig- nificantly increase their sales compared with nonprivatized SOEs. At the same time, there is a statistically significant reduction in cost per unit, and direct employment falls significantly, in line with the restructuring process that the privatized firms went through. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 433 Table 8.6 Changes in Performance for the Privatized Firms Difference in First difference difference Mean before Mean after t-statistica t-statistica Variable Median before Median after t-statisticb t-statisticb Profitability Operating income/ 0.053 0.187 2.70 2.700 sales 0.083 0.207 5.57 4.850 Net income/sales 0.293 0.028 2.41 2.410 0.128 0.074 0.30 0.300 Net income/PPE 0.062 0.010 1.41 1.410 0.027 0.042 1.56 1.090 Operating efficiency Cost per unit 0.947 0.813 2.70 2.700 0.917 0.793 2.81 2.980 Sales/employeec 110.317 242.909 11.91 11.910 105.089 249.802 2.12 1.970 Log(sales/ 5.336 5.770 10.85 10.850 employees) 5.377 5.789 3.35 2.550 Sales/PPE 1.215 1.007 1.43 1.430 1.167 0.936 0.25 0.260 Log(sales/PPE) 0.443 0.321 1.98 1.980 0.377 0.323 0.14 0.240 Output Log(sales) 6.590 11.636 8.47 8.470 7.596 11.674 1.25 0.350 Labor Employeec 114.621 68.621 13.93 13.930 116.612 67.377 4.88 2.740 Log(employee) 6.145 5.635 14.65 14.650 6.183 5.622 0.28 5.740 Assets Log(PPE) 11.641 11.722 0.73 0.730 11.594 11.768 1.22 0.490 Note: PPE property, plant, and equipment. See text for an explanation of the empirical results; see appendix table 8B.1 for variable definitions. Pre- and postprivati- zation categories are defined according to the privatization date for each company. Second differences (last column) calculated from a panel regression are presented using as a control group all the firms not privatized in the period of analysis. a. Corresponds to the test of significance of an ordinary least squares regression with fixed effects. b. Corresponds to the test of significance of an LMS (Lagrange multiplier statistic) regression with fixed effects. c. Index (year 1993 100). Source: Author's calculations. 434 TORERO Moreover, the profitability indicators and all the operating efficiency indicators show significant improvement for both the privatized firms and in comparison with the nonprivatized firms. The ratio of net income to assets (property, plant, and equipment, or PPE) shows no significant difference between the pre- and postprivatization periods. This result was expected because both the denominator and the numerator increase with the privatization given the high levels of investments made by new com- panies to increase efficiency. Results for Public Utilities Sector. A detailed analysis of the two major privatized sectors (public utilities and finance) is carried out for the firms included in the panel and the different time periods over which they were privatized. As previously mentioned, these two sectors repre- sent about 80 percent of the total revenue collected by the privatization process. The results of estimating equation 8.5 for public utilities (electricity, telephones, and water as a control) are shown in table 8.7. As expected, the results are consistent with the literature: privatized firms are more profitable and productive than are public firms (Boardman and Vining 1989; Vining and Boardman 1992; La Porta and López-de-Silanes 1999). The results of the privatization date dummy are significant only for the debt indicators. At the same time, the time trend is positive and significant, which means that over time all the performance measures improved. This finding, together with the insignificant dummy that cap- tures the date of privatization, suggests that the performance indicators started to improve before the privatization process and that only the debt indicators improved significantly faster after the privatization process. When examining the dummy that captures a firm's privatization sta- tus, through a treatment on the treated type of analysis (similar to the second difference) in the return on sales, debt indicators, sales efficiency, and the ratio of assets to employment, the privatized utilities show a sig- nificant improvement compared with SEDAPAL, which is used as a con- trol group. However, the variable of time trend times the privatized firm dummy is significant and negative, meaning that over time, the priva- tized firms are converging to the lower performance of SEDAPAL. This finding is very important because it sheds light on the decline in recent years in the financial performance of both the telephone company and the electric utilities, which could be a consequence of increased market competition. At the same time, the coefficient on the percent of government par- ticipation has a negative and significant sign for two of the three prof- itability indicators (return on assets and return on sales), although it shows a positive and significant sign in sales efficiency and ratio of as- sets to employment. These results sharply contradict expectations. In Table 8.7 Performance Indicators of Privatized Utilities, Difference in Difference (Generalized method of moments instrumental variables estimator) Net Debt to Debt to Sales income Assets/ Variable ROS ROA ROE assets equity efficiency efficiency employment Dummy for date of 0.5355 0.0788 0.0263 0.5594 1.9622 3.9619 20.346 4,453.08 privatization (fpriv) (0.8867) (0.1070) (0.1706) (0.2133)*** (0.6373)** (324.041) (186.8696) (4209.4528) Dummy if firm is 1.3155 0.0114 0.008 0.4892 1.4444 233.929 40.0353 5,574.28 privatized (epriv) (0.4665)*** (0.0242) (0.0425) (0.0549)*** (0.2800)*** (47.0005)*** (32.4350) (1,771.816)*** Time trend (t) 0.1517 0.0113 0.0176 0.0097 0.0162 34.3105 11.3793 35.6078 (0.0292)*** (0.0038)*** (0.0063)*** (0.0046)** (0.0130) (4.6841)*** (2.7264)*** (84.3298) t * fpriv 0.1455 0.0026 0.0077 0.0502 0.1705 53.5472 13.581 192.7965 (0.0885) (0.0074) (0.0123) (0.0174)*** (0.0525)*** (28.5794) (15.1040) (372.0849) t * epriv 0.1305 0.0067 0.0093 0.0345 0.1115 26.3847 6.3136 443.6183 (0.0425)*** (0.0032)** (0.0061) (0.0071)*** (0.0306)*** (5.0188)*** (3.3763) (148.8121)*** Concentration index 0.004 0.0002 0.0003 0.0000 0.0017 1.1352 0.1334 15.8996 (0.0025) (0.0003) (0.0006) (0.0006) (0.0025) (1.0208) (0.4399) (14.9523) Percent change of 0.0484 0.0012 0.0015 0.0013 0.0033 20.0098 4.6774 72.6033 GDP per capita (0.0141)*** (0.0018) (0.0030) (0.0022) (0.0101) (2.0111) (1.9953)** (66.4124) Percent government 1.1136 20.0448 0.0635 0.1108 0.1387 513.6258 104.3681 6,369.42 participation (0.7313) (0.0184)** (0.0310)** (0.0603) (0.1850) (122.735)*** (78.5154) (1,345.163)*** (Table continues on the following page.) 435 436 Table 8.7 (continued) Net Debt to Debt to Sales income Assets/ Variable ROS ROA ROE assets equity efficiency efficiency employment Dummy regulation 0.4064 0.0602 0.1393 0.1374 0.3629 66.4621 58.5637 626.2109 by price cap (0.2648) (0.0352) (0.0569)** (0.0622)** (0.1887) (77.8893) (36.9691) (1017.4182) Dummy regulation 0.0522 0.0177 0.0137 0.0415 0.0214 222.233 22.2343 21,485.57 based on costs (0.3372) (0.0303) (0.0530) (0.0469) (0.2030) (56.4033)*** (39.7200) (1524.1113) Log(employment) 0.2742 0.0247 0.0463 0.0138 0.0598 104.4154 40.8697 572.6417 (0.1328)** (0.0094)*** (0.0176)*** (0.0182) (0.0818) (19.5389)*** (13.6881)*** (714.1734) Assets/employment 0.0002 0.0000 0.0000 (0.0001) (0.0000)*** (0.0000)*** Constant 4.4631 0.2313 0.4215 0.0884 0.4686 1,450.22 505.2756 1,151.88 (1.5426)*** (0.0860)*** (0.1583)*** (0.1853) (0.6785) (256.104)*** (175.467)*** (6526.9513) Observations 93 93 93 96 96 98 98 93 Pseudo-R2 0.349 0.335 0.335 0.569 0.502 0.741 0.371 0.426 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: ROA return on assets, ROE return on equity, and ROS return on sales. This table reports difference-in-difference panel data GMM-IV (generalized method of moments instrumental variables estimator) results of performance indicators for utilities. SEPADAL, the most important water and sanitation enterprise, is used as a control group. There are eight different regressions with distinct dependent performance variables; see appendix table 8B.1 for variable definitions. Standard errors are in parentheses. Log(employment) has been instrumented using one period lag, fpriv and epriv by GMM-IV. Pseudo-R2 is the R2 using IV regression with ro- bust standard errors. Hansen J-statistic (1982) has been used to test for overidentifying restrictions. In every case the null hypothesis that the additional moment condi- tions are approximately satisfied was rejected, validating the use of the instruments. Source: Author's calculations. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 437 addition, the dummy variable for price cap regulation is significant and positive for the return on equity and on the debt-to-assets ratio, imply- ing that the type of regulation carried out by the regulatory agency also has an important impact on firm performance. Finally, the concentration index is not significant, possibly because there is competition in the elec- tricity sector, especially in distribution, and a monopoly in the telephone sector: the effects of the two different market structures go in opposite directions. When examining each of the privatized firms, the results of the first- and second-difference statistics for the privatized SOEs are completely consistent with the findings shown in table 8.7. In the cases of Tele- fónica del Perú S.A. and Electrolima, both the first and second differ- ences are significant and in the expected direction. In the case of the telephone company, only the difference in means in the leverage indi- cators in both the first and second differences were not significant (see appendix table 8E.1). Similarly for Electrolima, all the performance indicators, including leverage, improved significantly (appendix table 8E.2). This also holds when a control group is included and the second difference is calcu- lated. The profitability ratios moved from being negative on average to positive in magnitudes from 8 percent to 20 percent. Another significant change occurred in debt ratios, which fell more than 50 percent. Fur- thermore, sales efficiency increased by 400 percent, and net income ef- ficiency reached high levels compared with previously negative results. One of the expected explanations for such a significant increase in op- erating efficiency would be the reduction of employment by more than half (55.6 percent) after privatization. This reduction, however, only doubles the ratio and is thus not sufficient to explain the improvement. It is safe to conclude that efficiency gains through restructuring and im- proved incentives play a key role in contributing to the rise in operating efficiency. In summary, the results for Electrolima showed that al- though the reduction in employment affected labor productivity, labor productivity and total factor productivity increased after the privatiza- tion process. The results for Electroperú, which was privatized between 1995 and 1996, two years after Electrolima, are shown in appendix table 8E.3. The results in this case are not as significant as the results for Elec- trolima. Only two indicators improved: operating efficiency, as observed in the first difference, and the ratio of debt to assets, as observed in the difference in difference when the performance of this firm is compared to other firms not yet privatized. The state's assumption of Electrolima's long-term debts just before privatization explains the improvement in the ratio of debts to assets. These unsatisfactory results could be a con- sequence of the incomplete privatization process in this sector. One of the major electricity-generating enterprises, Central Hidroeléctrica del 438 TORERO Mantaro, and all of the distribution enterprises in the south of Peru are not yet privatized. Results for the Financial Sector. The results for simple differences in mean for the financial sector were similar to those of the public utilities sector. There was no significant impact on profitability measures and leverage in- dicators, but there was a significant increase in the operating efficiency and coverage of the privatized banks (Interbanc and Banco Continental). The increase in operating efficiency is mainly explained by the 50 percent reduction in employment, which practically duplicates the indicator of op- erating efficiency. At the same time, the difference-in-difference indicators are similar to the first-difference indicators when public banks are used as a comparison. When a comparison is made with private banks of similar size, the private banks still demonstrate better performance measures than privatized banks. Finally, when the control group was defined with propensity scores, and the difference-in-difference estimation for priva- tized banks against matched private banks was carried out, private banks still performed better. Furthermore, when analyzing indicators specific to banks, such as personnel expenses per employee, bad loan portfolios, administrative expenses, and financial margin per branch, an increase in the perfor- mance was discovered. These results even hold for personal expenses per employee in the difference-in-difference indicators despite the fact that the comparison group is preexisting private banks (see table 8.8). The results of estimating equation 8.5 can be observed in table 8.9. The table shows that, unlike the results for the utilities sector, the date of privatization (fpriv) is significant for all the performance indicators. This finding reveals that there was an important change after the priva- tization process. At the same time, there is no direct difference between the privatized banks (epriv 1) and the already private banks (the con- trol group). These results are consistent with the fact that privatized banks are being compared to banks that were always private, and their performance is therefore expected to be better or similar to the private banks. When analyzing the interaction between the time trend and the dummy for privatized banks (epriv), there is a significant and positive effect for the return on sales, indicating that over time newly privatized banks improve in comparison with banks that were always private. This result reflects a possible convergence of performance. As expected, there is also a negative time trend, which could be explained by the international financial crisis that affected all the banks in the region. Additionally, the interaction be- tween the time trend and dummy for year of privatization is significant and negative, implying that the growth rate of performance since priva- tization is declining. The decline can also be explained by the interna- tional effects of the Japanese and Russian financial crises. However, the Table 8.8 Changes in Performance in the Financial Sector after Privatization (Differences between means and difference-in-difference tests) Difference in Means First Differences difference S-Franciac Kolmogorov- Sector Preprivatization Postprivatization t-testa Hotellingb Hotellingb Prob Z Smirnov Performance measure (Pi) Profitability Return on sales (ROS) 0.0545 0.0784 1.320* 1.743 2.971 0.116 0.476 (0.010) (0.016) Return on assets (ROA) 0.0112 0.0099 0.383 0.147 0.167 0.314 0.994 (0.002) (0.002) Return on equity (ROE) 0.1467 0.1193 0.629 0.395 0.429 0.291 0.994 (0.029) (0.027) Operating efficiency Sales efficiency (SALEFF)d 188.85 404.10 7.122*** 50.725*** 35.071*** 0.125 0.007*** (17.921) (24.582) Net income efficiency (NIEFF)d 10.70 22.60 2.641** 6.973** 12.031** 0.178 0.082** (2.403) (4.249) Employment Total employment (EMPL) 3132.5 1831.2 7.101*** 50.419*** 44.613*** 0.146 0.001*** (95.085) (177.859) Leverage Debt to assets (LEV) 0.9125 0.9175 0.555 0.308 0.758 0.077 0.329 (0.007) (0.002) Debt to equity (LEV2) 12.134 11.189 0.485 0.235 0.733 0.029 0.476 439 (1.432) (0.304) (Table continues on the following page.) 440 Table 8.8 (continued) Difference in Means First Differences difference S-Franciac Kolmogorov- Sector Preprivatization Postprivatization t-testa Hotellingb Hotellingb Prob Z Smirnov Coverage Loans per worker (LOAW) 354.17 1440.36 13.329*** 177.654*** 120.006*** 0.015 0.001*** (40.902) (81.520) Deposit per worker (DEPW) 513.23 1742.60 9.216*** 84.940*** 93.298*** 0.051 0.001*** (52.243) (154.157) Indicators specific to banks Personal expenses per employee 31.45 53.48 6.74*** 45.41*** 22.79*** 0.665 0.007*** (1.98) (2.54) Bad loan portfolio 0.08 0.08 0.26 0.07 0.60 0.671 0.721 (0.07) (0.01) Administrative expenses 0.49 0.42 1.19 1.43 3.91* 0.010 0.216 (0.04) (0.01) Financial margin per branch 1913.80 2374.20 1.27 1.62 1.30 0.724 0.476 (253.46) (145.10) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: This table reports results for two privatized banks, Banco Continental and Interbank First, and second differences for changes in perfor- mance are shown, for each empirical proxy, using the mean values (see appendix table 8B.1 for variable definitions). Columns 1 and 2 present the mean values before and after privatization (the year of privatization is 1995). Columns 3 and 4 show results for two tests for significance in changes in performance after privatization (first differences): t-test (two-tailed Wilcoxon signed-rank test) and the Hotelling test. Column 5 displays signifi- cance of difference-in-difference results using a Hotelling test (control group is based on propensity score matching). Column 6 presents Shapiro- Francia test for normality. Finally, column 7 shows the results for the Kolmogorov-Smirnov (K-S) statistic, a nonparametric test used to formally test the equality of the empirical hazards functions of the different pre- and postprivatization performance indicators. a. t-test for Ho (null hypothesis) for difference between means. Numbers are unequal. (X1 X2) (X1 X2) t S x1 x2 (n1 1) s1 2 (n2 1) s2 2 1 1 B (n1 n2 2) *Bn1 n2 where x is an l k matrix of the means and S is the estimated covariance matrix. b. Test of equality: T2 (x1 x2) S1(x1 x2) c. Shapiro-Francia test for normality. Ho (null hypothesis) variable is normally distributed. d. Thousands of nuevos soles. 