34433 Keeping the Promise of Social Security IN LATIN AMERICA Indermit S.Gill Truman Packard Juan Yermo THE WORLD BANK STANFORD UNIVERSITY PRESS Keeping the Promise of Social Security in Latin America Keeping the Promise of Social Security in Latin America Indermit S. Gill, Truman Packard, and Juan Yermo With the assistance of Todd Pugatch 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. First printing October 2004 Revised June 2005 1 2 3 4 07 06 05 04 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, Calif. 94304 Washington, DC 20433 The findings, interpretations, and conclusions expressed here are those of the au- thors 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 cannot 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 on the part of the World Bank any judgment of the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is copyrighted. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photo- copying, recording, or inclusion in any information storage and retrieval system, with- out the prior written permission of the World Bank. The World Bank encourages dis- semination of its work and will normally grant permission promptly. For permission to photocopy or reprint, please send a request with complete infor- mation 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 ad- dressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, fax 202-522-2422, e-mail pubrights@worldbank.org. ISBN 0-8213-5817-0 (World Rights except North America) ISBN 0-8047-5181-1 (hardcover) (North America) ISBN 0-8047-5238-9 (softcover) (North America) Library of Congress Cataloging-in-Publication Data Gill, Indermit Singh, 1961- Keeping the promise of social security in Latin America / by Indermit S. Gill, Truman Packard, and Juan Yermo. p. cm. -- (Latin American development forum) Includes bibliographical references (p. ) and index. ISBN 0-8213-5817-0 1. Social security--Latin America. I. Packard, Truman. II. Yermo, Juan, 1972- III. Title. IV. Series. HD7130.5.G55 20004 368.4'0098--dc22 2004050929 Latin American Development Forum Series This series was created in 2003 to promote, debate, and disseminate information and analysis, and convey the excitement and complexity of the most topical issues in economic and social development in Latin Amer- ica 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 publica- tion represent the highest quality in each institution's research and activ- ity output, 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 Re- gion, World Bank Steven Haber, A. A. and Jeanne Welch Milligan Professor, Department of Political Science, Stanford University; Peter and Helen Bing Senior Fel- low, 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 Indermit S. Gill is an economic adviser in the Poverty Reduction and Economic Management Network at the World Bank. He received a Ph.D. in economics from the University of Chicago, and has extensive experience in the areas of labor policy and social insurance in Latin America, includ- ing service as lead economist for the World Bank's work on pension re- form in Brazil. Before joining the Bank staff in 1993, he taught economics at the State University of New York at Buffalo. Truman Packard is a senior economist in the Latin America and the Caribbean department of the World Bank. He began work on pensions in 1995, when he worked with Bank staff in Mexico to develop adjustment operations needed to support pension reform. Since then he has partici- pated in analytical and lending assistance to support both pension reform and broader efforts to strengthen social insurance. His research focuses on how pensions and social insurance affect the efficiency of labor markets and incentives to save. He holds a Ph.D. in economics from the University of Oxford. Juan Yermo is an administrator at the Organisation for Economic Co-Operation and Development in Paris, where he directs private pen- sions projects in the financial affairs division. He is currently secretary of the OECD Working Party on Private Pensions, which hosts biannual meetings where policymakers and representatives from the private sectors of 30 member countries survey pensions and advise members on policy is- sues. He manages a broad research program on the operation, regulation, and supervision of pension plans and funds, and has consulted for the World Bank Group on pension reform in Latin America. He holds mas- ter's degrees in economics from Cambridge and Oxford universities. Todd Pugatch is a junior professional associate in the Poverty Reduction and Economic Management Network at the World Bank. Prior to joining the Bank he studied economics as a Morehead Scholar at the University of North Carolina at Chapel Hill, and as a Rotary Ambassadorial Scholar at the Universidad Nacional de Costa Rica. In 2005 he will begin a doctoral program in economics. Contents Foreword xvii Acknowledgments xix 1 RETHINKING SOCIAL SECURITY PRIORITIES IN LATIN AMERICA 1 PART I. RETROSPECTIVE: FISCAL, FINANCIAL, AND SOCIAL BENEFITS FROM PENSION REFORM 17 2 STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 19 3 THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES IN LATIN AMERICA 39 4 THE FINANCIAL BENEFITS OF PENSION REFORM 57 5 SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 89 PART II. ANALYTICAL: HOW GOVERNMENTS CAN HELP INDIVIDUALS DEAL WITH RETIREMENT 107 6 HOW INDIVIDUALS VIEW SOCIAL SECURITY 109 7 HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED FROM THE INDIVIDUAL'S PERSPECTIVE? 125 8 THE PREFERENCES THAT INDIVIDUALS REVEAL 165 ix x CONTENTS PART III. PROSPECTIVE: THE FUTURE OF SOCIAL SECURITY IN LATIN AMERICA 197 9 PREVENTING POVERTY IN OLD AGE: IMPROVING THE POOLING COMPONENT 199 10 FACILITATING CONSUMPTION SMOOTHING: IMPROVING THE SAVINGS COMPONENT 227 11 THE WAY FORWARD 267 Technical Annex. Assumptions of PROST Simulations 283 Appendix. Background Papers for Keeping the Promise 291 Bibliography 293 Index 321 BOXES 1.1 Three Central Dilemmas of Pension Privatization 3 3.1 Bolivia's Pension Reform: A Transition Considerably More Costly Than Expected 45 3.2 Virtuous and Vicious Fiscal Circles: The Cases of Chile and Argentina 47 4.1 The Importance of Concomitant Reforms in the Financial System 64 4.2 Inflation-Indexed Securities in Latin America 74 5.1 Defining the Coverage Problem: What Is It and Why Do We Care? 95 5.2 Was Increasing Coverage an Objective of Pension Reform? 97 6.1 The Welfare State as a Piggy Bank to Reduce Old-Age Income Insecurity 110 6.2 A Theory of "Comprehensive Insurance" 113 7.1 The Four Fundamental Risks in Retirement Income Security Systems 127 7.2 Comparing Notional Defined Contribution Systems and Pension Funds Invested in Inflation- and Wage-Indexed Government Bonds 128 7.3 Argentina's System in Crisis: Do Private Accounts Protect Workers from Policy Risk? 133 7.4 Points to Keep in Mind When Comparing Risks in Pension Fund Portfolios in Latin America 137 CONTENTS xi 8.1 PRIESO: Social Risk Management Surveys in Chile and Peru 176 8.2 Peru's Reformed Pension System: Multipillar in Name Only 179 9.1 The Cuota Social: Preventing Poverty among Elderly Men and Women in Mexico 208 9.2 BONOSOL: Bolivia's Universal Pension Program 210 10.1 The Big-Bang Approach to Voluntary Pension Savings Reform in Chile 248 10.2 The Role of the Financial System in Brazil's Voluntary Pension Savings Plans 250 11.1 The Role of the Second Pillar 278 FIGURES 1.1 Pension Systems Cover between 10 and 60 Percent of the Economically Active Population in Latin American Countries 6 1.2 Between One-Tenth and Two-Thirds of the Aged Populations Receive Pensions in Latin American Countries 7 2.1 Rising Life Expectancy Increases the Share of Elderly People in the Population and Upsets the Balance of Pure PAYG Pension Systems 20 2.2 Destination of Mandatory Pension Contributions 29 3.1 Simulated Implicit Pension Debts (IPDs) with and without Structural Reforms 40 3.2 Total Pension Debt (Explicit) Accumulated after 2001, with and without Structural Reforms 43 3B.1 Cash Flow Gap in Bolivia 45 3B.2a Chile Was Fiscally Strong prior to Reform; Argentina Was Not 47 3B.2b Pension System Deficits Contributed Significantly to Deteriorating Fiscal Balance in Argentina 48 3.3 There Is No Indication That Pension Reform Increased Mexico's Country Risk 52 4.1 Pension Funds Are Major Investors in Government Debt 63 4.2 Interest Rate Spreads Have Declined in Peru and Bolivia Since the Reforms (1993­2002) 79 4.3 Stock Market Turnover Ratios in Selected Latin American Countries (1990­2001) 81 5.1 Structural Reforms Are Likely to Improve Equity by Lowering Regressive Transfers and Returns 91 xii CONTENTS 5.2 Structural Reforms Make the New Systems More Gender Neutral, but Women's Average Benefits Can Be Significantly Lowered by the Use of Gender-Specific Mortality Tables 92 5.3 Pension Income Increases Inequity Relative to Earned Income from Labor 94 5.4 Has Participation Increased? There Is No Clear Pattern in Data on Contribution to National Pension Systems 100 7.1 Pension Funds Invest Mainly in Government Bonds and Instruments Issued by Financial Institutions 132 7.2 The Standard Trade-Off between Risk and Return Has Not Materialized in Latin America (Returns from Inception to December 2000) 136 7.3 In Chile Intercohort Differences in Returns Have Been Large 139 7.4 Returns on Deposits in Chile Have Been Lower but More Stable Than Pension Fund Returns 140 7.5 Brady Bonds Would Have Been a Better Investment Than Domestic Equities for Peruvian Pension Funds 141 7.6 Variable Commission Rates Have Risen as a Result of Declines in Contribution Rates in Argentina and Peru 147 7.7 Half of the Pension Contributions of the Average Chilean Worker Who Retired in 2000 Went to Fees 148 7.8 Participation in the Second Pillar Is Costlier for Poorer Workers in Chile 149 7.9 Annuities Have Yielded Varying Levels of Retirement Benefits in Chile 154 7.10 Chile's Pension Funds Did Not Do a Good Job of Educating Participants: Fondo 2 Had Few Takers Despite Earning Higher Returns 156 7.11 Chilean Workers' Choices in the New Multifund System Largely Correspond with the Default Options 158 7.12 Choices Are Similar for Different Income Groups in Chile If Controlling for Age 159 8.1 Chile: Reported Contribution Density 180 8.2 There Is No Difference in Risk Preferences between Employees and Self-Employed Workers Who Responded to the PRIESO Survey in Chile 182 8.3 Self-Employed Workers Who Contribute to the AFP System in Chile Have a Greater Tolerance for Risk 183 CONTENTS xiii 8.4 Peru: Reported Contribution Density 184 8.5 There Is No Difference in Risk Preferences between Employees and Self-Employed Workers in Peru 187 8.6 Self-Employed People in the AFP System in Peru Are More Risk Averse 188 9.1 Rising Life Expectancy Increases the Share of Elderly People in the Population and Upsets the Balance of Pure Pooling Pension Systems 200 9.2 Accumulated Wealth Increases with Age and Is Greatest among Old People 204 9.3 Relative Generosity and Cost of Alternative Public Poverty Pension Arrangements in Selected Countries 212 9.4 Average Noncontributory Pensions Are between 30 Percent and 60 Percent of Average Contributory Pensions 217 10.1 In Peru, Fees Remain Persistently High, Despite Increasing Returns and Declining Administrative Costs 234 10.2 Savings Products Offered by Insurers Are More Popular Than the AFPs' Liquid Cuenta 2 249 10.3 Mutual Funds Have Grown Significantly Only in Brazil, Where Pension Funds Are Voluntary 252 10.4 Fees for Equity Mutual Funds in Chile Have Remained Stubbornly High 255 10.5 Economies of Scale in Fund Management Kick in at Low Asset Levels 256 10.6 Insurance Company Bankruptcy Is a Threat to Policyholders 258 11.1 Coverage Can Vary at Similar Stages of Economic Development 274 TABLES 1.1 Instruments of Old-Age Income Security 10 2.1 Principal Features of Structural Reforms to Social Security Systems (Old-Age Disability and Death) in Latin America during the 1980s and 1990s 23 2.2 Principal Features of Structural Reforms to Social Security Systems (Old-Age Disability and Death) in Latin America during the 1990s and 2000s 26 2.3 Contribution Rates and Earnings Ceilings in the Mandatory Funded Systems 32 2.4 Voluntary Funded Pillars in Latin America 35 xiv CONTENTS 3.1 Current Pension Deficits (Benefit Expenditures Less Contribution Revenue) Financed by Government Transfers 42 4.1 Portfolio Ceilings by Main Asset Classes (December 2002) 61 4.2 Portfolio Floors by Country 62 4.3 Assets Held by Pension Funds Have Doubled as a Percentage of GDP (December 1998­December 2002) 62 4.4 Pension Funds Invest Mainly in Debt of Governments and Financial Institutions (December 2002) 67 4.5 