441 442 Table 8.9 Performance Indicators of Privatized Banks, Difference in Difference (Generalized method of moments instrumental variables estimator) Debt to Debt to Sales Net income Assets/ Variable ROS ROA ROE assets equity efficiency efficiency employment Dummy for date of 0.3092 0.0538 0.4506 0.0922 11.4328 1.4313 111.1536 900.2763 privatization (fpriv) (0.1237)** (0.0182)*** (0.2096)** (0.0323)*** (4.1852)*** (112.163) (36.1325)*** (627.2483) Dummy if firm is 0.1219 0.0114 0.0778 0.0334 7.1600 96.5512 40.8681 27.2020 privatized (epriv) (0.0508)** (0.0080) (0.0696) (0.0227) (4.0179) (82.2342) (20.4276)** (461.7338) Time trend (t) 0.0208 0.0023 0.0132 0.0002 0.0470 6.4866 2.9459 121.4275 (0.0085)** (0.0009)*** (0.0070) (0.0010) (0.0648) (2.8684)** (1.1379)*** (19.2942)*** t * fpriv 0.0247 0.0041 0.0260 0.0108 1.4272 12.5945 8.1146 175.8407 (0.0102)** (0.0017)*** (0.0185) (0.0036)*** (0.5373)*** (13.5645) (3.7083)** (77.6577)** t * epriv 0.0116 0.0014 0.0096 0.0054 1.0251 4.9709 4.2447 51.8588 (0.0059)** (0.0011) (0.0080) (0.0029) (0.5275) (11.7789) (2.9500) (68.2979) Participation in total 0.1551 0.0567 1.1762 0.0521 12.3772 1748.686 274.5105 14982.4047 credit allocations (0.6042) (0.0601) (0.7128) (0.0696) (6.2735)** (202.047)*** (61.9442)*** (1,549.272)*** (share) Percent change of 0.0006 0.0005 0.0035 0.0019 0.0754 12.4515 1.8681 56.0758 GDP per capita (0.0017) (0.0003) (0.0021) (0.0006)*** (0.0423) (1.8376)*** (0.8185)** (9.3183)*** Dummy for closed 0.0548 0.0067 0.0125 0.0127 1.4007 76.3966 24.4158 67.1054 state-owned banks (0.0301) (0.0048) (0.0818) (0.0106) (1.0590) (27.8619)*** (10.8307)** (157.1675) Log(employment) 0.0283 0.0026 0.0549 0.0126 0.3098 115.6341 19.1690 942.4289 (0.0451) (0.0045) (0.0479) (0.0055)** (0.3564) (14.6869)*** (6.4199)*** (97.4552)*** Loans per worker 0.0001 0.0000 0.0001 (0.0001)** (0.0000) (0.0000) Constant 0.0258 0.0347 0.4267 0.8055 7.0200 878.7896 154.4391 5849.7005 (0.2340) (0.0252) (0.2199) (0.0325)*** (1.8918)*** (95.6279)*** (46.8079)*** (594.3191)*** Observations 285 285 285 285 285 285 285 285 Pseudo-R2 0.111 0.115 0.06 0.162 0.144 0.411 0.146 0.4755 * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: ROA return on assets, ROE return on equity, and ROS return on sales. This table reports difference-in-difference panel data GMM-IV (generalized method of moments instrumental variables estimator) results of performance indicators for privatized banks. Private banks are used as a control group. There are eight different regressions with distinct dependent performance variables (see appendix table 8B.1 for variable definitions). Standard errors are in parentheses. Log(employment) and loans per worker have been instrumented using one period lag, fpriv and epriv by GMM-IV. Pseudo-R2 is the R2 using IV regression with robust standard errors. Hansen J-statistic (1982) has been used to test for overidentifying restrictions. In every case the null hypoth- esis that the additional moment conditions are approximately satisfied was rejected, validating the use of the instruments. Source: Author's calculations. 443 444 TORERO size of the coefficient is less than a tenth of the coefficient of the date of the privatization dummy, implying that the overall effect of privatization on performance is positive. Employment The impact of privatization on the welfare of displaced public workers has received little attention in the literature. One of the main problems faced by researchers is the lack of available data, since information for displaced workers is provided for only one period of time (the moment of displacement from the SOE). In fact, it would require large amounts of time, effort, and resources to trace displaced public workers for the pur- pose of analyzing the long-term impact of privatization on their welfare and earnings. This is possibly the reason why most of the studies analyze the effects of privatization on employment, welfare, and wage levels us- ing gross national figures or industry and firm information. The employment indicators make clear that direct employment was reduced significantly after privatization (table 8.10). On average, employ- ment in the former SOEs fell almost 35 percent after privatization, al- though when the figures are compared with industry averages, the average decrease in employment is 0.7 percent higher than the average for the in- dustry.22 When decomposing the employment data by white- and blue- collar employees, the privatized firms show a smaller overall reduction than the industry average of 4.1 percent, but the reduction in employment of blue-collar workers is 17.6 percent higher for the privatized firms than Table 8.10 Changes in Employment after Privatization Before privatization After privatization Change Variable 1993 1997 (percent) Total Mean 2461.17 1605.75 34.76 Median 2237.50 1393.50 37.72 (1508.536) (1300.427) White-collar Mean 2049.17 1387.00 32.31 Median 1746.50 1102.00 36.90 (1457.232) (1153.019) Blue-collar Mean 412.00 218.75 46.91 Median 210.00 149.50 28.81 (495.463) (247.820) Note: See text for explanation. Standard deviations are in parentheses. Source: Author's calculations. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 445 for the industry average, which means that the main reduction of direct employment took place among blue-collar workers. One of the chief criticisms of the privatization process is that the major reason for improvement in performance is a significant reduction in the number of employees, rather than a real increase in total factor produc- tivity. To address this issue, the author follows La Porta and López-de- Silanes (1999) to calculate the impact on privatized companies if all the layoffs were instead kept on at their original salaries.23 For this purpose, the cost of layoffs was calculated as: (Lpre L1994)*Wpre, where Wpre is the average wage in the year preceding privatization, Lpre is the average number of employees in the years preceding privatization, and L1994 is the level of employment in 1994 after privatization. The results are shown in table 8.11. There is no significant difference between the postprivatization performance indicators and the postprivati- zation performance indicators under the assumption that there were no layoffs. For utilities, the percentage change in the profit indicators goes from 2 percent to 5 percent, while in the case of the banks the impact is bigger, averaging 12 percent and 26 percent for Banco Continental and Interbanc, respectively. After including all laid-off workers at their original salaries, net income efficiency (a variable that is directly affected by the number of employees) fell substantially; the percentage change for TdP was 12 percent, for Interbanc 34 percent, and for Continental 14 percent. In the case of TdP, contracting service companies created a significant amount of indirect employment. These service companies frequently consisted of personnel laid off as a result of the privatization. This re- quired an additional exercise that involved subtracting the costs from all service payments carried out by the company to find the net employment layoff. This resulted in a positive percentage change for some of the profitability indicators since the wage costs before privatization were smaller than the fees the company pays the service companies. Further- more, the number of employees in the telecommunications sector rose from 13,000 employees in 1993 to 34,000 employees in 1998, according to OSIPTEL. In general, for all the companies studied, the results of the modified ver- sion of La Porta and López-de-Silanes' (1999) exercise can be explained through the following reasons: total wages of laid-off employees repre- sented only 1.4 percent of total sales, since the average wages paid before privatization were extremely low; postprivatization sales increased signif- icantly, which spread labor costs over a wider base; and there was also a significant increase in the productivity of other factors, especially capital, because of the increase in coverage and the new investments undertaken by privatized firms.24 In the case of utility companies, there was a clear and significant increase in sales. In telecommunications, the number of fixed-line phones per 100 446 Table 8.11 Impact of Layoffs on Performance Indicators for Major Privatized Firms Postprivatization Postprivatization without layoffs Net income Net income Firm ROS ROA ROE efficiency ROS ROA ROE efficiency Telefonica 0.385 0.162 0.287 81.385 0.366 0.155 0.272 71.883 (0.09) (0.05) (0.08) (42.84) (0.09) (0.05) (0.07) (40.57) Electrolima 0.170 0.056 0.072 136.742 0.165 0.054 0.069 132.152 (0.08) (0.03) (0.03) (72.01) (0.08) (0.03) (0.03) (69.05) Electroperú 0.257 0.030 0.050 258.850 0.252 0.029 0.049 252.863 (0.22) (0.02) (0.04) (206.78) (0.22) (0.02) (0.04) (206.65) SEDAPAL 0.160 0.027 0.034 33.496 0.154 0.026 0.032 25.160 (0.04) (0.01) (0.01) (7.70) (0.04) (0.01) (0.01) (11.61) Continental 0.091 0.011 0.130 23.434 0.080 0.010 0.115 20.051 (0.05) (0.01) (0.09) (13.67) (0.06) (0.01) (0.09) (13.70) Interbank 0.066 0.010 0.116 17.323 0.048 0.007 0.087 11.457 (0.06) (0.01) (0.10) (14.27) (0.06) (0.01) (0.11) (16.37) Note: ROA return on assets, ROE return on equity, and ROS return on sales. See text for explanation. See appendix table 8B.1 for variable definitions. Source: Author's calculations. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 447 inhabitants grew from 2.9 in 1993 to 7.8 in 1998. Cellular phones grew from 50,000 to 735,000 in the same two years. In the electricity sector, the coefficient of electrification grew an average of 27 percent, and the genera- tion of electricity grew an average of 25 percent as a result of the heavy vol- ume of investments (approximately $682 million). Even more important, prices for telephone and electricity services were brought into line with the costs of producing those services; before privatization, prices typically cov- ered 75 percent or less of the costs (see Torero and Pascó-Font 2001). In summary, the results show a clear improvement in firm performance following privatization; a relative improvement compared with industry control groups; and an improvement in both labor productivity and total factor productivity. Conclusions Privatization began in Peru in 1991 to generate vital fiscal revenues for the government and improve the quality and coverage of infrastructure and other services. Privatization took place in the telecommunications, elec- tricity, mining, financial services, and hydrocarbons sectors. The process was accompanied by sector reforms aimed at establishing competitive markets and autonomous regulatory agencies. The result was one of Latin America's greatest privatization success stories. A strong record of eco- nomic policy and performance underpinned the success of the privatiza- tion process, however. Macroeconomic stability, an open policymaking environment, and competitive sector markets gave the firms a stable and certain environment through 1998. Without such conditions, success would not have been achieved. Unfortunately, the depth of the reforms, especially the extent of priva- tization, was uneven across sectors. Despite this reform mixture, the re- sults in terms of improvement on the supply side are positive and very sig- nificant. The analysis clearly shows a significant improvement in firm performance since privatization. It is clear from the analysis that privatization had an impact on priva- tized public utility firms in comparison to firms that went through a simi- lar preprivatization restructuring but then were not privatized. Although over time there was a decrease in the performance of the privatized utili- ties, implying lower profits, this could have resulted from increased com- petition in the sector and a slowdown in certain services such as electric- ity due to the incomplete privatization process. Results for the financial sector are similar. The main difference is that privatized banks performed better than they had before they were priva- tized, but not as well as the private banks in the control group. The results also show that privatized banks have a convergence tendency toward the best performers in the private sector. 448 TORERO It is clear that in the short run, the impact of privatization on employ- ment is negative since SOEs usually hired people based on political rather than technical criteria. The privatized firms had to adjust to the new mar- ket conditions and reduce the level of employment by 35 percent, on aver- age. Two effects have been demonstrated: a significant increase in indirect employment through services, and an average growth of 28 percent in to- tal employment--both direct and indirect--since privatization. Neverthe- less, to measure the real impact on employment in the medium run, it is inadequate to study only employment in a specific sector. One also has to study the effects on other sectors stemming from the higher demand for services by the privatized firms. Despite the success in terms of firm performance, service quality and consumer benefits must be taken into account to make a balanced judg- ment of the privatization process. As mentioned in Torero and Pascó-Font (2001), there exist important problems in the privatization process that could explain why positive welfare impacts on consumers were not very significant, or were even negative in electricity. Although, the electricity sector has shown important improvements, the positive effects of privati- zation had not yet reached important regions of urban Peru. This could ex- plain why, on average, consumers are not experiencing an increase in wel- fare. In contrast, telephony is the sector that has experienced the most significant improvements since privatization. In terms of both supply and demand, the results show a positive balance, including an increase in ab- solute levels of consumer surplus and in progressiveness, for the telecom- munications sector. However, since 1997 the positive trend of gains in consumer surplus was reduced. In summary, improved firm performance clearly suggests the necessity of continuing the privatization process, especially in electricity, water, and other SOEs where major reforms need to be concluded--or in some cases begun. This is only a supply-side analysis, but when combined with results from the demand-side analysis (see Torero, Schroth, and Pascó-Font 2004), it is apparent that firms and regulatory agencies must develop ad- equate policies to facilitate the transfer of performance benefits from pri- vatized firms to consumers. In so doing, even further increases in the gains in welfare derived from the process will be possible. Policymakers must fortify the regulatory agencies and increase their independence. They also must work with privatized firms to identify vulnerable groups and to de- velop tailored measures, such as appropriate consumption plans, that will help increase consumer welfare. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 449 Appendix 8A Data Sources Banco Central de Reserva del Perú (BCRP). 1986­2000. Annual Report. Lima. ------. Web page: http://bcrp.gob.pe. Cementos Lima S.A. 1988­98. Annual Report. Lima. Comisión de Promoción de la Inversión Privada (COPRI). 1996. White Book. Cementos Lima S.A. Lima. ------. White Book. CEPREL ­ Electrolima S.A. Lima. ------. White Book. Electroperú, S.A. Lima. ------. White Book. Electro Sur Medio, S.A. Lima. ------. White Book. Empresas Regionales de Electricidad: Electro Norte, S.A., Electro Norte Medio, S.A., Electro Noroeste, S.A., and Electro Centro, S.A. Lima. ------. Web page: http://www.copri.org. Comisión de Tarifas Eléctricas (CTE). 