For the Average Latin American Country, the Two Largest Institutions Control Two-Thirds of the Pension Funds (December 2002) 68 5.1 There Is Evidence That Pension Reforms Positively Affect Incentives 102 5.2 Participation Rates and Incidence of Benefits Show a Regressive Pattern of Coverage by Formal Pension Systems 104 6.1 Justifying a Government-Imposed Mandate to Save for Old Age--Microeconomic Reasons 117 6.2 Justifying a Government-Imposed Mandate to Save for Old-Age--Macroeconomic Reasons 120 7.1 Gross, Real Returns to Pension Funds Have Been High 130 7.2 Asset Allocation and Portfolio Limits 142 7.3 Workers Still Pay High Commissions in Some Countries (December 2002) 145 7.4 The Form of Benefits Is Usually Inconsistent with an Assumption of Retiree Irrationality with Respect to Saving 151 7.5 The Five New Funds in Chile Vary by Proportion of Equity Investment 157 8.1 Probability That Workers Contribute to Social Security Is Determined by Individual, Household, and Labor Market Factors 171 8.2 Reformed Pension Systems in Chile and Peru Are Similar, but There Are Important Differences 178 9.1 Poverty among Elderly People Is as Frequent as among Other Age Groups When Measured by Current Income 202 9.2 Expenditure on Noncontributory Pension Programs 215 9.3 Noncontributory, Poverty Prevention Pensions Cover a Significant Portion of Pension Recipients 216 CONTENTS xv 9.4 As a Share of Average Wages, Minimum Wages Are Relatively High in Chile, Colombia, and Some Other Latin American Countries 218 9.5 Cost of Providing and Guaranteeing a Public Pension Equal to the Minimum Wage 220 10.1 Reform Options to Reduce the Operational Expenses of the New Private Second Pillar 235 10.2 Reform Options to Ensure That Savings Are Passed on to Affiliates as Lower Commissions 240 10.3 Reform Options to Improve Investment and Longevity Risk Management 243 TA.1 Assumptions Common to All Country Cases 283 TA.2 Country-Specific Details, Assumptions, and Sources of Data 284 TA.3 Assumed Profiles of Representative Affiliated Men and Women 289 Foreword AT PRESENT, NATIONS AROUND THE WORLD (both large and small, rich and poor) are engaged in debate over how to reform their social security sys- tems and care for the aged. For many countries this debate requires spec- ulation on hypothetical scenarios, but in Latin America a rich body of ex- perience on social security reform has been accumulating for more than a decade (for Chile, more than two decades). Keeping the Promise of Social Security in Latin America takes stock of those reforms, evaluates their suc- cesses and failures, and considers the lessons that can be drawn for the fu- ture of pension policy in the region. The authors draw on a series of back- ground papers and surveys commissioned specifically for this inquiry, as well as existing research conducted by themselves and other pension ex- perts. The report was also enriched by discussions with policymakers, pen- sion fund industry managers, and academics at a regional conference held in Bogotá, Colombia, on June 22­23, 2004. The task of assessing reforms that are still in progress must be under- taken with both caution and humility. But the stakes are high, and it would not be advisable to wait until the reforms have run their course. This study is intended to inform an ongoing debate, not to end it. The authors find that structural reforms undertaken in Latin America in the last two decades mark a major improvement in many dimensions with respect to the earlier pay-as-you-go systems. In particular, reforms reduced fiscal liabilities, contributed to financial sector development, and improved the equity of pension systems, although there is still much to be done in these areas. Most importantly, the shift to individual ac- counts was a major structural improvement to the income-smoothing objective of pension systems for most current contributors. But there have also been significant disappointments, chief among them the failure to extend access to social security to a broader segment of society. More than just an empirical assessment of reforms, this volume is also an attempt to rethink the priorities of social security systems in the region. The authors argue that the main priority of any publicly established pen- sion system should be to prevent poverty in old age. If preventing poverty among old men and women is the most important policy priority, then the extent of coverage must be the most important criterion by which to judge xvii xviii FOREWORD any formal social insurance system. Detractors of social security reform in Latin America have rightly criticized the failure of reforms to increase cov- erage, despite any positive effects reforms may have had on fiscal balance sheets or financial sector development. This report meets the detractors in that arena, by analyzing why coverage has not adequately increased fol- lowing reform and discussing the range of appropriate policy responses. The authors argue, thus, for increased attention to the poverty pre- vention function of social security and a less prominent role for mandated savings. They claim that Latin America has not paid enough attention to pillars "zero" or "one"--those pillars whose main purpose is precisely to avoid the risk of falling into poverty in old age. They also argue that ex- cessive attention has been paid to "pillar two" (mandated individual sav- ings accounts) and not enough to the "third pillar" (voluntary savings). They do not, however, propose a "corner solution" composed of just pil- lars one (or zero) and three in all cases. The authors recognize the need to strike a balance between people's improvidence in planning for old age, on the one hand, and a government mandate to save, on the other--both of which can cause damage. Finding a suitable equilibrium between these tensions is a delicate task and one on which even experts can disagree. The appropriate final balance will depend, among other things, on the degree of development of financial systems, institutional capabilities, and past history of pension systems. The proper design of pillars "zero" and "one" is also difficult, as bad designs may encourage moral hazard and provide additional disincentives to participate in the formal labor market. Furthermore, the transition toward the desired system is not easy. Developing a strong zero or first pillar will be demanding in fiscal terms, requiring many countries to first finalize the "first generation" re- forms in order to reduce the large fiscal liabilities that still exist today in favor of highly privileged middle- and high-income individuals. Mean- while, developing a zero pillar must be weighted against the priorities of other poverty prevention programs. Some of the ideas presented in this book may strike some readers as rad- ical. However, in the debate on pension reform there is no orthodoxy, as reflected in major differences of opinion among leading experts in this area of policy. Despite more than a decade of experience with pension reform in Latin America, although undoubtedly a major step forward, reforms are still works in progress. No magic formula for success exists. We hope that this report will further enrich an already vibrant policy dialogue that is of crucial importance to the future of the region. Guillermo Perry Chief Economist, Latin America and the Caribbean Region World Bank Group July 2004 Acknowledgments THIS BOOK HAS BEEN PREPARED BY A TEAM led jointly by Indermit S. Gill (Senior Economic Adviser to the Poverty Reduction and Economic Man- agement Network Vice President, World Bank) and Truman G. Packard (Senior Economist, Social Protection, World Bank), consisting of Juan Yermo (Economist, Private Pensions and Insurance Unit, Organisation for Economic Co-operation and Development), Todd Pugatch (Junior Professional Associate, Poverty Reduction and Economic Management Network, World Bank), Asta Zviniene (Social Protection Specialist, World Bank), Norbert Fiess (Economist, Office of the Chief Economist, Latin America and the Caribbean Region), Salvador Valdés-Prieto (Con- sultant, Catholic University, Santiago de Chile), Rafael Rofman (Senior Economist, Latin America and the Caribbean Region), Abigail Barr (Consultant, Oxford University), Oliver Azuara (Consultant, World Bank), and Federico Escobar (Consultant, World Bank). This volume was prepared under the overall guidance of Guillermo Perry (Chief Economist, Latin America and the Caribbean Region). Luis Servén (Regional Studies Coordinator) and Xavier Coll and Ana-Maria Arriagada (Sector Directors, Human Development during 2000­02) ensured generous funding. The team also acknowledges the financial sup- port and technical guidance of the World Bank's Research Support Bud- get. Chris Chamberlin (Sector Manager, Social Protection, Latin America and the Caribbean Region) provided encouragement, general guidance, and technical comment in the final stages of preparation, and Augusto de la Torre (Senior Adviser, Office of the Chief Economist, Latin America and the Caribbean Region), Hermann Von Gersdorff (Lead Economist), Anita Schwarz (Senior Human Resources Economist), Robert Holzmann (Sector Director, Social Protection, Human Development Network), Michal Rutkowski (Sector Manager, Social Protection, Eastern Europe and Central Asia Region), Robert Palacios (Senior Pensions Economist), Jean-Jacques Dethier (Lead Economist, Development Economics Research xix xx ACKNOWLEDGMENTS Group), Roberto Rocha (Lead Economist, Financial Sector Operations and Policies), Fernando Montes-Negret (Sector Manager, Finance Cluster, Latin America and the Caribbean Region), David Lindeman (Principal Pensions Analyst, Organisation for Economic Co-operation and Develop- ment), Andras Uthoff (Coordinator, Special Studies Unit, Economic Com- mission for Latin America and the Caribbean, United Nations), Maureen Lewis (Interim Chief Economist, Human Development Network), Marcelo Giugale (Country Director, Bolivia, Peru, Ecuador, and Venezuela), and William Maloney (Lead Economist, Office of the Chief Economist, Latin America and the Caribbean Region) provided valuable feedback and comment on the concept paper and earlier drafts of the book. Neither these individuals nor the World Bank are, however, re- sponsible for its contents. This book synthesizes 16 background papers specially commissioned for the book. Simulation analysis conducted by Asta Zviniene with the World Bank's Pension Reform Options Simulation Toolkit (PROST) is cited throughout the book. The background papers were written by Oliver Azuara, Abigail Barr, Federico Escobar, Norbert Fiess, Truman Packard, Rafael Rofman, Salvador Valdés-Prieto, Juan Yermo, and Asta Zviniene. The background papers are listed in the appendix at the back of this vol- ume and are available online at the Web site of the Office of the Chief Econ- omist, Latin America and the Caribbean Region: www.worldbank.org/ keepingthepromise. Naoko Shinkai, Ricardo Fuentes, Enrique Mezanotte, Elizabeth Dahan, and Mauricio Cifuentes provided invaluable assistance in con- ducting research and analysis. This book has also benefited immensely from the comments of peer re- viewers inside and outside the World Bank. The peer reviewers for this task were Anita Schwarz, Dimitri Vittas (Senior Adviser, Financial Sector Development, World Bank), Dean Baker (Center for Economic and Policy Research), and Olivia Mitchell (University of Pennsylvania­Wharton Business School), who provided detailed comments, and to whom the team owes a debt of gratitude. Again, these individuals are not responsi- ble for the contents and messages of this book. The book was also greatly enriched by the presentations and comments of participants in a regional conference on pension reform, hosted by the World Bank and FEDESARROLLO, held June 22­23, 2004, in Bogotá, Colombia. Of particular value were discussions with Luis Fernando Alarcón (Private Funds Association, Colombia), Mariano Paz Soldán (Pro- futuro AFP, Peru), Bernhard Lotterer (Integra AFP, Peru), Fernando Muñoz-Najar Perea (AFP Association, Peru), Ligia Borrero (Banking Su- perintendency of Colombia), Kristhian García Aranda (Finance and Public Credit Ministry, Nicaragua), Francisco Perez (FEDESARROLLO, Colom- bia), Carmelo Mesa-Lago (University of Pittsburgh), Fabio Bertranou (In- ternational Labour Organization), Guillermo Larrain (Superintendency ACKNOWLEDGMENTS xxi of Pension Fund Administrators, Chile), Augusto Iglesias (PrimAmerica Consultants, Chile), and Amparo Ballivian (World Bank). Finally, the authors would like to thank the social security and pensions supervisory authorities in Argentina, Bolivia, Brazil, Chile, Colombia, El Salvador, Mexico, Peru, and Uruguay for providing much of the data used in this analysis. 1 Rethinking Social Security Priorities in Latin America TWO DECADES AGO CHILE'S GOVERNMENT radically altered its approach to old-age income security. Simply put, it changed the basis of public pen- sions from collective to individual: instead of the widely used system that pooled the risk of being unable to earn while aged, the Chileans adopted a system that relied on mandatory individual savings accounts. The shift was seen by its detractors as a retreat from "solidarity," and by its sup- porters as a move toward greater "personal responsibility." Neither char- acterization is entirely correct, but a debate has raged ever since on the shift's main effect: has the change left Chileans better or worse off? An important event in this debate was the World Bank's 1994 publica- tion of Averting the Old Age Crisis. The report explained that the existing approach to ensuring income support for elderly people was unsustain- able, it neatly characterized the institutions involved using a novel termi- nology, and it prescribed a new doctrine for better addressing the chal- lenges in this difficult area of public policy. Using the terminology suggested in Averting the Old Age Crisis, the new approach has come to be called the "multipillar" model of old-age income security.1 Although this approach has been used by reformers worldwide, it can be safely asserted that no region has taken it more seriously than has Latin America. In addition to Chile, governments in 11 countries--Peru (1992), Colombia (1993), Argentina (1994), Uruguay (1996), Mexico and El Salvador (1997), Bolivia (1998), Costa Rica and Nicaragua (2000), Ecuador (2001), and the Dominican Republic (2003)--representing about half of all Latin Americans, have adopted or are in the process of adopting various forms of the multipillar model as suggested by the World Bank. These changes have been seen by policymakers as necessary, but many of their citizens see in them a relinquishing of responsibility by government. The debate rages on: Will these changes make Argentines and Mexicans and Colombians and the citizens in these other countries better or worse off? 1 2 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA The Benefits of Hindsight In this book, using both the experience of these countries and simple an- alytical principles, we try to shed light on this question. Thus, as in Averting the Old Age Crisis, we analyze public policy toward pensions over the last two decades, but especially since the early 1990s. But there are differences between that volume and this one, principally because of the developments in the last decade. We have benefited from advances in thinking that Averting the Old Age Crisis substantially stimulated. What is perhaps more important is that we have the benefit of more experience so we can replace informed conjecture based on the reforms in one coun- try (Chile) with empirical evaluation of the reform experience of more than two decades in Chile; about a decade's worth of experience in Colombia, Peru, and Argentina; and somewhat shorter but still informa- tive experience with reforms in several other countries, especially Bolivia and Mexico. Latin America is not alone in its experience with structural pension re- form. Eight countries in Eastern Europe also have undertaken multipillar re- forms in their transition to market economies. Although these countries also offer important lessons, their institutional and demographic context (adapt- ing formerly socialist systems to meet the needs of a population with an older age profile) is sufficiently different from that of Latin America to pres- ent a distinct set of issues that lie beyond the scope of this book. Further- more, a much larger set of countries, in Latin America and elsewhere, have engaged in "parameter tinkering"--adjusting the size and scope of their sin- gle-pillar social security systems. Our purpose here, however, is to focus on multipillar pension reform in Latin America and present policy implications appropriate for the region, not to offer a global study. This book is based both on specially commissioned background pa- pers and on other work done at the World Bank and elsewhere. Some of the background papers address general questions such as the need to mandate participation in pension schemes (e.g., Packard 2002 and Valdés-Prieto 2002b). Some papers focus on more specific issues such the fiscal, labor market, and capital market effects of social security reforms (e.g., Fiess 2003; Zviniene and Packard 2002; Packard, Shinkai, and Fuentes 2002; Packard 2001, 2002, Yermo 2002a) and more country- specific experiences (e.g., Rofman 2002 for Argentina; Escobar 2003 for Bolivia; Valdés-Prieto 2002c and Yermo 2002c for Chile; Azuara 2003 for Mexico). Other papers assess how workers fared under the new pen- sion system (Yermo 2002b) and their reactions to the reformed systems using data collected in purpose-built household surveys (e.g., Barr and Packard 2002 and Packard 2002 for Chile; Barr and Packard 2003, for Peru). We also take advantage of efforts at the World Bank and elsewhere to collect quantitative information and to refine actuarial techniques, again RETHINKING SOCIAL SECURITY PRIORITIES 3 inspired by the debates initiated in good measure by Averting the Old Age Crisis.2 There are some differences in our approach as well, principally because of differences in the circumstances that have prompted this inquiry. Whereas a primary (although not exclusive) objective of prior efforts has been to improve the public pension system's fiscal health or to help govern- ments better administer and regulate the systems, the principal objective of this inquiry is to try to determine whether participants (not just the admin- istrators or providers) in pension systems are better or worse off since the reforms. That is, we evaluate reforms from the viewpoints of individuals (and their households) and of the policymakers who represent them. A payoff to emphasizing the perspective of individuals is that this en- ables us to exploit well-accepted insights provided by the economics of insurance to answer the critical questions raised in the debate on social security, even some of those raised in Averting the Old Age Crisis (see box 1.1). Matching insights gained from applying an analytical frame- work to the problem of old-age income security with the experiences of countries that have reformed their social security systems can help point the correct way forward. It is reasonable to ask, however, whether enough time has passed to ex- pect tangible results, and whether we are too quick to assert that some fresh thinking is required. We think the time is right. Although most countries that have implemented the multipillar reform model improved incentives for workers to participate in the system, the main concern among policymakers is that the degree of coverage--measured as the number of workers partici- pating in formal pension arrangements--is now stagnant at levels less than half of the labor force. Covering the largest possible number of citizens Box 1.1 Three Central Dilemmas of Pension Privatization 1. If mandatory schemes are needed because of shortsighted workers, how can these same workers be counted on to make wise investment de- cisions? That is, if workers are myopic (which is the primary justification for the mandated private pillar) how can they be trusted to make good in- vestment decisions? 2. If governments have mismanaged their centrally administered pen- sion plans, how can they be counted on to regulate private funds effec- tively? That is, if governments mismanage pay-as-you-go (PAYG) systems, how can they be trusted to properly regulate mandated private pillars? 3. If government regulates and guarantees the plans, won't it eventu- ally end up controlling these funds? That is, does it really make a differ- ence whether the funds are privately or publicly managed? Source: World Bank (1994), p. 203. 4 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA against the risks associated with aging is among the objectives of policy- makers in every country of the region. We intend to persuade the reader that our approach, with its focus on the individual--and the role of government emerging from individual welfare maximization--provides useful pointers for policymakers who wish to increase the reach of pension systems. We provide analytical and empirical evidence to show that the reforms have been in the right direction. But we do not stop here. We go on to consider how this progress can be continued by meeting the many detractors of pen- sion reforms where they have somewhat cynically taken the debate (i.e., the concern for low coverage) and examine the problems raised from the per- spective of the participants in the new multipillar pension systems. Significant Progress, but Stalled Coverage Latin America is the right place to study pension reform. The longest and most varied experience with the multipillar approach is in Latin America. Starting with Chile in 1981, 12 countries in the region have adopted mul- tipillar arrangements, best distinguished from earlier systems by the prominence of a mandatory funded component administered by purpose- built and dedicated private providers. But often overlooked are the con- siderable differences in the systems adopted, most notably in the degree of choice allowed to workers between the old pay-as-you-go (PAYG) system and the new multipillar arrangement, and the level of benefits in the PAYG component. Costa Rica and Uruguay, for example, have kept a large earnings-related and defined-benefit (DB) system, whereas Argentina, Colombia, and Peru offer workers a choice between a reformed defined- benefit PAYG and the new defined contribution (DC) funded component to finance the bulk of their retirement income. In Mexico, on the other hand, workers rely fully on the funded system but have a guarantee of benefit levels equal to what they would have re- ceived under the old system. Chile and El Salvador also rely largely on the funded pillar, and the government has limited the PAYG component to providing a basic pension or "topping-up" to ensure a minimum level of retirement income. What is common, however, is the "individualization" of social security (see also Lindbeck and Persson 2003). When judged against the objectives of the reform, the multipillar approach can be credited with considerable success. The fiscal burden of pensions has been reduced: the most illustrative example is that total pen- sion debt-to-GDP (gross domestic product) ratios have fallen in most of these countries, as a result of both reduced benefits in the reformed PAYG component and a lower rate of accumulation of new liabilities (see Holz- mann, Palacios, and Zviniene 2001; Zviniene and Packard 2002). The re- forms appear to have improved the incentives to contribute to the formal system: recent analysis indicates that the introduction of individual RETHINKING SOCIAL SECURITY PRIORITIES 5 accounts (in which benefits are closely linked to contributions) lowers la- bor market distortions and improves incentives for workers to participate in formal pension arrangements (Packard 2001). The reforms have also increased equity: internal rates of return have become less regressive (Zviniene and Packard 2002). There also has been an increase in the depth of capital markets, at least in part attributable to pension reform (Yermo 2002a). These successes warrant that future reforms should build upon the new systems. In each of these areas, the experience has revealed shortcomings as well. In Argentina, Colombia, and Peru, the option given to new workers to choose between the old and new systems creates uncertainty regarding the fiscal liability of government. In Colombia, where workers can change their choice every three years, the disequilibrium between benefits and contributions in the old defined-benefit PAYG system is particularly acute and severely weakens the reformed system. Chile is increasingly concerned about the rising costs of the minimum pension guarantee, driven in part by falling numbers of active contributors in the labor force. And pensions of government workers continue to exercise a serious fiscal burden in coun- tries such as Argentina, Mexico, and Peru, although these constitute a greater burden even in countries, such as Brazil, that have not adopted the multipillar approach. These and other shortcomings, if not a failure of the reform model, are indeed failings of the actual reforms undertaken in the region. The ability