1984, 1985, 1986­89, 1990­91, 1996, 1999. Annual Report. Lima. ------. 1992­93, 1994, 1995, 1996, 1997, 1998, 1999. Statistical Year- book. Lima. Comisión Nacional Supervisora de Empresas y Valores (CONASEV). Web page: http://www.conasevnet.gob.pe. Compañía Peruana de Teléfonos S.A. 1985­87, 1990­93. Annual Report. Lima. Electrolima, S.A. 1985­88, 1990­94, 1997. Annual Report. Lima. Electroperú, S.A. 1985­99. Annual Report. Lima. Entel Perú, S.A. 1990, 1991, 1993. Annual Report. Lima. ------. 1985­92. Statistical Yearbook. Lima. SEDAPAL, S.A. 1984­99. Annual Report. Lima. ------. 1997. "Historia del abastecimiento del agua potable de Lima 1535­1996." Lima. Superintendencia de Banca y Seguros (SBS). 1986­2000. Weekly Notes. Lima. Superintendencia Nacional de Servicios de Saneamiento (SUNASS). 1996­99. Annual Report. Lima. ------. 1998. "Indicadores de Gestión 1996­1998." Lima. 450 TORERO ------. Centro de Documentación, Web page: http://www.sunass.gob.pe/ cendoc.html. Telefónica del Perú, S.A. 1994­2000. Annual Report. Lima. Other Data Sources From the Commission of Energy Tariffs: Commercial Information (First quarter, 2000). Publication containing results from the processing and analysis of com- mercial information provided by the electric sector companies. CTE Informs (June 1999­November 2000). Monthly publication with news about regulation, markets, agents, and other current topics of interest in the electricity and hydrocarbon sec- tors in Peru and all around the world. Electric Sector Operations (January 2000­November 2000). Monthly publication containing information on production and demand of electricity, prices, and other information related to the operation of the sector. El Informativo (June 1996­November 2000). Periodical publication containing technical articles, market information, evolution of rates, company news, statistics, and sector news. The Statistical Yearbook (1994­98). A yearly publication that details regulation of electricity rates and eco- nomic results for the Peruvian electricity market. Yearly Report (1994­1999). From OSIPTEL (Supervisory Agency for Private Investment in Telecom- munications): The Transformation of Telecommunications in Peru. 1995 Report. Regulation and the Telecommunications Market. 1996 Report. The Opening of the Telecommunications Market. 1997 Report. Consumers and Telecommunications. 1998 Report. Five Years in the Telecommunications Market. 1999 Report. Institutional Report. 2000. Technical studies (various titles). Studies in telecommunications (various titles). Bulletins (various titles). PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 451 Appendix 8B Table 8B.1 Description of Variables Variable Description Performance variable ROS Return on sales is the ratio of net income to sales. Net income is equal to total income minus operating and administrative expenses, plus financial income minus financial expenses, and net taxes paid. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. ROA Return on assets is the ratio of net income to assets. Net income is equal to total income minus operating and administrative expenses, plus financial income minus financial expenses, and net taxes paid. Assets are the total value of the entire property of the firm on December 31. ROE Return on equity is the ratio of net income to equity. Net income is equal to total income minus operating and administrative expenses, plus financial income minus financial expenses, and net taxes paid. Equity is the value of the participation that the partners or owners have in the company. Debt/assets Debt/assets is the ratio of liability to assets. Liability is the value of the debt owed by the company. Assets are the total value of the entire property of the firm on December 31. Debt/equity Debt/equity is the ratio of liability to equity. Liability is the value of the debt owed by the company. Equity is the value of the participation that the partners or owners have in the company. Sales efficiency Sales efficiency is the ratio of sales to employment. Sales are equal to the total value of products and services sold, nationally and internationally, minus sales returns and discounts. Employment is measured as the total number of employees in the firm, including white- and blue-collar workers. (Table continues on the following page.) 452 TORERO Table 8B.1 (continued) Variable Description Net income efficiency Net income efficiency is the ratio of net income to employment. Net income is equal to total income minus operating and administrative expenses, plus financial income minus financial expenses, and net taxes paid. Employment is measured as the total number of employees in the firm, including white- and blue-collar workers. Total employment Total employment is the total number of employees. The employees include white-collar workers and blue-collar workers at full time. Assets/employment The ratio of assets to employment indicates the proportion of assets that on average, would correspond to each employee. Assets are the total value of the entire property of the firm on December 31. Employment is measured as the total number of employees in the firm, incuding white- and blue-collar workers. Sector variable The share rate is the ratio of participation of the Share rate firm in its economic sector. In the case of utilities, participation is constructed by the contribution of the firm to total invoicing of the sector. In other words, it is the ratio of the firm's invoicing to the total invoicing of the sector. In the case of banks, the rate is constructed by the participation of the bank in total credit allocations. The concentration index of the sector is the Concentration index by CI4. This index is the sum of the share rate sector of the four firms with the highest participation in the sector. The share rate is the ratio of participation of the firm in its economic sector. In the case of utilities, participation is constructed by the contribution of the firm to total invoicing of the sector. In the case of banks, the rate is constructed by the participation of the bank in total credit allocations. The highest value the CI4 can take is 1, which indicates the existence of a monopoly. Utilities A dummy that takes the value 1 if the firm belongs to the utilities sector, and 0 otherwise. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 453 Table 8B.1 (continued) Variable Description Privatization variable Privatization A dummy that takes the value 1 for all years the firm has been in private hands, and 0 otherwise. Concession A dummy that takes the value 1 for all years the firm has been in private hands, and 0 otherwise. Value of transactions The value of transactions is the amount, in current US$ millions, paid by the investor for the available share package or management rights. Projected investment The projected investment is the amount, in current US$ millions, that the investor commits to invest. Operator's characteristic Foreign participation A dummy that takes the value 1 if foreign investors hold a greater percentage of stocks than any other investor, and 0 otherwise. Buyer's experience A dummy that takes the value 1 if the buyer owns and manages one or more companies that belong to the same sector of the privatized firm, and 0 otherwise. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. Regulatory agency variable Regulated industry A dummy that takes the value 1 if the firm belongs to a regulated industry, and 0 otherwise. A regulated industry is characterized by the presence of a regulatory agency. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. Regulatory agency A dummy that takes the value 1 if a regulatory agency operated in the firm's sector in that year, and 0 otherwise. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. Price cap regulation A dummy that takes the value 1 if there was a price cap regulation in a specific year, and 0 otherwise. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. (Table continues on the following page.) 454 TORERO Table 8B.1 (continued) Variable Description Rate of return A dummy that takes the value 1 if the rate of regulation return is regulated, and 0 otherwise. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. Discretionary prices A dummy that takes the value 1 if the regulatory agency adopts a discretionary price regulation in that year, and 0 otherwise. Industries or sectors considered are telecommunications, electricity, water and sanitation, and banking. Macro variable Peruvian per capita real Per capita real GNP is the gross national GNP product by inhabitant at 1994 prices. Average precipitation Average precipitation is the average annual rain precipitation, measured in cubic millimeters. Average exchange rate The annual average exchange rate, measured as the value of a U.S. dollar in Nuevos Soles. IPG The IPG (generalized price index) is a modification of the consumer price index (CPI). It has the advantage of solving problems that are found in traditional estimations of changes in prices, and that become particularly severe in hyperinflation periods. It is a measure of how much the average price has changed since the base year (1994). GNP The gross national product is the value of goods and services produced by a country's residents over a year at 1994 prices. GNP is the sum of the value of consumption, investment, government expenses, and exports, minus imports. Population The total number of inhabitants in Peru. Agriculture GNP The value of the goods and services produced in the agricultural sector in a specific year at 1994 prices. Fishing GNP The value of the goods and services produced in the fishing sector in a specific year at 1994 prices. Mining and The value of the goods and services produced hydrocarbons GNP in the mining and hydrocarbons sector in a specific year at 1994 prices. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 455 Table 8B.1 (continued) Variable Description Manufacturing GNP The value of the goods and services produced in the manufacturing sector in a specific year at 1994 prices. Construction GNP The value of the goods and services produced in the construction sector in a specific year at 1994 prices. Total domestic savings The change in the value of the economy's assets (percent of GNP) as a whole, calculated as a percentage of GNP. Public savings (percent The change in the value of the Peruvian public of GNP) sector's assets calculated as a percentage of the GNP. Private savings (percent The change in the value of the Peruvian private of GNP) sector's assets, calculated as a percentage of the GNP. Total investment The value of total purchases of capital goods (percent of GNP) as a percentage of GNP. Public investment The value of public purchases of capital goods (percent of GNP) as a percentage of GNP. Private investment The value of private purchases of capital goods (percent of GNP) as a percentage of GNP. Current account balance The value of goods produced by domestic (percent of GNP) residents (including the net factor income from abroad) plus net transfers from abroad, minus the expenditure by domestic residents on goods as a percentage of the GNP. When the current account balance is positive (negative), there is a surplus (deficit) in the current account. Trade balance (percent The value of exports of goods minus the value of GNP) of imports of goods, expressed as a percentage of GNP. Services balance The value of exports of services minus the (percent of GNP) value of imports of services, expressed as a percentage of GNP. Net factor income The net income from labor or capital factors, (percent of GNP) which includes claims on assets abroad and debt interest payments. It is calculated as a percentage of GNP. Current transfers The value of received foreign assets minus (percent of GNP) assets transferred outside the country. Any extraordinary contributions, such as donations, are included. It is calculated as a percentage of GNP. (Table continues on the following page.) 456 TORERO Table 8B.1 (continued) Variable Description Total exports Total exports, measured in $US millions at 1994 prices. Total imports Total imports measured in $US millions at 1994 prices. Private and public total The value of debt contracted with foreign external debt agents by public and private organizations, measured in US$ millions at 1994 prices. Public total external The value of debt contracted with foreign debt agents by the government, measured in US$ millions at 1994 prices. RIN (net international The value, in US$ millions at 1994 prices, of reserves) liquid assets, including international currency, which the Central Bank uses for international transactions. Inflation (percent) The percentage change in the general level of prices, as measured by the IPG (defined above). Export (percent of The value of exports as a percentage of GNP. GNP) Exports are goods that are produced by the residents of a country and sold to foreigners. Import (percent of The value of imports as a percentage of GNP. GNP) Imports are goods that are produced by foreign- ers and sold to the residents of a country. Export import The value of exports plus imports as a (percent of GNP) percentage of GNP. Exports are goods that are produced by the residents of a country and sold to foreigners. Imports are goods that are produced by foreigners and sold to the residents of a country. Terms-of-trade index The price of Peru's tradable goods expressed relative to the price of a market basket of the world's tradable goods. It is approximated by the ratio of Peru's export prices to import prices. Potable water The national production of potable water in a year, measured in cubical meters. Electricity The national production of electricity in a year, measured in kilowatts per hour. Telephony The number of telephone calls in a year. Vital minimum wage Legal minimum wage that a firm can pay. Index of total The number of persons working at jobs in the employment (Jan. market sector divided by the number of workers 1995 100) in January 1995 and multiplied by 100. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 457 Table 8B.1 (continued) Variable Description Strikes Number of strikes in a year for all industries of the economy. The Ministry of Labor registers the strikes, the number of affected workers, and the total loss of work hours as a result of strikes. Affected workers Total number of workers affected by strikes in a year for all industries of the economy. Man-hours lost The sum of all lost work hours per worker due to strikes. Subversive activity The number of subversive attacks in a year within the Peruvian territory. Financial variables Personal expenses per The ratio of total personal expenses to the employee total number of employees. Personal expenses are equal to all resources dedicated to labor issues and services in a firm. Total number of employees is the total employment. The employees include white-collar workers and blue-collar workers at full time. Bad loan portfolio The ratio of bad loans to the net loan portfolio. Bad loans are defined as expired loans plus legal costs. Net loan portfolio is equal to current accounts less discounts, plus long- and short-term loans, refinanced loans, mortgages, and other loans. Administrative expenses The ratio of total administrative expenses to financial income. Administrative expenses are equal to personal expenses, general expenses, depreciation, and amortization. Financial income is equal to income obtained by commissions and interests on loans. Financial margin per The ratio of the financial margin to the branch number of branches. Financial margin is equal to financial income (income obtained by commissions and interests on loans) less financial expenses. Branch is defined as the total number of offices. Source: Author's calculations. 