of the multipillar model to isolate the pension system from abuse by governments may also have been oversold by reformers. It is now clear that unsustainable fiscal and monetary policies can jeopardize even well-implemented funded schemes. Whereas this was highlighted most dramatically in Argentina during the economic crisis in 2001 when the government made the administrators of second-pillar pension funds in- crease their holdings of increasingly risky government bonds and eventu- ally even confiscated their deposits in banks, similar threats to the viabil- ity of funded pension schemes can emerge in other countries of the region. In Bolivia, for example, there have been attempts to force a swap of dol- lar-denominated government debt held by pension funds for less attractive bonds denominated in the local currency (Escobar 2003). After rising modestly as a result of the reforms, coverage ratios have stalled at levels of about half of the labor force in those countries where workers' participation is highest. In most countries the ratios are much lower (see figures 1.1 and 1.2). Although many factors other than pension reforms (e.g., changes in labor and social legislation3) can affect partici- pation, stagnant coverage ratios are indicative of skepticism of the new system, despite its virtues. In Chile, for example, special survey data indi- cate that workers may perceive AFP (administradora de fondos de pen- siones; pension fund administrator) accounts as a relatively risky retire- ment investment (see Barr and Packard 2002 and Packard 2002); this is 6 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA confirmed by the low number of workers who use AFP accounts as in- struments for long-term saving despite their tax advantages. These same data show that many workers cease to contribute to the pension system af- ter completing the minimum contribution requirement, preferring other long-term savings instruments to those offered by dedicated pension providers, and even more importantly revealing a preference for govern- ment schemes for pooling resources to insure against old-age poverty, compared with government-mandated saving instruments (Packard 2002). Preliminary indications are that these may be more widespread phenomena: analysis of the contribution behavior of a sample of Peruvian affiliates suggests that the longer workers have contributed to the re- formed pension system, the less likely they are to continue contributing, and that where workers are free to choose how to save, many prefer to in- vest in housing and in their children. Figure 1.1 Pension Systems Cover between 10 and 60 Percent of the Economically Active Population in Latin American Countries (Percent) 70.0 with 60.0 sector 50.0 formal affiliated in 40.0 EAP or of 30.0 system 20.0 10.0 Proportion pension 0.0 Chile Peru Colombia Argentina Mexico Bolivia El Salvador Costa Rica Nicaragua 2000 1999 1999 2002 2001 2000 1998 2000 1999 Male 63.6 12.8 19.8 35.2 45.1 11.8 24.2 23.2 8.9 Female 60.9 9.1 26.9 37.2 46.6 8.3 27.6 22.5 15.3 Total 62.7 11.2 22.3 36.0 45.7 10.3 25.5 22.9 11.0 EAP Economically active population. Source: Household surveys between 1997 and 2002, analyzed by Todd Pugatch. RETHINKING SOCIAL SECURITY PRIORITIES 7 Figure 1.2 Between One-Tenth and Two-Thirds of the Aged Populations Receive Pensions in Latin American Countries (Percent) 80.0 70.0 older income 60.0 or 50.0 65 40.0 pension 30.0 20.0 10.0 Population receiving 0.0 Chile Peru Colombia Argentina Mexico Bolivia El Salvador Costa RicaDominican Republic 2000 1999 1999 2002 2001 2000 1998 2000 1997 Male 57.1 32.2 21.3 69.6 32.3 19.2 12.4 42.9 18.7 Female 27.6 7.9 9.9 64.0 10.5 5.6 5.8 23.9 10.6 Total 41.4 19.5 15.2 66.2 20.0 11.9 8.8 33.2 14.6 Source: Household surveys between 1997 and 2002, analyzed by Todd Pugatch. The coverage ratios shown in figures 1.1 and 1.2 are the best illustra- tions for why further thought has to be given to how to close the coverage gap, beyond the simple multipillar recipe of reforms. Although ensuring fiscal stability was the primary impetus behind the region's pension re- forms, advocates of multipillar reform, including the World Bank, saw the potential for increased coverage as an additional motivation (see box 5.2). But there are other reasons as well. Although the precarious fiscal posi- tions of governments in the region have resulted in high gross returns on the portfolios of dedicated pension providers (through high interest rates on government debt issues), several factors provide cause for concern. The first factor is how long these high returns can be maintained as fis- cal adjustment lowers the spreads on government debt. The second factor (where fiscal adjustment is slow) is that a good part of these high returns reflects the risk of default, as the experience in Argentina illustrates. The solution is greater diversification into domestic nongovernment and for- eign assets, and while these changes are difficult to engineer during times of fiscal stress, they should be introduced gradually as conditions improve. The third factor is that in some countries management fees have remained relatively stable, while the operational costs of the pension fund industry have fallen. This indicates deficiencies in regulation that should be cor- 8 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA rected, but also the impact of weak overall fiscal positions of governments in the region. In Chile and a few other countries, commission rates are slowly com- ing down to reasonable levels (less than 20 percent of contributions or 1 percent of assets). This raises questions of intergenerational fairness: the first generations of workers pay higher commissions in order to cover the start-up costs of the new industry of dedicated providers. The fee struc- tures in Chile and elsewhere also result in within-generation inequity as poorer workers end up paying a larger share of their contributions as com- missions (Yermo 2002b).4 But most worrisome are findings that point to difficulties in regulating the firms that manage mandatory pension funds. In Peru, for example, management fees have remained steady even though the ratio of operating costs to fees fell by almost half between 1998 and 2002 (Lasaga and Pollner 2003). Whereas these high management fees would be troubling if discovered for voluntary pension funds, such diver- sions from worker contributions that are mandated by governments should be cause for concern. Only Bolivia acknowledged the natural oli- gopolistic nature of the industry and from the start decided to minimize administrative costs within this constraint. The bidding contest used led to the lowest commissions in Latin America, although since then the fees charged in other countries have approached the Bolivian level. Governments also face a dilemma as they aim to make their funded sys- tems more flexible by permitting workers to choose among a range of in- vestments. In most countries, mandatory pension savings are backed by minimum pension guarantees (or even more generous guarantees) that ex- pose the state to a contingent liability that depends on the investment per- formance of the pension funds. The ensuing incentive to take unwarranted risks (what is commonly referred to as a moral hazard) is a reality in Chile, where workers can choose among five funds with different allocations to equities. The tension between benefit guarantees and individual choice is evidence of the inherent weakness of pension systems that rely exclusively on mandatory contributions to funded accounts.5 In summary, it is safe to conclude that these reforms have been a major step forward. As part I (chapters 2 through 5) of this book shows, the re- forms have led to lower fiscal burdens, a slower rate of growth of pension- related public liabilities, financial deepening and less regressive pension expenditures, although there is still much to do in these areas. As we rea- son in part II (chapters 6 through 8), the shift from defined benefit to de- fined contribution schemes as the mainstay of old-age income security ac- cords well with the basic principles of the economics of insurance. But it may also be reasonable to question the effectiveness of the currrent multi- pillar systems in creating an attractive instrument for retirement savings. Concerns over stalled coverage indicators and the vulnerability of all the multipillar components to macroeconomic instability--a fact of life in Latin America--reflect weaknesses that merit reexamination of the multipillar RETHINKING SOCIAL SECURITY PRIORITIES 9 parameters. In part III (chapters 9 through 11) we argue that fiscal, cov- erage-related, equity, and financial indicators over the last decade show that a return to single-pillar, defined-benefit PAYG systems as the main- stay for old-age income security is not the answer, although in some coun- tries in the region even this alternative is being considered.6 But the way forward is far less obvious. Using a blend of theory and em- pirical analysis, in part III of this volume we examine the options for the future and propose a rationale for continuing, redirecting, or strengthen- ing various aspects of the reforms initiated during the last two decades. Distinctions That Matter--Pooling vs. Saving, and Mandatory vs. Voluntary In this volume we categorize the components of a multipillar pension sys- tem by their objective, rather than by whom they are administered (the public or private sector), how benefits are structured (final-salary benefit formula or defined contributions), or their financing mechanism (PAYG or full funding). Thus we use the term "first pillar" or "pillar one" to re- fer to the part of a pension system intended to keep elderly people out of poverty,7 "second pillar" to refer to the mandated part intended to help individuals smooth consumption over their life cycle (i.e., to prevent a dra- matic fall in income at the time of retirement), and "third pillar" to iden- tify the instruments and institutions available on a voluntary basis for workers to increase their income in retirement. A simple way to charac- terize the main differences among these pillars is the differing role of gov- ernment: in the case of the first pillar government defines benefits, in the second it defines contributions,8 and in the third it defines incentives for retirement savings. Table 1.1 summarizes the main features of instruments for old-age income security that together constitute the multipillar system. In general the issue of interest is not whether a country should have three pillars (or tiers) or just one. It can be shown without much difficulty that in- dividuals are better off diversifying the risks to adequate income in old age and thus that a country does better for its elderly population by instituting more than one of these components (see Lindbeck and Persson 2003). The critical question is: What should be the relative importance of the three pil- lars--that is, their "weights" in the system? That is one of the questions this volume addresses. In doing so it exploits two fundamental dichotomies: 1. The nature of the instrument: pooling, where there is by definition a transfer in every period from the more to the less fortunate; and saving, where there is by definition a transfer of one's income from one period to another but no redistribution between individuals.9 2. The role of government or the main reason for participation by the individual: mandatory, where the government mandates participation and 10 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 1.1 Instruments of Old-Age Income Security Nature of Mainstay: Pooling Mainstay: Savinga instrument Mandatory Mandatory Voluntary Common name First pillar Second pillar Third pillar Main function Insure against Smooth con- Smooth con- poverty in old sumption sumption over age, lower over life cycle life cycle income inequal- ity Main role of Defines benefits Defines contri- Defines incen- government butions tives Principal risk- Government Worker Worker bearer Financial instru- Unfunded PAYG Funded: indi- Funded: tax- ment vidual preferred indi- accounts vidual accounts a See chapter 6 for an important qualification: the use of annuities in the saving components implies that the risks associated with unexpected longevity are pooled. defines the rules of the game; and voluntary, where the rules are made clear but people have the choice to participate or not. Exploiting the first dichotomy enables the use of well-developed insights from the economics of insurance to answer the question of relative weights of the first pillar on the one hand, and the second and third on the other. The first pillar is pooling-based whereas the second and third are primarily savings-based, although there is an important role for pooling in the form of annuities and survivors and disability insurance. Throughout this book we refer to the first pillar as the pooling component and the second and third pil- lars as the savings component of a pension system. Combined with insights from the study of household behavior and financial institutions, the second dichotomy helps in deciding the relative weights of the second and third pil- lars. Together these can help determine the unfinished reform agenda needed in countries that have already adopted the multipillar approach, and the options that should be considered in those that are seriously contemplating pension reform. It may even persuade obstinate nonreformers to reexamine their strategies to help elderly people achieve income security. A Summary of the Main Findings This volume approaches the problem of old-age income security princi- pally from the viewpoint of individuals, rather than that of governments RETHINKING SOCIAL SECURITY PRIORITIES 11 alone. Although previous analyses have certainly not ignored the individ- ual's perspective, an explicit focus on the individual has been employed too infrequently in the literature in our view. Consistent with the advice of Barr (2001), who pointed out that analysis of pensions requires an un- derstanding of macroeconomics, microeconomics, financial economics, and the theory of social insurance, this book examines pensions in Latin America using all four of these lenses. Taking this perspective, we find that the successes of pension reform are not where commonly believed: the successes of the reform are not in the much-touted shift to "pre-fund- ing," but in the switch from pooling to saving as the mainstay of old-age income security. Put another way, the merit of the reform is not in the pri- vatization of schemes for old-age income support but in their "individu- alization." And contrary to the claims of proponents of reforms, the strength is not in arriving at a durable and permanent system, but in breaking with a past of approaches that demographic and economic changes have made defunct. It should therefore be emphasized once again that future reforms should build upon the efforts that countries have undertaken. The objective of fu- ture efforts should be to improve the functioning of all components--first, second, and third pillars--in countries that have adopted the multipillar approach, within the fiscal and administrative constraints they face. In summary, this volume uses available evidence, including that pre- sented in background papers and surveys commissioned specifically for our inquiry, to draw several important lessons from the Latin American experience with pension reform. First, and most important, the poverty prevention pillar should get a lot more attention than it has in Latin America during the last decade. This poverty prevention role of government only increases in importance with economic development--as the likelihood of poverty in old age de- clines, the fundamentals of insurance make pooling of this risk across in- dividuals more, not less, appropriate. A government mandate is necessary for such a defined benefit system because private insurance markets are unlikely to provide such coverage. Systems of this type are also best fi- nanced and managed separately from the savings component, which is not the case in countries such as Chile, El Salvador or Mexico. The de- fined benefit formulas of such programs call for conservative investment strategies that can clash with the need for individual portfolio choice in the savings pillar. Second, it should be emphasized that although such poverty prevention tiers will provide a minimum pension to those people who are unfortunate or unwise, saving should be the mainstay for earnings replacement during old age (i.e., mechanisms to cover the loss of earnings capacity while living). Retirement programs on top of the poverty prevention pillar should closely link benefits to contributions and do so in a similar way for most work- ers. The individual capitalization schemes that have been introduced in 12 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Latin America are fully consistent with this objective. In particular, rising life expectancy does not affect the link between benefits and contributions in these new schemes. On the other hand, countries that still have large earn- ings-related defined-benefit PAYG pillars (e.g., Colombia, Costa Rica, and Peru) will see benefits grow out disproportionately relative to contributions as life expectancy increases. These countries may therefore consider adjust- ing their benefit formulas to replicate a savings system (through notional in- dividual accounts) or may provide more space to the funded system by clos- ing down the earnings-related PAYG pillar to new entrants to the labor force. Third, incentive compatibility between the poverty prevention and in- dividual savings pillars is key. Excessively generous or badly designed poverty prevention pillars may create further incentives to informalization of employment and reduce contributions to the mandatory and voluntary savings pillars. Fourth, more attention should be paid to the size of the mandatory savings pillar. High contribution rates and maximum taxable earnings can discourage worker participation, especially by young and poor pop- ulations with other urgent, competing demands on their disposable in- come. Large second-pillar pensions may be a useful instrument for ef- fecting a transition from overly generous defined-benefit PAYG systems. They may also provide an initial boost to capital and insurance markets. However, these are needs that become less important over time, calling for a reduced mandate. Mandatory savings plans may not even be nec- essary in countries such as Brazil, which have overcome many political hurdles to social security reform and already have the foundations for thriving capital and insurance markets. Careful consideration of country circumstances, backed by solid country-specific analytical work, is nec- essary to determine the appropriate size of the second pillar. Fifth, for countries that do have mandatory savings plans, the priority should be to lower costs to affiliates and improve risk management. Fur- ther reductions in commissions would also improve the attractiveness of the funded pillar, and there are various options for achieving this goal. At one extreme there is the centralized management model. At the other ex- treme there is a fully contestable market where different providers com- pete by offering diverse products. Latin American countries will need to assess which is the best solution for them, and we do not propose country- specific solutions in this volume. The choice will depend largely on the ex- tent to which other financial institutions are appropriately regulated and supervised and the population's ability to make difficult choices. With re- spect to risk management, more consideration should be given to the value of international diversification of pension fund portfolios, the pros and cons of worker choice, and improving benefit options at retirement. Countries with high fiscal net liabilities (and thus regressive transfers) may need to eliminate those net liabilities to make the "fiscal space" needed to fund a poverty reduction pillar. RETHINKING SOCIAL SECURITY PRIORITIES 13 A Roadmap to this Volume This volume consists of three parts. In part 1, following a brief description of the reforms in the region in chapter 2, in the next three chapters we provide evidence on the performance of countries that have adopted the multipillar approach in three dimensions: fiscal, financial, and social. The numbers show that countries in Latin America that have adopted the mul- tipillar approach have done well in terms of the objectives of reforms. This finding should form the basis of future reform efforts. Part 2 makes the point that these countries have made progress in an- other (much less widely acknowledged) aspect: they have made or begun the transition to a more sustainable and meaningful social contract. The main reason for judging the changes with optimism is the switch to sav- ings from pooling as the basis or mainstay of old-age income security. One can reasonably make the case that the shift from unsustainable defined- benefit PAYG to sustainable old-age income security systems had to go through this stage for reasons of political economy. So it would be a mis- take to go back to the unsustainable structures that existed prior to the in- stitution of the multipillar systems. But it would also be a mistake to think of this stage as the "final structure." In fact, the report argues that the 1990s could be seen as a transition in reforming Latin American countries to structures that are sustainable at their levels of institutional sophistica- tion. Chapter 6 asks and answers the question, how do we know what is the appropriate structure? The main proposals stem from well-accepted principles in the economics of insurance that argue for saving as the main- stay of a comprehensive insurance strategy against a frequent loss--of be- ing without earning capacity while living--and pooling as an auxiliary in- strument for a risk that is now smaller--that of poverty in old age. Chapter 7 continues to address the issue using this relatively simple an- alytical framework. It examines how well the mandatory and voluntary savings components have done from the individual worker/contributor's perspective. High contribution rates may have discouraged young and poor workers from participating in the formal pension system. Most coun- tries also have very high earnings ceilings for calculating mandatory con- tributions and some have no ceiling at all. High contribution rates and earnings ceilings may explain why few workers have found it worth mak- ing voluntary contributions to their individual accounts in most countries, even when they are relatively liquid and offer attractive tax benefits. In ad- dition, the operation of the savings component (both the mandatory and voluntary parts) presents some weaknesses arising primarily from high commissions, lack of international diversification, and inadequate risk management over the life cycle. Chapter 8 provides some evidence on what may be wrong with seeing the multipillar structure with its heavy reliance on mandated saving as a final structure. Using survey data from Chile and Peru, the two countries 14 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA with the longest experience with mandated private savings, we provide ev- idence that supports worker rationality and reveals their preference for government provision of instruments to insure against old-age poverty over those that enable individual saving. Although the evidence is not definitive given the small size of the samples, it is more than merely suggestive. In view of this rationality (evidence of the ability to distinguish between risky and less risky instruments of old-age income security, and preferences for alternative long-term savings and investments) and in light of governments' difficulties in providing either PAYG pensions or efficient savings instruments, we propose here that the size of the mandate to save in the form of a rigidly defined and not-so-easily regulated instrument de- serves to be reassessed. Part 3 of the book proposes that the time is right to reflect on the gov- ernment mandate to save for old age. Chapter 9 discusses how best to insure against the risk of poverty in old age, and chapter 10 treats the equally important issue of facilitating saving for retirement--the mainstay of old-age income security. The volume proposes continuing the move to- ward a system that consists of a sustainable first pillar to address the risk of poverty and a vibrant, competitive savings pillar to address the need for consumption smoothing over the life cycle. Chapter 9 proposes what may be seen by some as a radical increase in attention of Latin American gov- ernments to the poverty prevention function of public pensions. In con- trast, chapter 10 proposes a gradual reform of plans to encourage saving for old age. But the principle that guides all the recommendations is that these changes be welfare-improving, institutionally feasible, and fiscally sustainable. In chapter 11 we look back and ask whether the decade or so of reform has been a success. The answer is that in many respects, it has. But if the new structures are viewed as a final design, the efforts may well be as- sessed harshly because scores of people are left uncovered just as they were under the old systems, there are still some adverse equity effects and fiscal liabilities, and the cost and risk management features of the savings pillar are somewhat deficient. On the other hand, if the current structures are viewed as a transitory stage, social security reforms should be viewed as successful because the movements have been in the right direction. This is true in all countries surveyed in this book. In countries such as Brazil this is also true, even though the country has eschewed a mandated private sav- ings pillar in favor of efforts to strengthen the third pillar.10 Even in Ar- gentina, where the second pillar is in the midst of a crisis, we believe that the current social security system is superior to the one it replaced; the re- forms of the 1990s have to be built upon, not abrogated. The greatest dangers to all that the reforms have achieved lie not in countries where the new approach to ensuring income support for the aged is being scrutinized and altered, but in countries where large mandated saving is viewed as a solution for the ages. RETHINKING SOCIAL SECURITY PRIORITIES 15 Notes 1. Barr (2001) correctly points out that "tier" is a better characterization than "pillar" because it "is linguistically more apt: pillars can only be effective if they are all in place and all, broadly of the same size . . ." (p. 133). Because the relative size of these components is a central concern of this book, Barr's point is especially pertinent. 