458 Appendix 8C Table 8C.1 Evolution of Privatization in the Electricity Sector 1986­93 1994 1995 1996 1997 1998 1999 Electroperú Electroperú Electroperú Electroperú Electroperú Egenor Egenor Egenor Egenor Egenor Electroperú Electroperúb Egesur Egesur Egesur Egesur Egesur Cahua Cahua Cahua Cahua Cahua Luz del Sur Luz del Sur Luz del Sur Luz del Sur Luz del Sur Luz del Sur Electrolimaa Edegel Edegel Edegel Edegel Edegel Edegel Edelnor Edelnord Edelnor Edelnor Edelnor Edelnor Electrolimac EDE-Chancayd EDE-Cañete EDE-Cañete EDE-Cañete EDE-Cañete EDE-Cañete Elec. Centro Elec. Centro Elec. Centro Elec. Centro Elec. Centro Elec. Centro Elec. Nor Oeste Elec. Nor Oeste Elec. Nor Oeste Elec. Nor Oeste Elec. Nor Oeste Elec. Nor Oeste Elect. Norte Elect. Norte Elect. Norte Elect. Norte Elect. Norte Elect. Norte Elec. Nor. Medio Elec. Nor. Medio Elec. Nor. Medio Elec. Nor. Medio Elec. Nor. Medio Elec. Nor. Medio Elec. Oriente Elec. Oriente Elec. Oriente Elec. Oriente Elec. Oriente Elec. Oriente Elec. Sur Elec. Sur Elec. Sur Elec. Sur Elec. Sur Elec. Sur Elec. Sur Este Elec. Sur Este Elec. Sur Este Elec. Sur Este Elec. Sur Este Elec. Sur Este Elec. Sur Medio Elec. Sur Medio Elec. Sur Medio Elec. Sur Medio Elec. Sur Medio Elec. Sur Medio Seal Seal Seal Seal Seal Seal Seal Emsemsa Emsemsa Emsemsa Emsemsa Emsemsa Etevensa Etevensa Etevensa Etevensa Etevensa Egasa Egasa Egasa Egasa Egasa Gera Gera Gera Gera Gera Egemsa Egemsa Egemsa Egemsa Egemsa Etecen Etecen Etecen Etecen Etecen Etesur Etesur Etesur Etesur Etesur Electro Ucayali Electro Ucayali Electro Ucayali Electro Ucayali Electro Ucayali Coelvisa Coelvisa Coelvisa Coelvisa Sers Sers Sers Sers C.H. Virú Electro Andes Electro Andes Electro Andes Eepsa Eepsa Eepsa Chavimochic Chavimochic Chavimochic Shougesa Shougesa Pariac Pariac Electro Pangoa Electro Pangoa Emseusa Electro Tocache Electro Puno San Gabán a. Electrolima was divided into four different companies: Edelnor (distribution), Luz del Sur (distribution), Edegel (generation), and Electrolima (residual company). b. Electroperú split into four different companies: Electroperú, Egenor, Egesur, and Cahua. c. Electrolima (residual company) was again divided into two companies: EDE-Chancay and EDE-Cañete. d. The merger between Edelnor and EDE-Chancay resulted in Edelnor. Source: Author's tabulations. 459 460 Table 8C.2 Evolution of Privatization in the Financial Sector Birth State Financial reform: Name year participation 1992­98 1986 1987 1988 1989 1990 1991 1992 Amazónico 1962 Yes Liquidación Amazónico Amazónico Amazónico Amazónico Amazónico Amazónico America 1966 America America America Bandesco 1980 Bandesco Bandesco Bandesco Bandesco Bandesco Bandesco Bandesco Central de 1984 Central de Central de Madrid Madrid Madrid CCC 1988 CCC CCC CCC Citibank 1920 Citibank Citibank Citibank Citibank Citibank Citibank Citibank Comercio 1967 Comercio Comercio Comercio Comercio Comercio Comercio Comercio Continental 1951 Yes Privatización Continental Continental Continental Continental Continental Continental Continental Continorte 1961 Yes Liquidación Continorte Continorte Continorte Continorte Continorte Continorte Crédito 1889 Crédito Crédito Crédito Crédito Crédito Crédito Crédito De los Andes 1962 Yes Liquidación De los andes De los andes De los andes De los andes De los andes De los Andes Del Norte 1960 Del Norte Del Norte Del Norte Del Norte Del Norte Del Norte Del Norte Extebandes 1982 Extebandes Extebandes Extebandes Extebandes Extebandes Extebandes Extebandes Financiero 1986 Financiero Financiero Financiero Financiero Financiero Financiero Interamericano 1991 Interamericano Interamericano Interandino 1990 Interandino Interandino Interandino Interandino Interbank 1897 Yes Privatización Interbank Interbank Interbank Interbank Interbank Interbank Interbank Latino 1982 Latino Latino Latino Latino Latino Latino Latino Lima 1952 Lima Lima Lima Lima Lima Lima Lima Londres 1936 Londres Londres Manhattan 1984 Manhattan Mercantil 1984 Mercantil Mercantil Mercantil Mercantil Mercantil Mercantil Mercantil Popular 1889 Yes Liquidación Popular Popular Popular Popular Popular Popular Probank 1990 Probank Probank Probank Sur Perú 1962 Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Surmebanc 1962 Yes Liquidación Surmebanc Surmebanc Surmebanc Surmebanc Surmebanc Surmebanc Tokyo 1965 Tokyo Tokyo Wiese 1943 Wiese Wiese Wiese Wiese Wiese Wiese Wiese Sudamericano 1993 Banex 1993 Nuevo Mundo 1993 Del Libertador 1994 Del Trabajo 1994 Solventa 1994 Serbanco 1996 Boston 1996 Republica 1980 Orion 1995 Del pais 1997 Mibanco 1998 BNP-andes 1999 Name 1993 1994 1995 1996 1997 1998 1999 2000 Amazónico America Bandesco Bandesco Bandesco Central de Madrid CCC Citibank Citibank Citibank Citibank Citibank Citibank Citibank Citibank Citibank Comercio Comercio Comercio Comercio Comercio Comercio Comercio Comercio Comercio Continental Continental Continental Continental Continental Continental Continental Continental Continental Continorte Crédito Crédito Crédito Crédito Crédito Crédito Crédito Crédito Crédito De los Andes Del Norte Del Norte Del Norte Del Norte Del Norte Del Norte Del Norte Del Norte NBK-Boston Extebandes Extebandes Extebandes Extebandes Extebandes Extebandes Standard Standard Standard Financiero Financiero Financiero Financiero Financiero Financiero Financiero Financiero Financiero 461 (Table continues on the following page.) 462 Table 8C.2 (continued) Name 1993 1994 1995 1996 1997 1998 1999 2000 Interamericano Interamericano Interamericano Interamericano Interamericano Interamericano Interamericano Interamericano Interamericano Interandino Interandino Interandino Santander Santander Santander Santander Santander Santander Interbank Interbank Interbank Interbank Interbank Interbank Interbank Interbank Interbank Latino Latino Latino Latino Latino Latino Latino Latino Latino Lima Lima Lima Lima Lima Lima Lima ¬N Fusión con Wiese Londres Manhattan Mercantil Mercantil Mercantil Mercantil ¬N Fusión con Santander Popular Probank Probank Probank Probank Probank Probank Probank ¬N Fusión con Del Norte Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Sur Perú Surmebanc Tokyo Wiese Wiese Wiese Wiese Wiese Wiese Wiese Wiese Wiese Sudamericano Sudamericano Sudamericano Sudamericano Sudamericano Sudamericano Sudamericano Sudamericano Sudamericano Banex Banex Banex Banex Banex Banex Banex Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Nuevo Mundo Del Libertador Del Libertador Del Libertador ¬N Fusión con Sur Peru Del Trabajo Del Trabajo Del Trabajo Del Trabajo Del Trabajo Del Trabajo Del Trabajo Solventa Solventa Solventa Solventa Solventa ¬N Fusión con Del Norte Serbanco Serbanco Serbanco Serbanco Serbanco Boston Boston Boston Boston Boston Boston Republica Republica Republica Republica Orion Orion Orion Orion Del pais Del pais Del pais ¬N Fusión con Nuevo Mundo Mibanco Mibanco Mibanco Mibanco BNP-Andes BNP-Andes BNP-Andes Source: Superintendencia de Banca y Seguros, Memorias 1986­1991. Superintendencia de Banca y Seguros, Información Financiera Mensual 1992­2000. Appendix 8D Table 8D.1 Basic Statistics of Privatized Firms Debt Debt Net Number Assets to to Sales income of to Net Firm Number ROS ROA ROE assets equity efficiency efficiency workers employment income Assets Nonfinancial firms 336 0.249 0.039 0.595 0.469 33.372 369.818 43.185 1,337 1,443.96 70,480 1,540,130 (0.72) (0.23) (12.18) (0.55) (519.60) (554.37) (262.36) (2,533.35) (3,508.70) (802,842) (3,612,434) 332 298 297 301 300 317 317 336 296 331 311 Telecommunications Telefonica 15 0.174 0.077 0.136 0.462 1.047 270.121 37.457 10,523 663.55 343,455 5,398,240 (0.22) (0.09) (0.17) (0.15) (0.77) (177.89) (51.47) (4,284.89) (490.97) (427,695) (2,266,784) Energy Electro Centro 14 0.443 0.034 0.053 0.263 0.407 146.745 43.999 618 722.48 31,089 388,853 (0.76) (0.05) (0.08) (0.14) (0.29) (72.39) (60.80) (150.04) (338.59) (42,847) (99,896) 14 10 10 10 10 14 14 14 11 14 11 Electro Nor Oeste 13 0.600 0.141 0.259 0.459 1.140 203.269 94.758 372 695.36 35,737 240,917 (0.95) (0.25) (0.43) (0.18) (1.07) (89.24) (114.65) (87.02) (283.27) (47,839) (103,963) 13 10 10 10 10 13 13 13 11 13 11 Electro Norte 14 0.473 0.105 0.534 0.455 2.030 156.803 54.046 327 405.53 18,137 125,990 (0.60) (0.08) (1.20) (0.22) (4.00) (85.14) (44.42) (46.00) (135.30) (15,795) (33,007) 13 10 10 10 10 13 13 14 11 13 11 Electro Norte Medio 14 0.367 0.025 0.066 0.387 1.010 260.329 69.090 550 1,024.26 41,454 468,014 (0.66) (0.04) (0.10) (0.24) (1.14) (183.68) (116.46) (140.05) (777.21) (71,836) (298,246) 463 14 10 10 10 10 14 14 14 11 14 11 (Table continues on the following page.) 464 Table 8D.1 (continued) Debt Debt Net Number Assets to to Sales income of to Net Firm Number ROS ROA ROE assets equity efficiency efficiency workers employment income Assets Electro Oriente 14 0.437 0.046 0.074 0.329 0.555 237.351 84.553 199 1,285.01 17,795 240,264 (0.61) (0.04) (0.07) (0.14) (0.39) (86.83) (95.37) (32.33) (475.32) (20,692) (82,307) 14 10 10 10 10 14 14 14 11 14 11 Electro Sur 14 0.408 0.124 0.010 0.419 0.333 187.524 50.600 152 519.93 8,429 74,913 (0.61) (0.23) (0.73) (0.41) (1.71) (67.67) (71.51) (23.42) (327.13) (11,603) (50,539) 14 10 10 10 10 14 14 14 11 14 11 Electro Sur Este 14 0.283 0.027 0.054 0.313 0.578 203.681 51.568 313 778.74 16,760 233,936 (0.46) (0.03) (0.08) (0.21) (0.62) (60.11) (70.13) (35.53) (333.17) (23,075) (87,921) 14 10 10 10 10 14 14 14 11 14 11 Electro Sur Medio 14 0.248 0.044 0.100 0.295 0.516 140.550 15.698 368 337.86 7,816 110,249 (0.45) (0.08) (0.18) (0.18) (0.47) (74.78) (26.94) (124.89) (135.52) (13,078) (36,632) 14 10 10 10 10 14 14 14 13 14 13 Electrolima 14 0.031 0.012 0.012 0.345 0.577 415.272 47.692 3,252 1,370.89 51,941 3,533,158 (0.30) (0.05) (0.08) (0.12) (0.29) (323.15) (94.74) (1,275.97) (967.23) (221,382) (1,480,396) Electroperú 14 0.475 0.010 0.017 0.453 0.905 540.289 69.958 1,503 8,463.11 26,213 12,500,000 (1.52) (0.03) (0.05) (0.10) (0.47) (497.82) (240.49) (806.17) (3,448.23) (320,191) (9,487,072) Sanitation SEDAPAL 14 0.279 0.002 0.001 0.192 0.241 133.730 0.536 2,514 1,059.30 47,248 2,500,485 (0.70) (0.03) (0.04) (0.05) (0.08) (78.19) (40.12) (954.74) (408.21) (183,016) (1,143,402) Mining Centromin 5 0.133 0.069 0.541 0.877 8.169 1,909.095 206.377 714 3,935.96 147,339 2,810,561 (0.17) (0.10) (0.70) (0.04) (3.71) (578.95) (257.10) (0.00) (2,321.44) (183,552) (1,657,888) Cerro Verde 6 0.191 0.095 0.157 0.445 1.025 257.143 25.517 556 545.46 15,259 293,492 (0.73) (0.36) (0.68) (0.22) (0.75) (127.13) (148.92) (114.88) (347.76) (80,744) (165,074) 5 5 5 6 6 5 5 6 6 5 6 Condestable 14 0.097 0.020 20.031 0.797 676.505 775.598 97.327 28 902.60 150 24,341 (0.45) (0.21) (52.62) (0.23) (2,395.99) (326.46) (160.30) (0.00) (694.27) (5,534) (13,553) 14 14 14 14 14 7 7 7 7 13 14 Hierro Peru 8 0.293 0.082 0.457 0.628 2.743 98.712 0.329 2003 177.58 26,824 493,081 (0.60) (0.14) (0.85) (0.20) (2.72) (8.29) (6.73) (145.42) (28.59) (67,554) (287,769) 8 8 8 8 8 3 3 3 3 8 8 Minero Peru 5 1.213 0.215 0.976 0.810 4.474 716.930 634.945 984 3,152.68 624,833 3,102,620 (1.81) (0.26) (1.10) (0.04) (1.24) (479.05) (717.66) (0.00) (2,747.58) (706,262) (2,704,137) Tintaya 4 0.190 0.050 0.805 0.232 13.902 3,292.367 541.116 88 21,114.60 47,634 1,858,315 (0.37) (0.08) (1.61) (1.89) (26.50) (786.45) (1,263.80) (0.00) (17,047.68) (111,257) (1,500,480) Industry Cemento Sur 13 0.135 0.059 0.076 0.365 0.407 198.682 24.380 141 270.92 2,873 35,505 (0.20) (0.16) (0.26) (0.28) (0.21) (86.87) (36.09) (15.04) (138.58) (4,010) (14,562) 8 8 7 8 7 8 8 13 8 8 8 Cemento Yura 10 1.152 0.087 2.199 0.608 6.184 299.780 274.507 199 1,436.22 56,090 284,337 (1.65) (0.23) (5.64) (0.22) (12.30) (166.06) (595.65) (10.12) (1,081.74) (119,813) (220,114) 7 7 7 7 7 7 7 10 7 7 7 Cementos Lima 14 0.144 0.080 0.115 0.241 0.353 785.835 138.690 340 1,886.88 43,628 657,215 (0.14) (0.07) (0.10) (0.13) (0.24) (268.28) (111.02) (56.83) (1,052.64) (35,490) (451,104) 14 14 14 14 14 14 14 14 14 14 CNP S.A. 15 0.234 0.100 0.823 0.747 0.421 49.333 6.827 399 57.97 2,157 21,189 (0.51) (0.21) (2.47) (0.26) (6.92) (17.35) (12.94) (48.77) (20.28) (4,419) (5,758) 465 9 9 9 9 9 9 9 15 9 9 9 (Table continues on the following page.) 466 Table 8D.1 (continued) Debt Debt Net Number Assets to to Sales income of to Net Firm Number ROS ROA ROE assets equity efficiency efficiency workers employment income Assets Cerper 3 0.065 0.105 0.176 0.412 0.707 0.000 0.000 0 0.00 1,213 11,460 (0.10) (0.13) (0.22) (0.04) (0.12) (0.00) (0.00) (0.00) (0.00) (1,469) (217) Sider Peru S.A. 12 0.149 0.051 0.105 0.505 1.414 138.729 14.741 3,271 262.68 73,433 848,168 (0.55) (0.21) (0.76) (0.19) (1.17) (64.26) (47.50) (1,052.99) (166.00) (153,536) (724,436) 10 10 10 10 10 10 10 12 10 10 10 Empresa de la Sal 10 0.051 0.023 0.041 0.358 0.575 113.040 1.783 187 162.10 458 19,505 (0.16) (0.09) (0.15) (0.07) (0.18) (60.39) (11.86) (147.11) (90.17) (1,858) (1,952) 10 10 10 10 10 10 10 10 10 10 Industrias Navales 6 0.018 0.069 0.062 0.333 0.675 90.090 6.622 34 140.94 279 4,483 (0.37) (0.12) (0.28) (0.20) (0.73) (40.56) (22.17) (13.29) (48.94) (657) (1,076) La Pampilla 1 0.024 0.025 0.039 0.362 0.566 1,486.441 35.453 341 1,442.04 12,089 491,734 . . . . . . . . . . . Petroperu S.A. 13 0.540 0.403 1.394 1.422 2.455 1,055.132 274.642 4,772 793.46 1,592,135 3,835,390 (1.10) (0.81) (2.28) (2.04) (11.70) (1,216.42) (657.35) (1,733.10) (759.62) (3,759,827) (4,588,286) 13 13 13 13 13 13 13 13 13 13 Quimica del Pacifico 6 0.129 0.081 0.118 0.314 0.497 154.684 18.946 323 210.63 6,379 66,508 (0.11) (0.05) (0.06) (0.11) (0.30) (77.20) (19.41) (95.18) (117.05) (8,553) (49,123) Renasa 13 0.050 0.024 0.042 0.357 0.576 170.530 7.394 82 221.96 1,235 17,262 (0.21) (0.14) (0.21) (0.08) (0.19) (64.52) (33.13) (31.24) (93.80) (4,067) (10,391) SEAL 14 0.224 0.005 0.017 0.377 0.781 263.449 27.685 433 636.37 16,490 245,730 (0.38) (0.06) (0.11) (0.17) (0.77) (167.86) (54.21) (146.53) (351.27) (27,328) (185,602) 14 10 10 11 11 14 14 14 11 14 11 Solgas 5 0.003 0.042 0.060 0.323 1.743 309.336 0.852 230 184.35 196 42,403 (0.07) (0.13) (0.14) (0.35) (3.32) (154.65) (29.10) (0.00) (124.02) (6,694) (28,530) Sufisa 6 0.047 0.075 8.294 0.757 24.466 339.473 6.461 401 451.71 7,384 159,632 (0.18) (0.20) (20.30) (0.32) (46.83) (81.74) (51.93) (124.37) (236.92) (26,100) (47,332) Financial firms 340 0.725 0.080 1.600 0.883 9.462 277.628 269.940 1179 1636.41 1,904,351 1,738,226 (12.85) (1.32) (28.00) (0.07) (4.43) (182.05) (4,621.28) (1,573.68) (1,182.42) (34,800,000) (2,983,462) Continental 15 0.080 0.012 0.167 0.920 13.073 273.887 18.716 2913 1,656.87 52,323 4,491,742 (0.05) (0.01) (0.11) (0.02) (6.10) (109.48) (12.73) (491.96) (923.11) (34,154) (1,929,788) Interbank 15 0.056 0.011 0.121 0.908 10.313 254.653 12.846 2361 1,310.12 28,045 2,408,220 (0.05) (0.01) (0.10) (0.02) (2.22) (141.65) (12.25) (911.15) (974.21) (27,476) (886,993) Note: This table reports raw results for performance indicators (see appendix 8B for variable definitions). The results are simple means for each one of the variables. Standard deviations are shown in parentheses. Column (1) presents the number of available observations per firm or group of firms used to obtain the estimations. However, when there are fewer observations for a specific variable, this number is shown under the standard deviation for such a variable. Source: Author's calculations. 467 468 Table 8E.1 Changes in Performance after Privatization for Telefónica del Perú Means First differences Diff. in diff. Pre- Post- S-Franciac Kolmogorov- Sector privatization privatization t-testa Hotellingb Hotellingb prob z Smirnov Performance measure (Pi) Profitability Return on sales (ROS) 0.0099 0.4083 10.2639*** 105.3480*** 49.6114*** 0.1974 0.001*** (0.029) (0.028) Return on assets (ROA) 0.0024 0.1714 6.9935*** 48.9086*** 24.4539*** 0.3279 0.001*** (0.014) (0.021) Return on equity (ROE) 0.0036 0.3128 7.8995*** 62.4022*** 33.4508*** 0.2083 0.007*** (0.032) (0.014) Operating efficiency Sales efficiency (SALEFF)d 143.9187 455.3162 6.3317*** 40.0909*** 42.5110*** 0.0382 0.007*** (23.373) (47.931) Net income efficiency (NIEFF)d 0.9794 93.7577 8.3231*** 69.2732*** 47.2743*** 0.0299 0.008*** (3.662) (12.355) Employment Total employment (EMPL) 14125.6 5992.17 9.9687*** 99.3749*** 38.4810*** 0.1090 0.001*** (575.074) (543.713) Leverage Debt to assets (LEV) 0.4999 0.4584 0.5444 0.2964 11.9149 0.8725 0.921 (0.055) (0.049) Debt to equity (LEV2) 1.2433 0.9228 0.7603 0.5781 11.6069 0.0026 0.921 (0.339) (0.170) Coverage Lines per worker (LINES) 39.6038 261.0051 8.0536*** 64.8610*** 0.0047 0.001*** (9.763) (78.008) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: First and second differences for changes in performance are shown, for each empirical proxy, using the mean values (see appendix table 8B.1 for variable definitions). Columns 1 and 2 present the mean values before and after privatization (the year of privatization is 1994). Columns 3 and 4 show results for two tests for significance in changes in performance after privatization (first differences): t-test (two-tailed Wilcoxon signed-rank test) and the Hotelling test. Column 5 displays significance of difference-in-difference results using a Hotelling test (control group is the water and sanitation enterprise SEDAPAL). Column 6 presents Shapiro-Francia test for normality. Finally, column 7 shows the results for the Kolmogorov-Smirnov (K-S) statistic, a non- parametric test used to formally test the equality of the empirical hazards functions of the different pre- and postprivatization performance indicators. a. t-test for Ho (null hypothesis) for difference between means. Numbers are unequal. (X1 X2) (X1 X2) t S x1 x2 (n1 1) s1 2 (n2 1) s2 2 1 1 B (n1 n2 2) *Bn1 n2 where x is an l k matrix of the means and S is the estimated covariance matrix. b. Test of equality: T2 (x1 x2) S 1(x1 x2) c. Shapiro-Francia test for normality. Ho (null hypothesis) variable is normally distributed. d. Thousands of nuevos soles. Source: Author's calculations. 