2. All 16 background papers for the book are available online at the Web site of the World Bank's Chief Economist for Latin America and the Caribbean region, www.worldbank.org/keepingthepromise. 3. We discuss the range of factors influencing the decision to participate in public pension systems more extensively in chapters 5 and 8. 4. In addition to the high costs borne by younger workers, a comprehensive generational accounting framework would necessarily account for many other fac- tors to estimate the net intercohort impact of reforms. Such a framework is beyond the scope of this book, however. 5. Even in Chile, a country with one of the highest participation rates, the gov- ernment considers that low participation in the private pension system will keep ef- fective replacement rates low and thus put mounting pressures on the minimum- income plan for retirees. 6. In Argentina, for example, a draft law that would allow workers to switch between the public and the private branches of the pension system was passed in 2002 in the Lower House with only one vote against and one abstention. In Peru in late 2002 some articles proposed during the rewriting of the constitution would have allowed affiliates to the funded system to return to the public PAYG system and to lower the retirement age from 65 to 60 years. These articles were only nar- rowly defeated. In Chile in early 2002 civil servants started demonstrations de- manding to be allowed to switch back to the pre-reform PAYG regime as a result of disappointing projected replacement rates from individual accounts. 7. We discuss the emerging distinction between "pillar one" and "pillar zero" poverty prevention pensions in chapter 9. 8. Although the second pillar can be implemented as a defined benefit scheme, its predominant association with defined contribution plans in both theory and practice overwhelms this distinction in our view. 9. Because the Ehrlich-Becker "comprehensive insurance" framework under- lying our approach (and presented in chapter 6) would use the term "insurance" in reference to both the consumption-smoothing ("self-insurance") and poverty pre- vention ("market insurance") functions of social security, we use the more con- temporary terms "pooling" and "saving" in order to avoid confusion. 10. The recent decision to raise the maximum taxable salary, however, will re- duce the clientele in the third pillar substantially. I Retrospective: Fiscal, Financial, and Social Benefits from Pension Reform 2 Structural Reforms to Social Security in Latin America AMONG DEVELOPING REGIONS, LATIN AMERICA has a relatively long tradi- tion of formally institutionalized social security. Governments at various levels, unions, and trade associations have been administering retirement, disability, survivor insurance, and in many cases unemployment insurance since the early 1900s. National public pension systems were first estab- lished in Chile in 1924 and then in Uruguay in 1928. In the past 20 years, however, a slow but dramatic shift has occurred across the region in the approach taken by governments to providing social security, primarily in retirement pensions.1 Demographic Changes in Latin America As in other regions, Latin America's population is aging. Advances in tech- nology and health care have increased average life spans dramatically in the last 50 years. Although the pace of the region's demographic transition varies widely, from relatively "younger" countries like El Salvador to rel- atively "older" countries like Uruguay, falling fertility rates combined with lengthening life expectancy are increasing the portion of the popula- tion in old age and lowering the number of new entrants into the work- force (see figure 2.1). This demographic change has been accompanied by economic liberal- ization and greater integration with the world economy. Structural adjustment after the debt crisis in the 1980s and the need for greater effi- ciency as countries opened their economies to competition from abroad in the 1990s forced a steady reallocation of the labor pool. Changes in the relative size of different branches of the economy show a clear increase in the number of workers employed in small firms, temporarily employed, and self-employed and a fall in the number of people working in large pri- vate firms and in the public sector. 19 20 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 2.1 Rising Life Expectancy Increases the Share of Elderly People in the Population and Upsets the Balance of Pure PAYG Pension Systems Rising life expectancy and falling fertility rates in Latin America and the Caribbean 90 6 80 5 70 Fertility 60 4 50 rate 3 (years) 40 (percent) Age30 2 20 1 10 0 0 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Life expectancy at birth Total fertility rate Growth in the share of elderly people and the push for competitiveness have forced policymakers in the region to reexamine labor market institu- tions--primarily public pension systems--to accommodate these trends. Latin America, however, is still a relatively young region. The fiscal deficits and mounting contingent liabilities of overly generous public pen- sion systems plagued by mismanagement and fraud have often proven a more immediate impetus for structural reforms. Taxonomy of Social Security Reform in Latin America2 Structural reforms to social security in the region were initiated by Chile in 1981 and continued in the 1990s by Peru, Colombia, Argentina, Uruguay, Mexico, Bolivia, and El Salvador. Costa Rica, Nicaragua, the Dominican Republic, and Ecuador enacted reforms between 2000 and 2001.3 Each reform involved a transition from purely public pension sys- STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 21 Figure 2.1 (continued) Increasing number of elderly people in Latin America and the Caribbean 100 90 80 70 60 population 50 of 40 30 Percentage 20 10 0 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Percentage of working- Percentage 65 years age population and older Source: ECLAC/CELADE (1998). tems (similar to those administered in Europe and the United States) to systems with explicitly defined "multiple pillars" administered and/or mandated by government. Although reforms have varied widely, most countries have retained or restructured a publicly mandated and administered first pillar operated on a PAYG, defined benefit basis with a redistributive safety-net function. To that pillar they have added a publicly mandated but privately administered second pillar of individual retirement savings accounts funded with de- fined contributions and managed by a new industry of dedicated pension fund managers. Finally, reformers have tried to increase savings for retire- ment by defining incentives through regulation of a third pillar of volun- tary retirement saving and pension plans arranged privately between em- ployers and their workers. Although the function (and sometimes the form) of formal pension in- stitutions that existed prior to these reforms could also be categorized in 22 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA three pillars,4 the multipillar model has come to be distinguished in Latin America by the prominence of mandated private savings, relative to mandatory public social insurance and voluntary forms of savings and in- surance. Most multipillar systems are designed so that the bulk of workers' re- tirement income is financed from mandated private savings in individual accounts. These funds are invested in bonds and equities by dedicated pri- vate pension fund managers. Although the activities of these fund man- agers and other financial service providers in the voluntary pillar are strictly regulated, the direct role of the state in the new model is reduced to enforcing the mandate to save, regulating the new fund management industry, and guaranteeing a minimum threshold income to keep individ- uals from falling into poverty in old age. What should be emphasized is that the general character of social se- curity reforms in the region has been similar. Despite claims to the con- trary made by several opponents of reforms, however, there is no single, cookie-cutter "Latin American" reform model. There are important dif- ferences that mark the introduction of the new pension systems from one country to the next. Tables 2.1 and 2.2 show some of the principal char- acteristics and differences between (structurally) reformed pension sys- tems in the region.5 In Bolivia, Chile, El Salvador, and Mexico, mandated individual retire- ment accounts replaced public defined-benefit PAYG institutions as the state's primary intervention to provide retirement income. In addition, in Chile there is a (broadly defined) first pillar that consists of guaranteed minimum contributory benefits, subsidies, and other social assistance to the elderly indigent population, financed through general taxes.6 In El Sal- vador the state also guarantees a contributory minimum pension but there is no social assistance program for elderly people who do not meet contri- bution requirements. In Mexico workers who contributed to the old sys- tem have a guarantee of benefits equal to their accrued benefits under the old system. As in El Salvador, there is no social assistance program for eld- erly men and women. Bolivia offers a basic pension to all Bolivians 65 and older who were at least 21 at the end of 1995. In contrast, in Argentina, Costa Rica, Ecuador (if and when the reform is implemented) and Uruguay, a PAYG-financed, contributory first pillar underpins pensions financed with accumulated individual savings. This first pillar provides a basic flat-rate pension in Argentina and earnings- related pensions (backed by minimum pension guarantees) in Costa Rica, Ecuador, and Uruguay. Benefits are conditional on a certain minimum number of years of contribution.7 Rather than pillars, in Ecuador and Uruguay the new systems are best characterized as "tiers." Participation in the new tier of private individual accounts is only mandatory for workers whose salary is above a certain level. In Uruguay only 10 percent of all contributors earn above the threshold level of approximately six minimum STRUCTURAL Table 2.1 Principal Features of Structural Reforms to Social Security Systems (Old-Age Disability and Death) in Latin America during the 1980s and 1990s Chile Peru Colombia Argentina Uruguay Mexico Bolivia El Salvador Year of reform 1981 1992/1993 1994 1994 1996 1997 1997 1998 REFORMS Contribution-related Closed Remains Remains Remains Remains Closed Closed Closed PAYG system? Total payroll tax 33.0 18.0 17.8 42.0 40.0 20.0 19.0 11.8 TO rate, pre-reform (%) SOCIAL Total payroll tax 20.0 20.5/22.0a 33.8 46.0b 40.0 26.0 24.0 13.5 rate, post-reform (%) SECURITY Participation of new Mandatory Voluntary Voluntary Voluntaryc Voluntaryd Mandatory Mandatory Mandatory workers? Participation of self- Voluntary Voluntary Voluntary Mandatory Mandatory Voluntary Voluntary Voluntary INLATINAMERICA employed workers? Remaining separate No Noe Yes Noe No Yes No No system for civil servants? Dedicated fund AFP AFP AFP AFJP AFAP AFORE AFP AFP managers Contribution to IRA 10.00 8.00 10.00 7.72 12.27 12.07 10.00 10.00 (%)f Fees and insurance 2.31 3.73 3.49 3.28 2.68 4.48 2.50 3.00 premiums (% of 23 wage) (Table continues on the following page.) 24 Table 2.1 (continued) Chile Peru Colombia Argentina Uruguay Mexico Bolivia El Salvador Switching between 2 times 1 time 2 times 2 times 2 times 1 time 1 time 2 times KEEPIN fund managers? annually annually annually annually annually annually annually annually Payout options Annuity or Annuity or Annuity or Annuity or Annuity Annuity or Annuity Annuity or scheduled scheduled scheduled scheduled only scheduled only scheduled G THE with- with- with- with- with- with- drawal drawal drawal drawal drawal drawal PROMISE Minimum return on Relative to Relative to Relative to Relative to Relative to Unregulated Unregu- Relative to investment? average average average average average latedg average Minimum contribu- Yes Yes (only Yes Yes Yes Yes No Yes tory pension? for affili- OF ates born SOCIAL before 1945) Social assistance Yes No No Yes Yes No Yes (only No SECURITY pension? affiliates born before INLATINAMERICA 1974) Notes: AFAP pension fund administrators (administradoras de fondos de ahorro previsional [Uruguay]); AFJP retirement and pension fund administrators (administradoras de fondos de jubilaciónes y de pensiones [Argentina]); AFORE retirement fund administrators (administradoras de fondos para el retiro [Mexico]); IRA individual retirement account. STRUCTURAL a. 20.5 for private pension system, 22.0 for national system. Maximum taxable earning for disability and survivorship insurance: S/6130.88 or US$1,751. b. Maximum allowed by law. The effective tax rate has been falling since the reform, and it varies by sector and region. The current rate is less than 30 percent. c. Although new affiliates can choose, up to 80 percent of them in each year fail to make an explicit choice. The private second pillar is the default option. REFORMS d. Participation in individual accounts in Uruguay is determined by income level. Workers below a threshold level choose to split contributions between PAYG and individual retirement accounts. e. Exceptions exist for some subnational systems. f. At time of publication. TO g. Guarantees required from the fund managers. SOCIAL Source: Adapted from Cerda and Grandolini 1998; Queisser 1998a; Devesa-Carpio and Vidal-Meliá 2002; AIOS 2002; FIAP 2002; and country pension supervisors. The authors wish to thank Ximena Quintanilla (Chile), Elio Sanchez (Peru), Carlos Grushka (Argentina), Maria Nela Seijas (Uruguay), and Francisco Sorto Rivas (El Salvador) for their assistance in compiling table 2.1. The authors take sole responsibility for factual errors SECURITY in the table. INLATINAMERICA 25 26 Table 2.2 Principal Features of Structural Reforms to Social Security Systems (Old-Age Disability and Death) in Latin America during the 1990s and 2000s Costa Rica Nicaragua Ecuador Dominican Republic KEEPIN Year of reform 1995/2000a 2000, as yet not 2001, as yet not 2001b implemented implemented Contribution-related public PAYG Remains Closed Remains Closed G system? THE Total payroll tax rate, pre-reform (%) 22.00 17.00 -- 9.25 PROMISE Total payroll tax rate, post-reform (%) 26.0 21.5 Varies, but no more 20.0 than 20.0 Participation of new workers? Mandatory Mandatory Mandatory Mandatory Participation of self-employed Voluntary Voluntary Mandatory Mandatory OF workers? SOCIAL Remaining separate system for civil -- No -- Yes servants? Dedicated fund managers OPC AFP EDAP AFP SECURITY Contribution to IRA (%) 4.25 7.50 8.33 10.00 Fees and insurance premiums (% of c 2.5 4.0 2.0 wage) INLATINAMERICA Switching between fund managers? 1 time annually 1 time annually -- 1 time annually Payout options Annuity or scheduled Annuity or scheduled Annuity or scheduled withdrawal withdrawal withdrawal Minimum return on investment? Unregulated Unregulated Relative to average Relative to average Minimum contributory pension? Yes Yes -- Yes Social assistance pension? Yes Yes Yes Yes STRUCTURAL Notes: EDAP future savings depositary entities (entidades depositarias de ahorro previsional [Ecuador]); IRA individual retirement account; OPC complementary pension fund operators (operadoras de pensiones complementarias [Costa Rica]); --not available. a. Costa Rica introduced voluntary retirement accounts in 1996 but made private individual retirement saving mandatory as a complement to the defined benefit system in 2000. b. Implemented in 2003. c. Fees are charged as a percentage of returns from investment and capped at a maximum of 8 percent of returns or 4 percent of contributions. REFORMS Source: Adapted from Cerda and Grandolini 1998; Queisser 1998a; Devesa-Carpio and Vidal-Meliá 2002; AIOS 2002; FIAP 2002; and country pension supervisors. The authors wish to thank Tomás Soley (Costa Rica) for assistance in compiling table 2.2. The authors take sole responsibility for factual errors in the table. TO SOCIAL SECURITY INLATINAMERICA 27 28 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA wages. Workers with lower salaries can if they wish deposit up to one-half of their mandatory contributions in the funded plan. In Argentina, Peru, and Colombia each new generation that joins the labor force (and that takes up formal employment) is allowed to choose between a significantly downsized, earnings-related PAYG pillar and indi- vidual retirement accounts as the primary financing mechanism for their pensions. Irrespective of their choice, Colombian workers are covered by the same first-pillar arrangements (i.e., minimum guarantees and basic poverty prevention pension).8 In Peru, as in Argentina's second pillar, new workers who choose the public pillar are always allowed to move to the private system at a later date. However, workers who choose private ac- counts cannot choose to move back to the PAYG pillar. As a political con- cession to pass the reform, workers in Colombia were given the option to switch back and forth between systems every three years. Largely in reac- tion against a policy of assigning all undecided workers to private individ- ual accounts by default, draft legislation to allow similar switching was considered in Argentina (Rofman 2002). Similar attempts by the legislature to allow affiliates to return to the public system were recently defeated in Peru. Reforms in Latin America, therefore, differ significantly in the extent of private provision of formal retirement income. There is not yet a consen- sus on how to measure the extent of "privatization" of what were once purely public national pension systems. Some authors present the manda- tory contribution rate earmarked for private individual accounts as a por- tion of total mandated contributions for retirement income security (as in Palacios and Pallares-Miralles 2000), or contributions to individual re- tirement accounts as a share of total payroll taxes for social insurance (as in Packard 2001). Others propose the projected portion of future pension benefits that will come from the new private pillars (as in Brooks and James 2001) as a proxy for the degree of privatization. From the perspective of new entrants to the labor force, the reformed Latin American pension systems can be divided into four main groups (see figure 2.2): 1. Countries where the funded component is the only source of earnings- related contributory pensions available for new workers: Bolivia, Chile, the Dominican Republic, El Salvador, and Mexico. 2. Countries where new workers must choose between the funded and the PAYG system for their earnings-related pension: Colombia and Peru. In Colombia workers may switch between the two systems every three years. In Peru the default option is the funded system. 3. Countries where new workers remain in a PAYG system providing a basic, flat-rate pension and choose between a complementary PAYG and a funded pillar for their earnings-related pension: Argentina. The default option is the funded system. STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 29 Figure 2.2 Destination of Mandatory Pension Contributions Choice of Choice of system system Mandatory PAYG FF FF FF PAYG FF contributions PAYG PAYG ·Bolivia ·Colombia ·Argentina ·Costa Rica ·Chile ·Peru ·Uruguay ·Dominican Republic ·El Salvador ·Mexico Note: FF Fully funded system. 4. Countries where new workers remain in a reformed PAYG system that provides earnings-related defined benefit pensions that are lower but still generous, and make additional mandatory contributions to the funded plan: Costa Rica and Uruguay. In Uruguay the funded plan is op- tional for low- and middle-income workers. Although not yet imple- mented, structural reforms in Ecuador would introduce a system similar to that of Uruguay. In most countries the reform did not equally affect private and public sec- tor employees. Usually out of political considerations, reforming govern- ments avoided making structural changes to the public pension systems ben- efiting the military and civil servants. After structural reforms to the social security systems for workers in the private sector, separate public pension systems remain for civil servants in Colombia, El Salvador, and Mexico. Al- though the plan is nominally closed to new entrants, organized groups of workers have continually tried to be covered by the generous PAYG pa- rameters of Peru's pension regime for civil servants (the cédula viva), with mixed success. The retirement regimes for federal civil servants in Argentina were integrated into the reformed national system, but separate, relatively generous pension systems remain for civil servants in roughly half of Ar- gentina's provinces. The military (and in many cases the police) retain sep- arate, special retirement and other income security arrangements in all of the countries where reformed systems are now in place. 30 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA There also are important differences among countries, even within each of the four groups with respect to the transition arrangements put in place for private sector workers who were already participating in the formal pension system prior to reforms. In Bolivia and Mexico all workers were switched to the new system. In Chile workers were given a choice between staying in the old system and switching to the new one. In the Dominican Republic all workers under 45 years of age are required to contribute to the new plan. In El Salvador all workers under 36 years of age at the time of the reform were required to switch to the new system. Men between 36 and 55 and women between 36 and 50 were given one year to choose be- tween the old and the new systems. The default option for these workers was the funded system. Older workers were left in the old system. In Uruguay, all workers over 40 were given the choice between the old and new systems, and those below that age were automatically transferred to the new plan. In Argentina, Colombia, and Peru all workers who were un- der the old PAYG system and did not express a wish to switch to the new privately funded pillar remained in the reformed PAYG system. Past contributions to the old system for those who decided or were re- quired to switch to the new system were acknowledged in most cases through recognition bonds. The main exceptions are Argentina and Uruguay, where a complementary benefit will be paid at retirement, and Mexico, where contributors in the old system at the moment of the reform were offered a guarantee that their retirement benefit would be no lower than that which would have been received in the old PAYG system. Except in Chile where the system has been in place for more than two decades, Latin American workers retiring from the reformed pension sys- tems over the next decades will receive most of their retirement income in the form of benefits that accrued under the old PAYG systems. Therefore, the analysis in this chapter on the performance of the funded pillar pro- vides only a partial view of the performance of the overall mandatory re- tirement income security system. Over time, however, benefits accrued in the old PAYG systems will dwindle in relation to those accrued in the new funded pillars, except in Argentina, Colombia, and Peru, where the rela- tive size of each pillar will depend on workers' choices, and in Costa Rica, which decided to retain a large PAYG system. Since the reform, however, new workers in Argentina and Peru have overwhelmingly chosen the funded pillar, signaling some degree of confidence in the new system.9 Structure and Implementation of the Mandatory Savings Component In this section and the next we will specify in