469 470 Table 8E.2 Changes in Performance after Privatization for Electrolima Means First differences Diff. in diff. Pre- Post- S-Franciac Kolmogorov- Sector privatization privatization t-testa Hotellingb Hotellingb prob z Smirnov Performance measure (Pi) Profitability Return on sales (ROS) 0.1811 0.2018 1 2.6239** 116.8848** 113.6440* 0.0032 0.017*** (0.320) (0.024) Return on assets (ROA) 0.0205 0.0661 1 4.2075*** 117.7033*** 117.5367** 0.2327 0.017*** (0.016) (0.004) Return on equity (ROE) 0.0335 0.0850 1 3.2998*** 110.8884*** 112.4812 0.0583 0.017*** (0.028) (0.005) Operating efficiency Sales efficiency (SALEFF)d 162.9284 803.5256 12.5352*** 157.1316*** 119.0269*** 0.0155 0.002*** (16.391) (60.559) Net income efficiency (NIEFF)d 19.0959 163.0455 1 9.4166*** 188.6730*** 133.3117*** 0.0424 0.017*** (11.569) (16.074) Employment Total employment (EMPL) 4210.3 1855.60 1 7.2221*** 152.1582*** 150.8770*** 0.1292 0.002*** (239.607) (138.342) Leverage Debt to assets (LEV) 0.4302 0.2208 1 5.1558*** 126.5819*** 115.2595*** 0.6949 0.001*** (0.023) (0.037) Debt to equity (LEV2) 0.7739 0.2952 1 4.7567*** 122.6259*** 148.3539*** 0.8744 0.002*** (0.069) (0.062) Coverage Lines per worker (LINES) 229.3598 794.4770 1 8.8517*** 178.3535*** 0.0183 0.002*** (57.742) (169.273) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: First and second differences for changes in performance are shown, for each empirical proxy, using the mean values (see appendix table 8B.1 for variable definitions). Columns 1 and 2 present the mean values before and after privatization (the year of privatization is 1994). Columns 3 and 4 show results for two tests for significance in changes in performance after privatization (first differences): t-test (two-tailed Wilcoxon signed-rank test) and the Hotelling test. Column 5 displays significance of difference-in-difference results using a Hotelling test (control group is basd on Electro Oriente, Electro Sur, Electro Sur Este, and Electro Sur Medio for SALEFF, NIEFF, and EMPL). Column 6 presents Shapiro- Francia test for normality. Finally, column 7 shows the results for the Kolmogorov-Smirnov (K-S) statistic, a nonparametric test used to formally test the equality of the empirical hazards functions of the different pre- and postprivatization performance indicators. a. t-test for Ho (null hypothesis) for difference between means. Numbers are unequal. (X1 X2) (X1 X2) t S x1 x2 (n1 1) s1 2 (n2 1) s22 1 1 B (n1 n2 2) *Bn1 n2 where x is an l k matrix of the means and S is the estimated covariance matrix. b. Test of equality: T2 (x1 x2) S 1 (x1 x2) c. Shapiro-Francia test for normality. Ho (null hypothesis) variable is normally distributed. d. Thousands of nuevos soles. Source: Author's calculations. 471 472 Table 8E.3 Changes in Performance after Privatization for Electroperú (Differences between means and difference-in-difference tests) Means First differences Diff. in diff. Pre- Post- S-Franciac Kolmogorov- Sector privatization privatization t-testa Hotellingb Hotellingb prob z Smirnov Performance measure (Pi) Profitability Return on sales (ROS) 0.8485 0.2229 1.1486 111.3194 11.6025 0.0001 0.274 (0.605) (0.096) Return on assets (ROA) 0.0008 0.0300 1.9676* 113.8716* 11.6639 0.4842 0.234 (0.008) (0.012) Return on equity (ROE) 0.0021 0.0483 1.7341 113.0070 10.2457 0.5607 0.234 (0.015) (0.021) Operating efficiency Sales efficiency (SALEFF)d 205.7400 1222.7810 10.7568*** 115.7089*** 45.2077*** 0.0177 0.003*** (57.770) (51.656) Net income efficiency (NIEFF)d 26.6490 285.7193 1 2.4842** 116.1711** 13.3408* 0.6345 0.197 (65.580) (119.717) Employment Total employment (EMPL) 1976.7 593.00 1 4.6217*** 121.3599*** 19.9168*** 0.3306 0.003*** (194.342) (30.257) Leverage Debt to assets (LEV) 0.4757 0.4010 1.1977 111.4345 3.5220* 0.0858 0.749 (0.035) (0.049) Debt to equity (LEV2) 1.0039 0.7000 1.0279 111.0565 2.4812 0.0005 0.749 (0.185) (0.125) Coverage Lines per worker (LINES) 0.0034 0.0216 6.8787*** 147.3159*** 0.0031 0.003*** (0.002) (0.008) * Significant at the 10 percent level. ** Significant at the 5 percent level. *** Significant at the 1 percent level. Note: For explanation, see appendix table 8E.2. Source: Author's calculations. 473 474 TORERO Notes This chapter has been prepared for the project "Costs and Benefits of Privati- zation in Latin America" of the Inter-American Development Bank's Latin Ameri- can Research Network. Dean Hyslop gave valuable advice on the econometric analysis of these results; and Florencio López-de-Silanes, Alberto Chong, Alberto Pascó-Font, and all the participants in the IDB Research Network on Costs and Benefits of Privatization offered numerous and helpful comments. The author is also indebted to the enormous effort and support of Virgilio Galdo throughout this project, as well as to his remarkably talented team of research assistants: Daniel Oda, Jorge de la Roca, Gissele Gajate, and Linette Lecussan. Finally, the author thanks the Comisión de Promoción de la Inversión Privada (COPRI) for granting access to the White Books of all the privatized firms. 1. All dollar amounts are U.S. dollars unless otherwise indicated. 2. For the period 1987­92 the annual change in gross domestic product at constant prices was 4.9 percent, annual inflation was 733.1 percent (reaching a high of 7,649.6 percent in 1990), and the real effective exchange rate depreciation for the period was 2.04 percent. 3. According to Apoyo (2002), during the second half of 1985, the amount of banking deposits reached 23 percent of gross national product (GNP); five years later, in May 1990, bank deposits fell to 5 percent of GNP. A similar drop also oc- curred in the net internal credit of the banking system to the private sector (inter- est rates rose between 200 and 400 percent annually in real terms). 4. These figures are for the Lima metropolitan area; see Perú en Números 1991, Cuánto S.A. 5. Apoyo (2002). 6. The first 26 CEPRIs initiated their operations in February 1992. 7. This entailed the creation of the Commission for the Promotion of Private Concessions in 1997, which was later absorbed by COPRI. 8. The slowdown in privatizations between 1997 and 2000 is attributable to domestic and foreign factors, among them, the Russian financial crisis, "El Niño," and the Peruvian political crisis. 9. Telefonica was no stranger to the acquisition of Latin American telecom- munications providers, having already bought the former Teléfonos de Chile, cur- rently known as CTC, and Argentina's ENTEL. 10. Although the monopoly was initially scheduled to expire in June 1999, the TdP moved the expiration forward to August 1, 1998. 11. This transaction corresponds to the shares not purchased by the workers of Banco Continental and subsidiaries as part of their preferential right conferred by LD 674. In agreement with the contract, these shares had to be purchased by Hold- ing Continental S.A. at the auction price. 12. The privatization year is the year in which the government sold, for the first time, a certain amount of shares. 13. The test evaluates the closeness of the distributions pre priv and post priv by computing the least upper bound of all pointwise differences | l post ^ priv(x) l ^ pre priv(x)|. The K-S statistic can be written as: D supx[ l ^ postpriv (x) l ^ pre priv (x) ] The null hypothesis (H0 : l ^ post priv l ^pre priv) is accepted if lpost privis sufficiently close to pre-priv , in other words, if the value of D is sufficiently small or smaller than the critical value at a certain significance level. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 475 14. Heterogeneity may also be present but can be controlled by pooling the data. 15. For further details, see Arellano and Bond (1991, 1998). 16. This statistic is distributed Chi-squared in the number of overidentifying re- strictions. The null hypothesis is that the additional moment conditions are ap- proximately satisfied. 17. Hansen (1982) showed that the optimal weighting matrix for this class of estimators is W AsyVar[1/N Z e], where Z is the N L matrix of instruments and e is the N 1 matrix of the GMM residuals. For the procedure followed for N observations, the optimal W is given by: N W (1/N2)a ziziei r 2 i 1 where zi is the ith row of Z and ei is the ith element of e, and GMMIV saves W in e(W). 18. Among the companies acquired, the most important are Lar Carbon, Sia, and Nisa, acquired by Cementos Lima; Petrolube, acquired by Mobil Oil; Enata, acquired by Tabacalera del Sur S.A.; Compania Minera Mahr Tunel and Compa- nia Minera Paragsha, acquired by Volcan; and Planta de Cemento Rioja, acquired by Cementos Norte Pacasmayo. 19. An unweighted and locally weighted smoothing is carried out. 20. Barber and Lyon (1996) suggest that sample firms must be matched to con- trol firms with similar pre-event performance, which is especially difficult in stud- ies of privatized firms, but SEDAPAL went through the same reform as the priva- tized firms. 21. For this purpose a probit model was used to estimate the propensity score. For further reference, see Heckman and Hotz (1989); Heckman and others (1996); Heckman, Ichimura, and Todd (1997). 22. It is important to mention that Peru had possibly the most restrictive and protective labor legislation of any Latin American country. After the successive waves of reform in 1991 and 1995, no other country had so liberalized its labor market (Lora and Márquez 1998; Márquez and Pagés 1998; Saavedra and Torero 1999). Such drastic reform must be considered when looking at the impact of pri- vatization on employment. 23. Wages of the privatized firms increased significantly after privatization, in both absolute and relative terms with respect to the industry. On average, salaries increased 180 percent in the privatized companies and were 91 percent higher than the average of the specific industry of the privatized firm. 24. Note that, as in La Porta and López-de-Silanes (1999), the methodology overstates the contribution of layoffs, given the assumption that laid-off workers had zero productivity. References Arellano, M., and S. Bond. 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations." Review of Economic Studies 58: 277­97. ------. 1998. Dynamic Panel Data Estimation Using DPD98 for Gauss: a Guide for Users. Oxford, U.K.: Oxford University. Apoyo. 2002. Análisis de coyuntura, perspectivas económicas, tendencias. Apoyo, Lima. 476 TORERO Barber, B., and J. Lyon. 1996. "Detecting Abnormal Operating Performance: The Empirical Power and Specification of Test Statistics." Journal of Financial Eco- nomics 41: 539­99. Boardman, Anthony, and Aidan R.Vining. 1989. "Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises." Journal of Law and Economics 32: 1­33. Boubakri, Narjess, and Jean Claude Cosset. 1998. "The Financial and Operating Performance of Newly Privatized Firms: Evidence from Developing Countries." Journal of Finance 53 (3): 1081­110. D'Souza, Juliet, and William L. Megginson. 1999. "The Financial and Operating Performance of Privatized Firms during the 1990s." Journal of Finance 54 (4): 1397­438. Galal, A., L. Jones, P. Tandon, and I. Vogelsang, 1994. Welfare Consequences of Selling Public Enterprises. Washington, D.C.: Oxford University Press for the World Bank. Greene, W. 2000. Econometric Analysis, 4th ed. New York: Prentice-Hall. Guasch, José Luis, and P. Spiller. 1999. Managing the Regulatory Process: Design, Concepts, Issues, and the Latin America and Caribbean Story. Washington, D.C.: World Bank. Hansen, L. 1982. "Large Sample Properties of Generalized Methods of Moments Estimators." Econometrica 50: 1029­54. Heckman, J., and J. Hotz. 1989. "Choosing among Alternative Nonexperimental Methods for Estimating the Impact of Social Programs: The Case of Manpower Training." Journal of the American Statistical Association 84: 862­80. Heckman, J., and J. Smith. 1995. "Assessing the Case of Social Experiments." Journal of Economic Perspectives 9: 85­110. Heckman, J., H. Ichimura, and P. Todd. 1997. "Matching as an Econometric Eval- uation Estimator: Evidence from Evaluating a Job Training Program." Review of Economic Studies 64: 605­54. Heckman, J., R. LaLonde, and J. Smith. 1999. "The Economics and Econometrics of Active Labor Market Programs." In O. Ashenfelter and D. Card, eds., Hand- book of Labor Economics, vol. III. Elsevier, Amsterdam: North-Holland Press. Heckman, J., H. Ichimura, J. Smith, and P. Todd. 1996. "Characterizing Selection Bias Using Experimental Data." Econometrica 66: 1017­98. Kikeri, Sunita, John Nellis, and Mary Shirley. 1994. "Privatization: The Lessons from Market Economies." World Bank Research Observer 9: 241­72. La Porta, Rafael, and Florencio López-de-Silanes. 1999. "Benefits of Privatization: Evidence from Mexico." Quarterly Journal of Economics 114(4): 1193­242. Lora, Eduardo, and G. Márquez. 1998. "El problema del empleo en América Latina: Percepciones y hechos estilizados." Paper prepared for the meeting of the Governors of the Inter-American Development Bank, Cartagena, Colombia. Mackenzie, G. 1998. "The Macroeconomic Impact of Privatization." IMF Staff Papers 45 (2): 363­73. Márquez, G., and C. Pagés. 1998. Ties That Bind: Employment Protection and Labor Market Outcomes in Latin America. Paper prepared for the meeting of Governors of the Inter-American Development Bank, Cartagena, Colombia. PERUVIAN PRIVATIZATION: IMPACTS ON FIRM PERFORMANCE 477 Megginson, William, and Jeffry Netter. 2001. "From State to Market: A Survey of Empirical Studies on Privatization." Journal of Economic Literature 39: 321­89. Megginson, William, Robert Nash, and Matthias van Randenborgh. 1994. "The Financial and Operating Performance of Newly Privatized Firms: An Interna- tional Empirical Analysis." Journal of Finance 49(2): 403­52. Rubin, D. 1974. "Estimating Causal Effects to Treatments in Randomized and Nonrandomized Studies." Journal of Educational Psychology 66: 688­701. ------. 1977. "Assignment to Treatment Group on the Basis of a Covariate." Journal of Educational Studies 2:1­26. ------. 1979. "Using Multivariate Matched Sampling and Regression Adjustment to Control Bias in Observational Studies." Journal of the American Statistical Association 74: 318­28. Saavedra, J., and Máximo Torero. 1999. "Labor Market Reforms and Their Im- pact on Formal Labor Demand and Job Market Turnover: The Case of Peru." Working Paper R-394. Inter-American Development Bank, Research Depart- ment, Washington, D.C. Shleifer, Andrei. 1998. "State versus Private Ownership." Journal of Economic Perspectives 12(4):133­50. Torero, Máximo, and Alberto Pascó-Font. 2001. "The Social Impact of Privatiza- tion and Regulation of Utilities in Peru." WIDER Discussion Paper 2001-XX. Torero, Máximo, Enrique Schroth, and Alberto Pascó-Font. 2004. "The Impact of Telecommunications Privatization in Peru on the Welfare of Urban Con- sumers." Economía 4(1). Vickers, John, and George Yarrow. 1988. Privatization: An Economic Analysis. Cambridge, Mass.: MIT Press. Vining, Aidan R., and Anthony Boardman. 1992. "Ownership versus Competi- tion: Efficiency in Public Enterprises." Public Choice 73: 205­39. Acronyms and Abbreviations AFPs Administradoras de Fondos de Pensiones (Private pension funds) (Chile) BCRA Banco Central de la Republica Argentina BNDES National Social and Economic Development Bank (Brazil) Btu British thermal unit CAP Compañía de Acero del Pacífico (Chile) CEO Chief executive officer CEPRI Comité Especial de Privatizaciones (Special Privatization Committee) (Peru) CIIU Clasificación Industrial Internacional Uniforme (Manufacturing Industry Survey) (Bolivia) CONPES Concejo Nacional de Política Económica y Social (National Council for Economic and Social Policy) (Colombia) COPRI Comision de Promoción de la Inversión Privada (Commission for the Promotion of Private Investment) (Peru) CORFO Corporación de Fomento (Chile) CPI Consumer price index CPT Compañía Peruana de Teléfonos (Peru) CREG Comisión de Regulación de Energía y Gas (Regulatory Commission for Energy and Gas) (Colombia) CST Companhia Siderúrgica de Tubarão (Brazil) 479 480 ACRONYMS AND ABBREVIATIONS DANE Departamento Administrativo Nacional de Estadística (National Statistics Department) (Colombia) DEA Data Envelopment Analysis EAM Encuesta Anual Manufacturera (Annual Manufacturing Survey) (Colombia) EDS Encuesta de Desarrollo Social (Argentina) ENAP Empresa Nacional del Petróleo (Chile) ENDE Empresa Nacional de Electricidad (Bolivia) ENDESA Empresa Nacional de Electricidad S.A. (Chile) ENFE Empresa Nacional de Ferrocarriles (Bolivia) ENTEL Empresa Nacional de Telecomunicaciones (Bolivia, Chile, Peru) EPM Public Enterprises of Medellín (Colombia) FECUs Ficha Estadística Codificada Uniforme (Chile) FFDP Fondo Fiduciario para el Desarrollo Provincial (Argentina, a World Bank­IDB-created fund) FGLS Feasible generalized least squares FGTS Fundo de Garantia de Tempo de Serviço (Workers' Tenure Guarantee Fund) (Brazil) GDP Gross domestic product GLS Generalized least squares GMM-IV Generalized method of moments instrumental variables GNP Gross national product GWh Gigawatt hour IFI Instituto de Fomento Industrial (Institute for Industrial Promotion) (Colombia) ISA Interconexión Eléctrica S.A. (Colombia) IV Instrumental variable kWh Kilowatt hour LAB Lloyd Aéreo Boliviano (Bolivia) LAD Least absolute deviation ACRONYMS AND ABBREVIATIONS 481 LD Legislative decree MPU Municipal public utility MW Megawatt NPD National Program of "Destatization" (Brazil) OLS Ordinary least squares OSINERG Organismo Supervisor de la Inversión en Energía (Supervisory Agency for Private Investment in Energy) (Peru) OSIPTEL Organismo Supervisor de Inversión Privada en Telecomunicaciones (Supervisory Agency for Private Investment in Telecommunications) (Peru) OSN Obras Sanitarias de la Nación (Argentina) PCS Personal communications services PPA Power purchase agreement PPE Property, plant, and equipment PPI Producer price index PQ Privatization Q PVR Present value of revenue ROA Return on assets ROE Return on equity RPU Regional public utility SEDAPAL Empresa de Servicio de Agua Potable y Alcantarillado de Lima (Peru) Sendos Servicio Nacional de Obras Sanitarias (Chile) SIC Standard Industrial Classification SIRESE Sistema de Regulación Sectorial (Bolivia) SOE State-owned enterprise Subtel Subsecretaría de Telecomunicaciones (Chile) SUNASS Superintendencia Nacional de Servicios de Saneamiento (National Office for Services of Sanitation) (Peru) TdP Telefónica del Perú S.A. (Peru) 482 ACRONYMS AND ABBREVIATIONS TFP Total factor productivity WSP Water and sewerage provider YPF Yacimientos Petrolíferos Fiscales (Argentina) YPFB Yacimientos Petrolíferos Fiscales Bolivianos Index Note that f indicates figure, n indicates note (nn is notes), and t indicates table. A Annual Survey of Social Data accounting information (RAIS), Brazil, 163, Chile, 200, 213­14 193n5 and noncomparable data, 17, antiprivatization sentiment, 409 58n15 Antitrust Commission, Chile, Peru, 423 225­26, 272n19 accounting standards, 56, 213 Anuatti-Neto, Francisco, 17, 28, Aerolineas Argentinas, 107 145­96 AFPs. see private pension funds Apoyo, 474n3 (AFPs) Argentina Africa, 4, 58n6 airlines sector, 107 age, and worker displacement, banking sector, 69, 79t2.4, 97­98, 99, 114n26 86­90, 108, 112, 113n19 agency, and privatization, 119­20 data collection, 15 Aguas Argentinas, 91, 92­93, 106­7 electricity sector, 106 Aguas de Corrientes, 92 employment, Aguas del Illimani, Bolivia, 127­28 banking sector, 87, 89t2.8, 90 Aguas del Tunari, Bolivia, 127 displaced workers, 96­103, airlines sector, 36 104, 114n24­29 Argentina, 107 gas sector, 105 Chile, 219 oil industry, 108 airports, Chile, impact of postal services, 108 privatization on, 238­42, privatization 268t5A.9, 269t5A.10, program overview, 70­73 272nn26­30 results for nonfinancial firms, Allende, Salvador, 197, 198, 205 74­86, 112­13nn6­18 allocative efficiency, Mexico, sample studied, 73­74, 370­72 75­76t2.2, 77­78t2.3, Alonso, Juan, 277 79t2.4, 112nn3­4 Annual Manufacturing Survey study overview, 67­70 (EAM), 291, 344n15 railroad sector, 107, 112nn3­4 483 484 INDEX Argentina (continued) Banco do Brasil, 172 sanitation services, 68, 70, banking sector 90­96, 106­7 Argentina, 108, 112 shipping, 108 impact of privatization on, steel manufacturing, 108 86­90, 104, 113n19 telecommunications sector, 85, included in research 105­6 database, 79t2.4 water services, 68, 70, 90­96, operating efficiency of, 69, 106­7, 113n17 87, 88t2.7, 89t2.8, 90 Argentine Financial System, 86 Brazil, 172 Arin, Kermin, 40 Chile Ariyo, Ademola, 55­56 government take over of, Artana, D., 92 208 assets regulation of, 208­9, 271n9 Brazil, 155, 186­92tt4C.1­4C.4 and corporate governance, 56 Chile, 218­19, 254f5A.2, Mexico, 382­83, 400n4, 255­56t5A.3, 257f5A.3, 402­3n25 263t5A.6, 272n18 loans, 384­88, 384f7.6 Mexico, 356t7.4, 357 performance indicators, 383, changes in, 359, 360t7.5 384f7.6 in competitive vs. reregulation of, 382­83, noncompetitive industries, 402­3n25 359, 361­62t7.6, 363 ownership of, 6 Peru, 433t8.6 Peru, 424, 430t8.5, 431, assets to employment ratio, Peru, 438­44, 460­61t8C.2, 474n3 430 performance of, 442­43t8.9 auctions bankruptcy, 56, 391 Bolivia, 123 Barber, B., 475n20 Brazil, 145, 146­47, 151, Barberis, Nicholas, 120 152t4.2, 165­66 Barja, Gover, 121 proceeds from, 171­72 Barossi-Filho, Milton, 17, 145­96 Chile, 202­3, 207, 211 Basser, G., 81 infrastructure, 238, 242, Bayliss, Kate, 36, 37, 49 272n27 BCRA. see Banco Central de la Mexico, 374, 375t7.12, 376, Republica Argentina (BCRA) 402n20 Berg, Sanford, 276 Peru, 414 bias, 6, 8­17, 407 requirements of, 40­41, 374, bidding process, 39 375t7.12, 376, 402n20 Birdsall, Nancy, 37, 59n19 for SOEs, 40­41 blue-collar workers Bolivia, 133 B Colombia, 304 bailouts, 27 layoffs, 32 Baltagi, B. H., 194n13 Mexico, 364 Banco Central de la Republica Peru, 444­45 Argentina (BCRA), 86 BNDES, 151, 171­72, 173, Banco Continental, 445, 446t8.11 180­81t4A.2, 194n9 INDEX 485 Bogota Power Company, 288, 305­6 at federal level, 147­48, Bolivia 175­79t4A.1, data collection, 15 180­81t4A.2 employment, 133, 135t3.4, 136, impact of capital markets on, 139­41t3A.2 168, 170 foreign investment, 125 impact of fiscal discipline on, privatization 169 adoption of capitalization as impact on employment, method of, 123­28 163­65, 166f4.1, analysis of study results, 194­95nn15­16, 130­36, 139­41t3A.2, 194­95nn15­16 142nn11­13 impact on prices, 165­67 arguments for and against, literature review of, 148­49, 119­20, 142n2 194nn6­7 data collection, 128­29, perspectives on, 171­73, 142nn8­10 195n22 empirical studies of, 121 program description and industry-adjusted data, sample coverage, 150, 134­36, 142n13 150t4.1, 151 list of privatized firms, public opinion of, 170­71, 137­39t3A.1 174, 195nn20­21 objectives of, 118, 119t3.1 revenues from, 28 overview, 121­24 study data set and variables, variables used to evaluate 149­53, 194nn8­9 impact of privatization, summary and conclusions 139­41t3A.2 about, 173­74 sanitation services, 128 Brazilian elites, 171 telecommunications sector, British Airways, 120 126­27 British Telecom, 302 water sector, 127­28 budget deficits, Brazil, 169 Bonilla, J., 277 budgets, Chile, 198 Bortolotti, Bernardo, 40 bureaucratic quality index, 40 Boubakri, Narjess, 120, 418 Brazil C data collection, 17 Campos-Méndez, Javier, 56 net-taxes-to-sales ratio, 28 Capital Federal, Argentina, 91, 95 privatization capital markets benefits and costs of, 162­69 absence of, 55­56 and capital ownership, Brazil, 168, 170 167­68 Chile, 198 components of, 146­47 and methods of capitalization, costs of, 162­69 40 empirical analysis of data, Mexico, 389­91 153­62, 153t4.3, capital ownership, Brazil, 167­68, 184­86t4B.1, 170 186­92tt4C.1­4C.4, capital-to-labor ratio, Colombia, 194nn10­11, 194nn13­14 299, 303 486 INDEX capitalization historical perspective, Argentina, banking sector, 90 197­98, 203­6, 206­13, Bolivia, 122, 123, 125 271nn6­10 as means of privatization, impact on airports, 238­42, 123­28 268t5A.9, 269t5A.10, Mexico, 389, 390f7.9 272nn26­30 Capra, Katherina, 15, 117­44 impact on assets and Cardoso, Fernando Henrique, 169 investments, 218­19, Carey, Alan, 56 254f5A.2, 255­56t5A.3, Cauca Valley Corporation, 305­6 257f5A.3, 263t5A.6, cellular phones, 447 272n18 cement mills, Colombia, 301­2 impact on efficiency, 218, Central Interconnected System, 252­53t5A.2, 261t5A.5 Chile, 223 impact on electricity sector, CEPRIs. see Special Privatization 221­22 Committees (CEPRIs), Peru impact on employment, child mortality 219­20, 265t5A.7 Argentina, 68, 70, 104, 113n17 impact on firms, 216­20, and water delivery services, 38, 249­67(Appendix 5A) 95 impact on highways, 238­42, Chile 268t5A.9, 269t5A.10, airline sector, 219 272nn26­30 Antitrust Commission, 225­26, impact on infrastructure, 272n19 202­3, 237­38, 271nn4­5 banking sector, 208­9, 271n9 impact on operating data collection, 15 efficiency, 218, electricity sector, 198, 200, 201, 252­54t5A.2, 261t5A.5 205, 212 impact on productivity, 219, node prices, 224, 227, 254f5A.2, 266­67t5A.8 229t5.12 impact on privatization process, 221­22 telecommunications and productivity, 227, sector, 221­22 228t5.10, 229t5.11 of SOEs, 199­200, 207t5.4, regulatory framework, 210­13, 271n2, 271n10 222­25, 226­30, and state intervention in 231t5.13, 272nn20­11 economy, 203­6, residential rtes, 227, 271nn6­7 229t5.12 of utilities industry, 201­2, ports, 202, 203, 271n4, 242­46, 271n3 272nn31­32 regulatory framework, 214, privatization, 20, 21f1.5 247­48 concessions in operations, sales-to-employment ratio, 20, 238­42, 268t5A.9, 21f1.5 269t5A.10, 272n26 Chilean Securities and Exchange data on privatized firms, Commission, 214 213­16, 271nn11­14, Chilectra, Chile, 205, 215, 221, 271­72nn16­17 227, 229t5.11, 230 INDEX 487 China, 4 natural gas and gasoline Chisari, Omar, 54 distribution, 284­86, Chong, Alberto 343nn10­12 "Privatization and Firm overview, 277­79, 343n3, Performance in Bolivia", 343n5 117­44 performance in power sector, "Privatization in Mexico", 305­13, 314t6.11, 349­405 345n25, 345n27 "The Truth about Privatization and productivity in in Latin America", 1­66 manufacturing sector, Claessens, Stijn, 120 294­95t6.6, 296t6.7, CNT, Chile, 221, 236 299­303, 344n22 Codelco, Chile, 204­5, 212 restructuring of power sector, Colbún, Chile, 212, 215 286­88, 289­90t6.4, Colombia 344n13 concession contracts, 278, profitability gap between 329t6A.1 privatized and private firms, data collection, 15 24­25, 58n16 and deregulation, 277­79, regulatory framework, 275, 276, 343n3, 343n5 343nn1­2 economic liberalization of, thermal plants 275­76, 343nn1­2 input and output variables, infrastructure, 278, 329t6A.1 339t6D.2 performance indicators for IFI statistical analysis of power firms, 332, 345n35 plant efficiency scores, privatization 324­27, 345n34 data sets, 291­92, 336, variables, electricity sector, 313, 337t6D.1, 344n15 314t6.11 econometric analysis of Commission for the Promotion of manufacturing sector, Private Investment (COPRI), 315­23, 345n31, 345n33 Peru, 410, 411 efficiency gains in Compañía Peruana de Teléfonos manufacturing sector, (CPT), 413­14, 424 197t6.7, 294­95t6.6, 299 competition, 408 impact on manufacturing Chile, 201, 225, 236­67 sector, 292­305, 344n17 Colombia, 316, 328 impact on profitability, 19, and deregulation, 53 293, 294­95t6.6, 295, Mexico, 359, 361­63, 296­97t6.7, 298 401n14 impact on thermal plants, and politics, 53­54 311, 312f6.5, 313­15, CONASTIL, Colombia, 284 340­42t6E.1, 345n27, concession contracts, 51, 54­55 345n29 Chile, 198 indicators for IFI firms in in construction of highways sample, 333­36t6C.1 and airports, 238­42, manufacturing, 279­84, 302, 268t5A.9, 269t5A.10, 343nn8­9 272n26 488 INDEX concession contracts (continued) CPT. see Compania Peruana de for port management and Telefonos (CPT) operation, 242­46, credit market, Argentina, 90 272nn31­32 creditor rights, Mexico, 389­91 utility services, 223 cross-country firm-level analyses, 39 Colombia, 277­78 bias of, 8, 58n9 infrastructure, 278, 329t6A.1 cross-section studies, 120 Mexico, 379­80 cross-subsidies, 28 Peru, 413, 417­18, 474n7 Crosser, Jean-Claude, 120 conglomerates, Chile, 208­9 CTC, Chile, 221 consumer price index (CPI), 401n9 currency, devaluation of, 169, Argentina, 80 195n19 Brazil, 166­67 consumer welfare, 36­39, D 59nn18­19 da Silva, Lula, 173 consumers, exploitation of, 32­36, data 58­59n17 adjusted ratios, 23, 24­36 contract theory, 119 availability of, and sample contracts selection, 8, 58n9 Argentina, for privatized SOEs, firm-level data, 40 73­74, 112nn3­4 industry-adjusted Bolivia, shared-risk, 122 Bolivia, 139­41t3A.2 Mexico, 353, 354t7.3 Brazil, 163 types of, 378­80 Colombia, 296­97t6.7, 302 types of, 49, 51, 378­80 Mexico, 358­59, 360t7.5, see also concession contracts 401n13 cooperatives, Bolivia, 125, 127 noncomparable, 17­18, 58n15 COPRI. see Commission for the nonpublic information, 15 Promotion of Private on privatization in Latin Investment (COPRI), Peru America, 15, 16f1.3 CORELCA, Colombia, 288 raw data, 19­23 Corporación de Fomento within-country data, 40 (CORFO), Chile, 204­6, see also names of specific 271n7 countries Corporación Nacional Forestal, De la Madrid administration, 380 Chile, 206 De Souza, A., 171 Corporación Regional de la Costa DEA efficiency scores, 324­27, Atlantica, 305­6 345n34 corporate governance, 55­56 debt Mexico, 388­91 Brazil, 169, 173 corporate structure, Colombia, 298 capitalization of, 206 Correos y Telégrafos, Chile, 221 Mexico, 369­70, 371t7.3, 377 corruption, 39 SOEs, 377 lack-of-corruption index, 40 debt-to-asset ratio, Chile, 208 Cosser, Jean Claude, 418 decentralization costs per unit. see unit costs Bolivia, 121­22 CPI. see consumer price index (CPI) Mexico, 355 INDEX 489 demand-side growth, Colombia, Chile 302, 344n23 impact of privatization on, deregulation, 52­55, 59n24 218, 252­53t5A.2, Chile, 233, 236 261t5A.5 Colombia, 277­79, 343n3, port services, 246 343n5 Colombia, 311, 328 Mexico, 380­81, 389 manufacturing sector, developing countries, proceeds 197t6.7, 294­95t6.6, 299 from privatization, 5, 5t1.1 power sector, 306 difference-in-difference estimates, statistical analysis of power 112n6 plant efficiency scores, Argentina, 82, 83t2.6, 84, 85, 324­27, 345n34 87, 90, 92, 94t2.10 and deregulation, 52­53 Bolivia, 130 Mexico, 402n19 Peru, 435­36t8.7, 437 and profits, 23 disclosures, 51 electricity sector and corporate governance, 56 Argentina, 85, 106 displaced workers, 29­32 Bolivia, 124, 125­26 divestitures, Chile, 215 reform through dividends-to-net-operating-income capitalization, 125 ratio, Brazil, 155­56, Brazil, 164t4.5, 165, 172­73 186­92tt4C.1­4C.4 impact of privatization on Djankov, Simeon, 120, 391 prices of, 167 downsizing programs, 43, Chile, 198, 200, 201, 205, 212 44­47t1.4, 48, 49 node prices, 224, 227, D'Souza, Juliet, 120, 408 229t5.12 privatization process, 221­22 E and productivity, 227, EAM. see Annual Manufacturing 228t5.10, 229t5.11 Survey (EAM) regulatory framework, Earle, John, 58n4 222­25, 226­30, Eckel, Catherine, 120 231t5.13, 272nn20­22 Eckel, Doug, 120 residential rates, 227, economic growth 229t5.12 Bolivia, 121, 134 Colombia Brazil, 168, 170 efficiency variables, 313, Chile, 198 314t6.11 Mexico, 352­53, 400nn1­2 restructuring of, 286­88, ECOPETROL, Colombia, 284­86, 289­90t6.4, 343n13 343nn10­11 impact of privatization on, 36, Edelnor, Chile, 212 38 EDS. see Encuestra de Desarrollo Peru, 413, 416­17, 416f8.3, Social (EDS) 431, 437, 470­71t8E.2, education, Mexico, 371­72 472­73t8E.3 efficiency gains privatization in, 458­59t8C.1 Argentina, 69, 80, 104 Electrolima, Peru, 437, Bolivia, 130, 134 470­71t8E.2 490 INDEX Electroperú, 437, 472­73t8E.3 Empresa Ferroviaria Andina, Eletrobras, Brazil, 172 Bolivia, 138t3A.1 employee-wealth-transfer Empresa Ferroviaria Oriental, hypothesis, 382­84 Bolivia, 138t3A.1 employment, 408 Empresa Nacional de Correos y Argentina Telégrafos, 108 banking sector, 87, 89t2.8, 90 Empresa Nacional de Electricidad displaced workers, 96­103, (ENDE), Bolivia, 124, 104, 114nn24­29 125­26 impact of privatization on, 84 Empresa Nacional de as measure of profitability, 81 Telecommunicaciones, Bolivia, 133, 135t3.4, 136, Bolivia, 124, 126­27, 139­41t3A.2 138t3A.1 Brazil, 163­65, 166f4.1, Empresa Nacional de 194­95nn15­16 Telecommunicaciones changes in postprivatization, 21, (ENTEL), Peru, 413­14, 22f1.6, 23 424 Chile, 200, 219­20, 265t5A.7 Encuestra de Desarrollo Social Colombia, 306, 307­8t6.9, (EDS), 93­94 309­10t6.10 ENDE. see Empresa Nacional de competitive vs. noncompetitive Electricidad (ENDE), Bolivia industries, 34, 34f1.13 ENDESA, Chile, 215, 221, 223 displaced workers, 29­32 Enersis, Chile, 223 hiring policies, 43, 48, 48f1.14, Engel, Eduardo, 51, 55, 241 50t1.6, 59n23 ENTEL. see Empresa Nacional de layoffs, 32 Telecommunicaciones Mexico, 356t7.4, 357 (ENTEL), Peru changes in, 359, 360t7.5 ENTEL, Chile, 221, 236 in competitive vs. ENTEL PCS, Chile, 235 noncompetitive industries, enterprises 359, 361­62t7.6, 363 Argentina, impact of during Salinas privatization on, 111t2B.2 administration, 354­55 availability of data, and sample impact of privatization on, selection, 8, 58n9 375t7.12, 376­78, Bolivia 402nn21­22 analysis of study results, postprivatization, 383, 130­36, 139­41t3A.2, 403n26 142nn11­13 role of transfers from data sample, 128­29, workers, 363­64, 365t7.7, 142nn8­10 402n15 list of privatized firms, Peru, 410, 433t8.6, 448 137­39t3A.1 impact of privatization on, Brazil, list of privatized firms, 444­47, 475nn22­24 151, 152t4.2 power sector, 306 Chile, data on privatized firms, postprivatization, 29 213­16, 271nn11­14, see also labor force 271­72nn16­17 INDEX 491 Colombia Fernandes, R., 194n15 IFI list used in sample, Ferrocarriles Argentinos, 107, 112n2 330­31t6B.1 Ferronor, Chile, 211 sale of under privatization FFDP. see Fondo Fiduciario para el program, 278, 279t6.1 Desarrollo Provincial (FFDP) and data collection efforts, 15, FGTS. see Workers' Tenure 16f1.3 Guarantee Fund (FGTS), Mexico Brazil auction process and Ficha Estadística Codificada requirements, 374, Uniforme (FECUs), Chile, 375t7.12, 376, 402n20 213­14, 215 data collection, 422­30, financial sector 449­50(Appendix 8A), Peru, 410, 413, 416f8.3, 418, 475nn18­19 424, 430t8.5, 438­44, firm and industry 460­61t8C.2, 474n3, characteristics, 374, 474n11 375t7.12 impact of privatization on, and market power, 359, 447 361­63, 401n14 see also banking sector raw data concerning, firm-level data, 18, 40, 58n15 355­58, 401nn9­12 firms. see enterprises restructuring programs, fiscal management 375t7.12, 376­78, Brazil, 169 402nn21­23 Chile, 198, 271n1 mixed-capital, 280 Mexico, 367­70, 371f7.3 noncomparable data concerning, Fischer, Ronald, 15, 51, 53, 55, 17­18, 58n15 197­274 performance after privatization, Fondo Fiduciario para el Desarrollo 8, 9­14t1.2 Provincial (FFDP), 87 see also state-owned enterprises foreign direct investment (SOEs) Brazil, 168­69 EPM. see Public Enterprises of Mexico, 369, 370t7.10 Medellin (EPM), Colombia foreign investment equity transfer programs, Bolivia, 125, 142n13 Colombia, 280, 284, 302 Colombia, 320 Estache, Antonio, 54, 56 Peru, 410, 411 Europe, and firm-level data, 18, foreign trade, 316 58n15 franchises ex ante rents, 272n32 Chile, 237­38, 272nn27­28 export diversification, 295 United Kingdom, 272n29 free press, Chile, 200 F Frei administration, 202 Fantini, Marcella, 40 freight road transport, Mexico, FECUs. see Ficha Estadistica 366­67t7.8 Codificada Uniforme frentes de atraque, 243­45 (FECUs), Chile Frydman, Roman, 120 Fepassa, Chile, 211­12 Fujimori, Alberto, 409, 410 492 INDEX G Government Development Plan Galal, Ahmed, 36, 120, 407 (1990), Colombia, 277 Galetovic, Alexander, 51, 55, 241 Greene, W., 422 Galiani, Sebastián, 15, 37­38, gross domestic product (GDP) 67­116 Brazil, tax burden as percent of, Garay, Luis Jorge, 344n23 173, 195n23 Garrón, Mauricio, 15, 117­44 Chile, losses of financial sector gas distribution industries, Brazil, as percent of, 208 165 market capitalization as percent Gas Natural, Colombia, 284, 286 of, 389, 390f7.9 gas sector, Argentina, 105 and SOEs, 3­4, 4f1.1 gasoline distribution, Colombia, gross margin rate, manufacturing 284­86, 343nn10­12 sector, 294t6.6, 296t6.7, GDP. see gross domestic product 344n19 (GDP) Grubel, Herbert, 316 Gehlbach, Scott, 58n4 Guasch, José Luis, 51, 55, 343n2 Gerchunoff, P., 71 Gutierrez, Luis, 276 Gertler, Paul, 15, 67­116 Gutiérrez, Rodrigo, 15, 37, Giambiagi, F., 148 197­274 Gledson de Carvalho, Antonio, 17, 145­96 H GMM-IV estimation, 422 Hachette, Dominique, 246, 271n2 goods and services Hansen's J statistic, 422 demand for, 320 Heckman, J., 420­21 public vs. private provision of, Hidronor, Argentina, 106 119­20 highways government, 40 Chile, 202, 238­42, 268t5A.9, Brazil 269t5A.10, 272nn26­30 and fiscal management, 169 Mexico, 379 privatization at federal level, Hotelling test, 420 147­48, 175­79t4A.1, hydrocarbons industry, Bolivia, 180­81t4A.2 124, 126 Chile, state intervention in Hydrocarbons Law, Bolivia, 126 economy, 203­6, 271nn6­7 maintenance of infrastructure, 49 I Mexico, 352, 400n3 Ichimura, H., 421 expropriation of private Icollantas, Colombia, 343n8 companies, 355, 401n6 IFI. see Instituto de Fomento fiscal impact on privatization Industrial (IFI), Colombia programs, 367­70 immigration, Peru, 411 and SOEs, 351­53, 400nn1­4 import substitution, 295 and types of privatization incentives, 51 contracts, 378­80 income Peru, role in privatization Argentina, banking sector, 87, process, 410­11 88t2.7, 89t2.8 and privatization contracts, 49, see also wages 51, 378­80 income guarantees, 51 INDEX 493 India, and SOEs, 4­5 performance indicators for IFI industry-adjusted data, 23, 24­26 firms, 332, 345n35 Bolivia, 139­41t3A.2, 142n13 Instituto Nacional de Investigacion Brazil, 163 Agropecuaria survey, changes in wages, 29, 30f1.10, 271n13 32 insurance mechanisms, 51 Colombia Inter-American Development Bank, for IFI firms in, 296­97t6.7, 121­22 302 Interbanc, 445, 446t8.11 manufacturing sector, 293, Interconexión Eléctrica S.A., 296­97t6.7, 344n17, Colombia, 286­88 344n19, 344nn22­24 intercountry studies, 120­21 power sector, 306, interest rates, Chile, 208, 209 309­10t6.10, 311, 313 investment-spending-on-machinery- Mexico, 358­59, 359, 360t7.5, to-total-investment ratio, 401n13 Colombia, 302 inflation, Chile, 200 investment-to-employee ratio infrastructure Chile, 218 Bolivia, 125 Colombia, 301 Chile, 198 investment-to-physical assets (PPE) franchises, 238, 272n27 ratio impact of privatization on, Chile, 218, 219, 254f5A.2 202­3, 237­38, 238­42, Mexico, 357 268t5A.9, 269t5A.10, investment-to-sales ratio 271nn4­5, 272nn26­30 Chile, 218, 218­19, 257f5A.3 Colombia, concession contracts Colombia, 344n23 for, 278, 329t6A.1 investments, 23, 44 government maintenance of, 49 Argentina, 85 Mexico, and privatization as measure of performance, contracts, 379­80 81 Peru, 411 and success of SOEs, 71 and privatization contracts, 49, Bolivia, 133­34, 139­41t3A.2 51, 379­80 Brazil, 185t4B.1 railroad, 37 Chile, impact of privatization Instituto de Fomento Industrial on, 218­19, 254f5A.2, (IFI), Colombia, 279­84, 255­56t5A.3, 257f5A.3, 327­28, 343nn8­9 263t5A.6, 272n18 analysis of performance in Colombia, power sector, 306, manufacturing, 292­305, 307t6.9, 309­10t6.10 344n17 Mexico, 356t7.4, 357, 369 data sets, 291­92, 336, changes in, 359, 360t7.5 337t6D.1, 344n15 in competitive vs. list of firms used in sample, noncompetitive industries, 330­31t6B.1 359, 361­62t7.6, 363 manufacturing plants, impact of privatization on, econometric analysis of, 377 315­23, 345n31, 345n33 Peru, 409, 413 494 INDEX investments (continued) LAN Chile, 219 impact of privatization on, land reform, Chile, 205, 206 313f8.1, 411, 412t8.1, Larraín, Felipe, 198 413 last-mile technology, 237 iron and steel industry layoffs Argentina, 108 Argentina, 84, 85, 96­97 Brazil, 156­57 Bolivia, 133 ISNSA, Chile, 214 British Telecom, 302 Mexico, 364 J Peru, 445, 446t8.11 Jerome, Afeikhena, 55­56 and profitability after privatization, 29, 31f1.11 K Lerner indexes, 293, 294t6.6, Koenker,, R., 81 296t6.7, 315, 344n17 Kolmogorov-Smirnov (K-S) test, Levine, Ross, 168 420, 432 Lloyd Aéreo Boliviano, 124, Kornai, J., 195n18 138t3A.1 Lloyd, Peter, 316 L loans, 28 La Porta, Rafael, 6­68, 8, 15, 29 Mexico, 385­88, 391 on creditor rights, 391 López-Calva, Luis F., 3 employment issues, 445 López-de-Silanes, Florencio, 67­68, intercountry studies, 120 82, 85 labor force in Mexico, 133 employment issues, 445 on positive aspects of intercountry studies, 120 privatization, 407­8 labor force in Mexico, 133 preprivatization process, 214­15 on positive aspects of privatization study in Mexico, privatization, 407­8 82, 85 preprivatization process, 214­15 use of difference in difference "Privatization and Firm estimators, 112n6 Performance in Bolivia", labor costs, 28­29, 31f1.11 117­44 Colombia, 304 "Privatization in Mexico", labor force 349­405 Argentina, 96­103, 104, "The Truth about Privatization 114nn24­30 in Latin America", 1­66 Mexico, 377 use of difference in difference postprivatization, 21, 22f1.6 estimators, 112n6 and restructuring prior to Lora, Eduardo, 145, 171 privatization, 42­43, Lüders, Rolf J., 16, 246, 271n2 44­47f1.4 Lyon, J., 475n20 unskilled workers, 32 see also employment M labor productivity. see productivity Macedo, Roberto, 17, 145­96 lack-of-corruption index, 40 Machicado, Carlos, 15, 117­44 Lalonde, R., 421 management Lamounier, B., 171 banking sector, 385 INDEX 495 Mexico, 376 privatization of SOEs, 3 auction process and manufacturing sector requirements, 374, Colombia 375t7.12, 376, 402n20 econometric analysis of, data on firms, 8 315­23, 345n31, 345n33 definitions of variables, performance in, 292­305, 394­400t7A.1 344n17 firm and industry privatization data sets, characteristics, 374, 291­92, 336, 337t6D.1, 375t7.12 344n15 fiscal impact of, 367­70, privatization of, 279­84, 371f7.3 343nn8­9 historical perspective, profitability of, 25 353­55, 401nn5­6, 401n8 market power impact on corporate abuse of, 32­36, 58­59n17 governance, 388­91 Mexico, 359, 361­63, 381­82, impact on employment, 401n14 375t7.12, 376­78, 383, market share, manufacturing sector, 402nn21, 403n26 294, 294­95t6.6, 295, industry-adjusted data, 296t6.7 358­59, 360t7.5, 401n13 markups and market power, 359, Colombia, 293, 295, 298f6.1 361­63, 401n14 for IFI firms, 315, 316, raw data utilized, 355­58, 317­19t6.12, 320 401nn9­12 Martin, Stephen, 120 restructuring programs media, Brazil, 171 utilized, 372­78, Megginson, William L., 120, 145, 402nn18­23, 350 174, 407, 408, 432 role of transfers from Melón Tunnel, Chile, 240, 272n27 workers, 363­64, 365t7.7, Menem, Carlos, 70­71 402n15 mergers and acquisitions, 17 social impact of, 357 Brazil, 156 types of contracts written, Chile, 215 378­80 Mexico public services, 364, 366­67t7.8 and allocative efficiency, 370­72 reregulation, 381­88, banking sector, 382­83, 402­3nn25­26 384f7.6, 384­88, 400n4, restructuring programs, 402­3n25 375t7.12, 372­78, data collection, 15 402nn18­23, 350 deregulation, 380­81 sales-to-employment ratio, 20, employment, 354­55, 375t7.12, 21f1.5, 357 376­78, 402nn21­22 variables, 356t7.4, 359, 360t7.5, growth of SOEs, 351­53, 393, 394­400t7A.1 400nn1­4 military government, Chile, 206 net-taxes-to-sales ratio, 28, Miller, Merton, 389 357 mining sector, Peru, 413, 416f8.3 496 INDEX mixed-capital enterprises, 280, 298 Netter, Jeffry M., 145, 174, 407, mobile phones, 232, 235, 272n23 432 Modigliani, Franco, 389 Newberry, David, 120 monitoring mechanisms, 51 nickel processing plants, 302 monopolies, Chile, 225, 248 node prices, Chile, 224, 227, Montero, Marcelo, 121 229t5.12 nominal interest rates, Mexico, N 382 Nash, Robert C., 120 nonfinancial firms National Council for Economic and Argentina Social Policy, Colombia, 277 data collection, 74, National Institute for Social 74­75t2.2, 77­78f2.3 Security, Mexico, 352 impact of privatization on, National Program of Destatization 74­86, 109­10t2B.1, (NPD), 146 112­13nn6­18 natural gas sector Chile, 203 Colombia, 284­86, 343nn10­12 Peru, 424, 427t8.4 export industry, 126 included in privatization Mexico, 366­67t7.8 study, 424, 425­26t8.3 Navajas, F., 92 not included in privatization Nellis, John, 37, 59n19 study, 427t8.4 net-income-to-equity ratio, Chile, NPD. see National Program of 216t5.6 Destatization (NPD) net-income-to-physical assets (PPE) ratio O Chile, 217, 260f5A.4 Obras Sanitarias de la Nacion Mexico, 359 (OSN), 91, 106­7, 114n20 net-income-to-sales ratio, 19, 20f1.4 oil industry Argentina, 69, 80, 82, 82t2.5 Argentina, 108 Chile, 215 Brazil, 172 Colombia, 293, 294t6.6 Okten, Cagla, 40 gains in, 29 oligopolistic powers, 52, 53 Mexico, 358­59 operating costs, Mexico, 357, pre- and postprivatization, 24, 401n12 25f1.8 operating-costs-to-sales ratio, and worker transfers, 29, Brazil, 155 31f1.11 operating efficiency, 19, 20f1.4 net prices, 40­41, 59n20 Argentina, 81, 84 and labor force restructuring banking sector, 87, 88t2.7, preprivatization, 42­43, 89t2.8, 90 44­47t1.4 Bolivia, 130 and restructuring methods, 42 Brazil, 154­55, 161­62, net-taxes-to-sales ratio, 156­57 185t4B.1, 186­87t4C.1 Brazil, 28 changes after privatization, 20, Mexico, 28, 357 21f1.5 net-worth-to-assets ratio, Chile, 218, 252­54t5A.2, Argentina, 90 261t5A.5 INDEX 497 Colombia, 306, 307t6.9, Brazil, 155, 185t4B.1, 309­10t6.10 186­92t4C.1­4C.4 Mexico, 356t7.4, 357 competitive vs. noncompetitive changes in, 359, 360t7.5 sectors, 34, 35f1.13 in competitive vs. Mexico, 356t7.4, 357 noncompetitive industries, changes in, 359, 359, 361­62t7.6, 363 360t7.5 Peru, 429, 430, 433t8.6 in competitive vs. operating-income-to-capital stock noncompetitive industries, ratio, Colombia, 293, 359, 361­62t7.6, 363 294t6.6 Peru, 429, 430, 433t8.6 operating-income-to-employment outsourcing, Brazil, 163 ratio ownership Bolivia, 130, 131­32t3.3, 133 banking sector, 6 Chile, 219 capital ownership, 167­68, 170 Colombia, 306 electricity sector, 126 operating-income-to-physical assets power sector, 311 (PPE) ratio private vs. public, 146 Argentina, 80 Bolivia, 139t3A.2 P Brazil, 154, 185t4B.1 Pan-American Highway, Chile, Chile, 217, 262f5A.5 240 Colombia, 311 Panizza, U., 145, 171 Mexico, 359 PAPELCOL, Colombia, 284 operating-income-to-sales ratio, 19, Parker, David, 120 20f1.4, 24­25 partial input productivity, Argentina, 69, 80, 82, 82t2.5, Colombia, 294­95t6.6, 85 296t6.7, 299, 344n22 Bolivia, 130, 131­32t3.3, Pasco-Font, Alberto, 37 139t3A.2 passenger road transport, Mexico, Brazil, 161 366­67t7.8 changes in, 32­34, 58­59n17 payrolls Chile, 217, 264f5A.6 Colombia, 297, 304 Colombia, 293, 294t6.6, 311 Mexico, 302 gains in, 29 Pemex, Mexico, 352 Mexico, 349, 357­58, 359 pension systems and worker transfers, 29, Chile, 208, 209­10, 211 31f1.11 Mexico, 355 OSIPTEL. see Supervisory Agency performance for Private Investment in Argentina Telecommunications banking sector, 87, 88t2.7, (OSIPTEL), Peru 89t2.8 OSN. see Obras Sanitarias de la changes in, 82, 83t2.6 Nación (OSN) indicators of, 80­86 output, 23, 24f1.7 Bolivia Argentina, 81 data results analyses, 130­36, Bolivia, 134 139­41t3A.2, 142nn11­13 498 INDEX performance (continued) privatization industry-adjusted data, banking sector, 424, 430t8.5, 134­36, 142n13 460­618C.2 Brazil, empirical analysis of, data collection, 15, 422­30, 153­62, 153t4.3, 449­50(Appendix 8A), 184­86t4B.1, 475nn18­19 186­92tt4C.1­4C.4, electricity sector, 437, 194nn10­11, 194nn13­14 458­59t8C.1, Colombia 470­71t8E.2, indicators for IFI firms, 332, 472­73t8E.3 345n35 financial sector, 438­44, 447 manufacturing sector, historical perspective, 292­305, 344n17 409­18, 474nn2­3, power sector, 305­13, 474nn6­11 314t6.11, 345n25, 345n27 impact on employment, thermal plants, 311, 312f6.5, 444­47, 448, 475nn22­24 313­15, 340­42t6E.1, impact on utilities sector, 447 345n27, 345n29 methodology of, 418­22, variables for IFI firms in 474n13, 475n14, sample, 333­36t6C.1 475nn16­17 Mexico overview, 431­34, 475n20 changes in, 356t7.4 revenues and investments, industry-adjusted data, 313f8.1, 411, 412t8.1, 358­59, 360t7.5, 401n13 413 and market power, 359, revenues by sector, 413, 361­63, 401n14 415f8.2, 416f8.3 raw data concerning, statistics on privatized firms, 355­58, 401nn9­12 430, 463­67t8D.1 Peru, 448 telecommunications sector, banking sector, 442­43t8.9 437, 445, 446t8.11, data concerning, 424, 468­69t8E.1 428­29f8.5, 451­57t8B.1 Petrobras, Brazil, 172 privatization study overview, petroleum products, 126 428­29f8.5, 431­34, physical assets per employee ratio, 475n20 Chile, 218 study methodology, 418­22, Pinheiro, Armando C., 148­49, 474n13, 475n14, 194n7, 194­95n16 475nn16­17 Pinochet, Augusto, 197 social view vs. agency view of, Pohl, Gerhard, 120 119­20 policymakers, support for studies on firms in Latin privatization, 1, 58n4 America, 9­14t1.2 Political Constitution, Peru, 411 see also variables politics, 6, 372 Peru Brazil, and privatization, 172, banking sector, 424, 430t8.5, 173 431, 438­44, 460­61t8C.2, and inefficiency of SOEs, 27­28 474n3 and market competition, 53­54 INDEX 499 Mexico and method of privatization, 40 and restructuring policies, Mexico, 357­58, 401n12 376­77, 402n23 in competitive vs. and types of privatization noncompetitive industries, contracts, 378­80 359, 361­62t7.6, 363 and privatization contracts, 49, paid for privatized firms, 51, 378­80 372, 402n18 and SOEs, 3 restructuring programs, Pollitt, Michael G., 120 375t7.12, 376­78, Pombo, Carlos, 15, 53, 275­348 402nn21­23 popular capitalism, 209­10, 211 and operating income to sales, port transfer rate index, 245 34, 58­59n17 ports private investors, Bolivia, 123­24 Chile, 202, 203, 271n4 private pension funds (AFPs), Chile, concessions for management 208, 209­10, 211 and operation of, 242­46, privatization 272nn31­32 benefits of, 2 regulatory framework, costs of, 162­69 244­45 empirical studies on, 120­21 Mexico, 366­67t7.8 failures of, 2 postal services, Argentina, 108 historical perspective, 2­6, Potable Water and Sewerage Law, 58n6, 58n8 Bolivia, 127 impact of length of on prices power companies, Bolivia, 124, paid for firms, 41 125­26 methods of, 39­41, 59nn20­21 power sector overview, 407 Colombia, 328 restructuring firms before, performance, 305­13, 41­49, 50t1.6, 59nn22­23 314t6.11, 345n25, support for, 1, 57­58n3 345n27 see also privatization under see also thermal plants names of specific countries PPI. see producer price index (PPI) Privatization Law of 1992, Bolivia, PQ. see privatization Q (PQ) 123 present-value of revenue (PVR), privatization Q (PQ), 372­74, 377, 241­42 402n18 prices definition, 395t7A.1 Argentina, 85­86 impact of firm and industry Brazil, 165­67 characteristics on, 374, Chile 375t7.12 energy sector, 227, 229t5.12 producer price index (PPI), 401n9, telecommunications sector, 408 233­34, 272n24 productive efficiency, Colombia, 320 Colombia, electricity sector, productivity 287­88 Argentina, 84, 104 impact of auction requirements Chile on, 374, 375t7.12, 376, electricity sector, 227, 402n20 228t5.10, 229t5.11 500 INDEX productivity (continued) changes in, 356t7.4, 357, impact of privatization on, 359, 360t7.5 219, 254f5A.2, in competitive vs. 266­67t5A.8 noncompetitive industries, Colombia 359, 361­62t7.6, 363 manufacturing sector, and role of transfers from 294­95t6.6, 296t6.7, workers, 363­64, 365t7.7, 299­303, 344n22 402n15 profitability Peru, 429, 433t8.6 and abuse of market power, post-privatization, 19, 20f1.4 32­36, 58­59n17 transfers from workers as Argentina, 103­4 percent of, 29, 31f1.11 banking sector, 86­90, Promigas, Colombia, 286 113n19 Promotion of Private Investment in changes in, 82, 82t2.5 State Companies, 410 as measured by performance property theory, 119 indicators, 80 public corporations, Chile, 200 Bolivia, 13, 131­32t3.3 public debt, Brazil, 169 industry-adjusted data, public disclosure requirements, 134­36, 142n13 213, 271n11 Brazil, 154, 157, 159­60t4.4, Public Enterprises of Medellin 161, 162, 184­86t4B.1, (EPM), Colombia, 306, 186­92tt4C.1­4C.4 307­8t6.9, 311 Chile public health, Argentina, 90­96, energy sector, 227, 230, 106­7, 113n17 231t5.13 public opinion impact of privatization on, Brazil, 170­71, 174, 216­17, 249­50t5A.1, 195nn20­21 258­59t5A.4 Peru, 413, 417f8.4 telecommunications sector, public services, 58n8 235­36 Bolivia, 124 utilities sector, 201­2, 224 Brazil, 145 Colombia Chile, 200 impact of privatization on, Mexico, 364, 366­67t7.8 293, 294­95t6.6, 295, Peru, 410 296­97t6.7, 298 public utility firms, Peru, 447 manufacturing sector, PVR. see present-value of revenue 315­23, 345n31, (PVR) 345n33 power sector, 305­13, R 314t6.11, 345n25, railway sector, 37 345n27 Argentina, 107, 112nn3­4 and consumer welfare, 37 Chile, 211­12 and labor costs, 29, 31f1.11 Mexico, 366­67t7.8 Mexico RAIS. see Annual Survey of Social banking sector, 382­83, 385, Data (RAIS) 402­3n25 Ramamurti, Ravi, 36­37, 120 INDEX 501 Ramírez, Manuel, 15, 53, 275­348 ex ante, 272n32 rate floor index, 245 Reordenamiento de las Empresas real average wages, Argentina, 81, Públicas, 121 84­85 Reorganization Unit, Bolivia, 129 real growth in sales, Mexico, 357 Requena, Mario, 121 real interest rates, Chile, 209 reregulation, 52­55, 59n24 redistribution hypothesis, 363, 364 Mexico, 381­88, 392­93, reduced-agency-cost hypothesis, 402­3nn25­26 382­83, 384­85 research reform, Bolivia, 125 adjusted ratios, 23, 24­26 Regulatory Commission for Energy noncomparable data, 17­18, and Gas, Colombia, 286­88 58n15 regulatory framework, 52, 408­9 raw data, 19­23 Bolivia, 125 sample selection bias, 6, 8­17 hydrocarbons industry, 126 see also privatization under telecommunications sector, names of countries 127 restructuring programs, 408 Chile, 214, 247­48 Brazil, 168 banking sector, 208­9 Colombia electricity sector, 222­25, manufacturing sector, 299, 226­30, 230, 231t5.13, 302 272nn20­22 power sector, 286­88, infrastructure concessions, 289­90t6.4, 344n13 240 methods of, 372, 373t7.11 ports, 244­45 Mexico, 350, 372­78, 392, telecommunications sector, 402nn18­23 222­23, 225­26, 230­37, cost of, 367­68 272nn23­24 impact on efficiency of public water services, 212­13 sector, 354­55 Colombia, 275, 276, 297, 306, Peru, 423 328, 343nn1­2 preprivatization, 41­49, 50t1.6, electricity sector, 286­88, 59nn22­23 344n13 and privatization contracts, 49 gas company sales, 284, 286 return on assets (ROA) thermal plants, 313, 315, Argentina, 87, 88t2.7, 89t2.8 340t6E.1 Brazil, 154, 161 gains from, 54 Peru, 428f8.5 Mexico, 352 return on equity (ROE) deregulation, 380­81 Argentina, 87, 88t2.7, 89t2.8 reregulation, 381­88, Brazil, 154, 161 392­93, 402­3nn25­26 Peru, 428f8.5 stock market, 389­90 revenues Peru, 410, 411 Argentina, 71, 72t2.1, 73f2.1 see also corporate governance Brazil, 147 rents Chile, 207, 210t5.5, 211 Argentina, 96, 97, 102f2.3, 103, in developing countries, 5, 114n24 5t1.1 502 INDEX revenues (continued) São Paulo State Privatization government, 27­28 Program, 147, 180­81t4A.2 in Latin America, 5, 5t1.1, 6, Scarpa, Carlo, 40 7f1.2 Schargrodsky, Ernesto, 15, 67­116 Mexico, 367 Schleifer, Andrei, 432 Peru, 411, 412t8.1, 413, Schroth, Enrique, 37 415f8.2, 416f8.3 seaport franchises, Chile, 248 Riaño, Aljandro, 39 SEDAPAL, Peru, 430, 431, 434, ROA. see return on assets (ROA) 475n20 ROE. see return on equity (ROE) SEGBA. see Servicios Electricos del Rogozinski, Jacques, 379 Gran Buenos Aires (SEGBA) Romero, Carlos, 54 Sendos. see Servicio Nacional de Obras Sanitarias (Sendos), S Chile Sáez, Raul, 16 Serra, Pablo, 15, 37, 53, 197­274 sales, Mexico, 357, 359, Servicio Nacional de Obras 401nn12­13 Sanitarias (Sendos), Chile, sales-to-assets ratio, 20, 21f1.5 212 sales-to-capital-stock ratio, Servicios Electricos del Gran Colombia, 299 Buenos Aires (SEGBA), 106 sales-to-employment ratio sewerage services. see sanitation Argentina, 81 services Chile, 214, 219 Shapiro-Francia test for normality, Colombia, 299 420 Mexico, 20, 21f1.5, 357 shared risk contracts, 122 sales-to-input ratio, Colombia, 299 shareholders sales-to-physical assets (PPE) ratio Brazil, 155­56, 168, 186t4B.1, Bolivia, 130 186­92tt4C.1­4C.4 Brazil, 154­55, 161­62, Mexico, 389­91 186­87t4C.1 Sheshinski, Eytan, 3 Chile, 218 shipping sector Colombia, 306 Argentina, 108 Mexico, 357 Chile, 202 Salinas administration, 354­55, see also ports 380­81, 389, 390f7.9 Shleifer, Andrei, 49 San Antonio, Chile, 243, 272n31 Singal, Vijay, 120 Sanchez de Lozada, Gonzalo, 122, Sistema de Participacion 142n3 Ciudadana, Peru, 416 Sánchez, Jose Miguel, 40, 54 Sistema de Regulación Sectorial sanitation services (SIRESE) Law, Bolivia, 125 Argentina, 68, 70, 90­96, Smith, J., 420­21 106­7 social rights, 194n15 Bolivia, 128 social Security of Government Brazil, 164t4.5, 165 Employees, Mexico, 352 Chile, 212, 213 social services, Mexico, 371­72 Mexico, 366­67t7.8 Sociedad Mixta Siderurgica Peru, 417, 430, 431 Argentina (SOMISA), 108 INDEX 503 SOEs. see state-owned enterprises wages in SOEs vs. private (SOEs) firms, 148, 194n6 SOMISA. see Sociedad Mixta Chile, 198, 216­20, Siderurgica Argentina 249­50t5A.1 (SOMISA) crisis of early 1980s, 207­10, Special Privatization Committees 271nn8­9 (CEPRIs), Peru, 410, 411, data on privatized firms, 474n6 213­16, 271nn11­14, Spiller, Pablo, 343n2 271­72nn16­17 state-owned enterprises (SOEs), 1, first round of privatization 21, 22f11.6, 29, 407 of, 206­7 and adjusted ratios, 23, 24­36 privatization of, 199­200, Argentina 210­13, 271n2, 271n10 impact of privatization on, state intervention in, 203­6, 74­86, 81­82, 271nn6­7 112­13nn6­18 utilities sector, 201­3, 271n3 nonfinancial, 74, 74­75t2.2, Colombia, profitability of, 19 77­78f2.3 and consumer welfare, 36­39, revenues from privatization 59nn18­19 of, 71, 72t2.1, 73f2.1 employment, 21, 22f1.6, 29 bias in selection samples, 6 evidence of failure of, 3 Bolivia, 122, 124 and government revenues, Brazil 27­28 benefits and costs of historical perspective, 58n6, privatization of, 58n8, 206 162­69 insolvency of, 156 empirical analysis of data, and labor force restructuring, 43 153­62, 153t4.3, Mexico, 298 184­86t4B.1, fiscal impact of sale of, 186­92f4C.4, 367­70, 371f7.3 194nn10­11, growth of, 351­53, 194nn13­14 400nn1­4 impact of privatization on, industry-adjusted data, 163­65, 166f4.1, 358­59, 360t7.5, 194­95nn15­16 401n13 literature review, 148­49, privatization program, 194nn6­7 353­55, 401nn5­6, owned by government but 401n8 not privatized, 147­48, restructuring programs, 182­84t4A.3 375t7.12, 376­78, privatization at federal level, 402nn21­23 147­48, 175­79t4A.1, noncomparable data concerning, 180­81t4A.2 17­18, 58n15 public opinion concerning, operating efficiency changes in, 170 20, 21f1.5 study data set and variables, performance of, 120 149­53, 194nn8­9 Peru 504 INDEX state-owned enterprises (SOEs) Bolivia, 126­27 (continued) Brazil, 146­47, 194­95n16 data collection, 422­30, impact of privatization on 449­50(Appendix 8A), prices of, 167 475nn18­19 public opinion concerning, employment, 444­47, 170 475nn22­24 Chile, 198, 201 privatization of, 411, 413 privatization process, 221­22 privatization study overview, regulatory framework, 431­34, 475n20 222­23, 225­26, 230­37, profitability of, 24­26 272nn23­24 see also enterprises impact of privatization on, stock markets 36­37 Brazil, 168 Mexico, 366­67t7.8 and corporate governance, 56 mobile phones, 232, 235, Mexico, 355, 389, 401n8 272n23 structural reforms, Peru, 411 Peru, 413­16, 416f8.3, 431, Sturzenegger, Federico, 67­116 437, 468­69t8E.1 Suarez, Hugo Banzer, 243n3 employment in, 445, subsidies 446t8.11 Chile, 238 preprivatization, 54 Mexico, 368, 368t7.9, 369 and regulation, 53 sunk costs, 345n31 Telefónica-CTC, Chile, 234, Supervisory Agency for Private 235­37 Investment in Telefonica de Argentina, 105 Telecommunications Telefónica de España, Spain, 414, (OSIPTEL), Peru, 414 474n9 supervisory system, Mexico, 385 Telefónica del Perú, 424 Telefónica del Perú S.A. (TdP), T 414, 415­16, 437, 445, taxation 468­69t8E.1, 474n10 Mexico, 357 terminals, Chile, 243­44 net-taxes-to-sales ratio, 156­57 TERPEL companies, Colombia, revenues from, 28 286, 343n12 TdP. see Telefónica del Perú S.A. TFP. see total factor productivity (TdP) (TFP) technical efficiency, thermal plants, thermal generation sector, 288 313 thermal plants technology, manufacturing sector, Colombia, 311, 312f6.5, 345n27 316, 317­19t6.12 input and output variables, Telcoy, Chile, 221, 236 339t6D.2 Telecel, Bolivia, 127 productive efficiency of, Telecom, Argentina, 105 313­15, 340­42t6E.1, Telecommunications Law, Bolivia, 345n29 127 statistical analysis of telecommunications sector efficiency scores, 324­27, Argentina, 85, 105­6 345n34 INDEX 505 Tobin's Q, 372, 402n18 Peru, 430, 445, 446t8.11 Tobit regressions, 324­27, 345n34 study results, 434­48 Todd, P., 421 see also specific utility toll roads, Mexico, 366­67t7.8, 379 V Torero, Máximo, 15, 37, 407­77 Valparaíso Port, Chile, total factor productivity (TFP), 245­46 299­300, 315 value added growth, 345n33 Tovar, Ramiro, 379 value added to capital, 299 trade, 295, 301, 316 van Randenborgh, Matthias, 120 traffic guarantees, Chile, 241, 242, variables 248 Argentina transfers from workers, Colombia, banking sector, 87, 88t2.7, 305t6.8 89t2.8 transition countries, move to used to evaluate impact of privatization programs, 5 privatization on financial transparency, 39, 40 firms, 1112B.2 Chile, 200, 213, 271n11 used to evaluate impact of transportation sector privatization on Chile, 211 nonfinancial firms, Peru, 411 109­10t2B.1 Tribasa, Chile, 248 Bolivia, used to evaluation Trujillo, Lourdes, 18n59, 56 impact of privatization, Turkey, 59n21 139­41t3A.2 Brazil U empirical analysis of data, unemployment rate, Argentina, 98, 157, 162, 184­86t4B.1, 114n29 192­93t4D.1 unit costs, 19, 20f1.5 and study data set, 149­53, Argentina, 81, 84, 104 194nn8­9 Bolivia, 130, 139­41t3A.2 Chile Chile, 218, 251f5A.1, 261t5A.5 merged firms, 215 Colombia, 294­95t6.6, 296t6.7, profitability, 216­17, 299, 306 249­505A.1, competitive vs. noncompetitive 258­59t5A.4 sectors, 34, 36 Colombia gap between private and electricity sector, 313, privatized firms, 26­27 314t6.11 Mexico, 357, 358, 359 indicators for IFI firms in United Kingdom, 57­58n3, 272n29 sample, 333­36t6C.1 unskilled workers, 32 manufacturing sector, Urbiztondo, s., 92 291­92, 315­23, 336, Urquiola, Miguel, 121 337t6D.1, 344n15, utilities sector 345n31, 345n33 Chile, 201­2, 271n3 thermal plants input and and consumer welfare, 36­39, output variables, 59nn18­19 339t6D.2 506 INDEX variables (continued) Mexico, 366­67t7.8 Mexico Peru, 417, 431 changes in performance, 359, welfare-enhancing outcomes, 28, 360t7.5 36­39, 59nn18­19 definitions of, 394­400t7A.1 Argentina, 68, 70, 95, 96t2.11 description overview, 393 Mexico, 364, 366­67t7.8 performance changes, 356t7.4 Peru, 448 Peru, 432, 433t8.6 white-collar workers difference-in-difference, Bolivia, 133 435­36t8.7, 437 Brazil, 194­95n16 of performance, 418­22, Colombia, wages, 304 474n13, 475n14, layoffs, 32 475nn16­17 Mexico, 357, 364 Verdoom law, 320 Peru, 444­45 Vergara, Rodrigo, 198 Wilcoxon signed-rank test, 165, Vishny, Robert W., 49 420 voucher programs, Mexico, 370­71 Winston, Clifford, 52 within-country data, 40 W WLL, Chile, 237 wages workers Argentina, impact of Argentina, displaced workers, privatization on, 84­85, 96­103, 114nn24­29 98­105, 114n25, 114n30 displaced, 29­32 Bolivia, 133, 140t3A.2 labor costs, 28­29 Brazil, private firms vs. SOEs, transfers from as percent of 148, 194n6 profitability, 29, 31f1.11 Colombia, manufacturing sector, see also blue-collar workers; 304, 305t6.8 employment; white-collar Mexico, 363­64, 402n15 workers and privatization, 29 Worker's Party, Brazil, 173 real and industry-adjusted Workers' Tenure Guarantee Fund changes in, 29, 30f1.10, 32 (FGTS), Brazil, 167­68, 170, see also income 174, 195n22 Wallsten, Scott, 53 World Bank, 122 water services, 37­38, 49 Argentina, 68, 70, 106­7 Y and child mortality, 95, Yacimientos Petrolíferos Fiscales 113n17 Bolivianos (YPFB), 124 impact of privatization on, 90­96 Z Bolivia, 127­28 Zamora, Jaime Paz, 142n3 Brazil, 164t4.5, 165 Zervos, S., 168 Chile, 212, 213 Zuleta, L. H., 276 LATIN AMERICAN DEVELOPMENT FORUM rivatization is under attack. Since the 1980s, thousands of failing state-owned P enterprises worldwide have been turned over to the private sector. But public opinion has turned against privatization. A large political backlash has been brewing for some time, infused by accusations of corruption, abuse of market power, and neglect of the poor. What is the real record of privatization and are the criticisms justified? Privatization in Latin America: Myths and Reality evaluates the empirical evidence on privatization in a region that has witnessed an extensive decline in the state's share of production over the past 20 years. The book is a compilation of recent studies that provide detailed coverage within a systematic econometric framework. Seven countries are investigated: Argentina, Bolivia, Brazil, Chile, Colombia, Mexico, and Peru. The quality of the data enables the researchers to comprehensively analyze the record of and accusations against privatization, with important recommendations for the future. This book will be vital to anyone interested in the privatization debate but especially to those involved in civil service reform, corporate governance, economic policy, finance, and anticorruption efforts. "Privatization is important but controversial. While economists typically favor it, others are skeptical. This book provides strong scientific evidence that privatization has been beneficial for many Latin American countries, although some privatizations failed and some groups in society lost out. As usual, the devil is in the details: how privatization is carried out and what reforms accompany it are crucial to its success. The book is definitely an invaluable contribution to the privatization debate." --Oliver Hart, Andrew E. Furer Professor of Economics, Harvard University "There are few issues in economics that elicit such passionate opinion as privatization in Latin America. Typically, privatization is berated because the 1990s failed to deliver the kind of growth and prosperity expected at the start of the decade, without taking into account the serious disruption in financial markets following the 1998 Russian crisis. This book, in contrast, addresses the issue by bringing to the table new, relevant, and rich information. It provides powerful evidence of the benefits of privatization in the region--benefits that largely offset any potential costs. Further, the evidence presents a compelling case that privatization failure is more likely due to poor contract design, lack of regulation, or opaque processes than to privatization per se. The book should be required reading for policymakers and academics alike." --Guillermo A. Calvo, Distinguished University Professor, University of Maryland, College Park, and Chief Economist, Inter-American Development Bank STANFORD UNIVERSITY PRESS THE WORLD BANK ISBN 0-8213-5882-0