greater detail the structure of the "savings" pillars of multipillar reforms in Latin America, address- ing first the mandatory and then the voluntary components. It should be STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 31 noted that in some countries, such as Argentina, Colombia, and Peru, the mandatory component is not mandatory in one sense of the term--work- ers can choose between a downsized earnings-related PAYG plan and the funded individual accounts. But it is mandatory in that the worker must pick one of these alternatives. The accumulation stage (of both the mandatory and the voluntary savings components) is based on a defined contribution formula whereby contributions are saved in pension fund accounts managed by dedicated pension fund administrators.10 A portion of workers' contri- butions to the private pillar pays for the services provided by the fund managers and covers the cost of premiums for group disability and life insurance policies they are required to provide to contributing work- ers.11 Workers are allowed to choose their fund manager from among the limited number in the closed industry, and workers who have been with a fund manager for some minimum period of time are allowed to switch to another manager. The minimum mandatory contribution is based on the minimum salary in all countries except Uruguay, where contributions to the pension fund system are mandatory only on the portion of the salary above approxi- mately six minimum wages for workers who earn above this level. Work- ers who earn less may direct up to half of their mandatory contributions to the pension fund of their choice. The contribution rates that are chan- neled into the individual's capitalization account (net of commissions or insurance premiums) and the maximum monthly earnings subject to the mandate to save are shown in table 2.3. Contribution rates for the indi- vidual account vary over time in four countries--Argentina, El Salvador, Mexico, and Uruguay--because commissions and insurance premiums are paid from the total contribution. In the remaining countries that have un- dertaken reforms, on the other hand, the contribution for the individual account is fixed as a percentage of the worker's salary. In Mexico work- ers make an additional mandatory contribution to the housing fund man- aged by a state body, INFONAVIT. Contributions and investment income are tax-exempt, but retirement income is taxed. In Mexico workers who participate in the funded system also get a public subsidy equivalent to 5.5 percent of the minimum wage in January 1997, indexed to the consumer price index (CPI; see box 9.1). Self-employed workers who contribute to the new mandatory funded sys- tem in the Dominican Republic will also receive a public subsidy, but its value is still to be determined. In Colombia another form of account sub- sidy has been introduced in a solidarity tax on affiliates with higher incomes. Contribution collection and record keeping are managed by the pension fund administrators in all countries except Argentina, the Dominican Repub- lic, and Mexico, where this function has been centralized. In Argentina the in- ternal revenue tax authority is responsible for collecting income taxes and all 32 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 2.3 Contribution Rates and Earnings Ceilings in the Mandatory Funded Systems (December 2002) (Percent of wages) Contribution into Maximum contributable fund/individual's earnings/average national Country salary earnings Argentina 2.75 5.80 Bolivia 10.00 12.50 Chile 10.00 3.10 Colombia 10.00 10.00 Costa Rica 4.25 No ceiling Dominican Republic 10.00 10.30 El Salvador 11.02 14.40 Mexico 6.27 6.40 Peru 8.00 No ceiling Uruguay 12.27 5.70 Notes: The total contribution rate (for capitalization, insurance, and commission) was cut to 5 percent in Argentina in December 2001. The contribution that is chan- neled into the individual's account therefore dropped dramatically between 2001 and 2002. During 2003 the total contribution rate was gradually raised back to its original value (11 percent). The contribution rate in the Dominican Republic is rising gradually until it reaches 10 percent in 2008. Similarly, the contribution rate in Costa Rica is be- ing increased gradually from 1.5 percent to 4.5 percent. In Uruguay poorer workers can split their 15 percent mandatory contribution equally between the PAYG and the funded system. The figures reported assume that the total contribution is made to the funded system. Figures for Chile are based on the average earnings for workers who contributed to the system in December 2002. (This average is higher than the national one because of the informal and self-employed workers.) Sources: Devesa-Carpio and Vidal-Meliá (2002) and Whitehouse (2001). payroll contributions for social insurance programs, including the pension system. In contrast, Mexico established a centralized record-keeping and collection system with the sole purpose of handling the flows of informa- tion and funds in the new pension system. Another important difference is that the Mexican agency is owned by the new dedicated pension fund man- agement industry and set up as a nonprofit company. In the Dominican Republic a nonprofit, private foundation is also responsible for record keeping. The Treasury monitors this institution and is ultimately responsi- ble for sanctions and for ensuring that funds are correctly allocated to each of the individual programs within the social security system. The pension fund administrators charge a fee or commission in remu- neration for their services. All countries except the Dominican Republic STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 33 and Mexico permit only charges on contribution flows. In the Dominican Republic a performance-related fee is allowed, whereas in Mexico there is complete freedom in the type of charge that the administrators may set. Two countries, El Salvador and the Dominican Republic, have set caps on charges. In El Salvador the total charge for commissions including insur- ance was limited by law to 3 percent of wages. In the Dominican Repub- lic the charge for commissions is set at 0.5 percentage points out of the to- tal contribution of 10 percent. Each pension fund administrator can manage only one fund, except in Chile, where each administrator offers five funds with different risk­re- turn characteristics. In Mexico the pension legislation contemplates mul- tiple funds, but as yet only one is permitted. Legislation passed in March 2003 but not yet implemented will allow the fund managers in Peru to of- fer multiple funds. All countries restrict the frequency of switching between providers. In Argentina, Colombia, El Salvador, and Uruguay two annual switches are permitted. Mexico permits only one switch annually. Chile and Peru have not established legal restrictions, but the switching procedure actually al- lows at most one per year. In Bolivia switching between the two fund man- agers is currently not permitted. The balance in affiliates' individual retirement accounts reflects changes in the market value of the financial assets in which the new funds are invested.12 Hence, during the fund accumulation stage (i.e., before re- tirement) both investment and longevity risk are borne fully by the indi- vidual.13 Moreover, the asset allocation of pension funds is constrained by quantitative investment limits, and in some countries (Argentina, Chile, Colombia, El Salvador, Peru, and Uruguay) the performance of each fund cannot stray too far from industry averages. As a result, the funds offer similar risk­return trade-offs. The retirement income that workers obtain from the mandatory funded pension pillar is protected by a minimum pension guarantee in those coun- tries where the PAYG pension is being phased out (Bolivia, Chile, Do- minican Republic, El Salvador, Mexico, and Nicaragua). The minimum pension guarantee is set at a level close to that of the minimum wage. In Mexico contributors in the old system at the moment of the reform were offered an additional guarantee that their retirement benefit from the new system would be no lower than that which would have been received in the old PAYG system. Workers in Colombia are also protected by a min- imum pension guarantee, regardless of whether they choose the PAYG plan or individual accounts. In Argentina, Costa Rica, and Uruguay there is no minimum pension guarantee in the funded pillar, but workers' ben- efits are underpinned by an explicit PAYG minimum benefit. Affiliates who choose the PAYG option in Peru are guaranteed a minimum benefit, but those who choose individual accounts are not guaranteed any mini- mum pension.14 34 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Structure and Implementation of the Voluntary Savings Component As mentioned previously, the new funded pillars usually have both a mandatory and a voluntary component. Conceptually, the new pillars can combine both second- and third-pillar features. As shown in table 2.4, workers may make additional, voluntary contributions to the individual accounts managed by the pension fund administrators. The voluntary component of the new funded pillars has a similarly restrictive design to the mandatory component in all countries (except in Chile since early 2002; see below). In some countries, such as El Salvador, these voluntary contributions must be deposited in the same fund as the mandatory contributions; in others, such as Colombia and Mexico, there is a separate fund for volun- tary contributions that is subject to a more flexible regulatory regime. In Colombia fiduciary societies are also able to manage these voluntary re- tirement savings. In October 2002 Chile became the first Latin American country to liberalize the market for tax-preferred voluntary retirement savings, by permitting other financial companies to act as pension plan providers and to offer products that could substitute for individual retire- ment accounts. In order to benefit from tax incentives on voluntary savings, deposits must normally be left in the individual account until retirement. There are two exceptions to this restriction: in Colombia and Mexico funds de- posited on a voluntary basis can be cashed out at any time with six months' notice. Chile also recently modified pension legislation permitting distributions at any time, although they are subject to a tax penalty. In addition to voluntary contributions to the funded system, workers in Latin American countries can save in a variety of financial instruments. In some Latin American countries employers can also set up pension plans for their employees. With the exception of occupational plans in Costa Rica, all these instruments are subject to a much less advantageous tax treatment than is the voluntary funded system. Conclusion Proponents of the multipillar reform model claim that the new systems dis- tribute and diversify the risks to retirement income more efficiently than do pure PAYG systems (World Bank 1994). Instead of government pri- marily bearing the risk as in a single-pillar system, the multipillar ap- proach's mix of government guarantees, mandated individual savings, and voluntary pension arrangements--including those between employers and their workers--spread the demographic (longevity), macroeconomic STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 35 Table 2.4 Voluntary Funded Pillars in Latin America Country Voluntary tax-favored savings vehicle Argentina Individuals and their employers can make voluntary contribu- tions to their individual pension fund accounts that cannot be cashed out until retirement. The tax treatment for these voluntary savings is the same as for mandatory savings. Bolivia Individuals and their employers can make voluntary contributions to their individual pension fund accounts that cannot be cashed out until retirement. The tax treatment for these voluntary savings is the same as for mandatory savings. Chile Until October 2002 individuals and their employers could make tax-favored (same treatment as mandatory) voluntary contributions to their individual pension fund accounts that could not be cashed out until retirement (Cuenta 1). After October 2002 these contributions could be deposited in any registered pension plan offered by AFPs, banks, or insurance companies. The funds can now be cashed out at any time but are subject to a 10 percent tax penalty. In addition, individuals can make voluntary deposits to the so-called voluntary savings accounts (Cuenta de Ahorro Voluntario [CAV], also known as Cuenta 2) that are also managed by the pension fund administrators. Individuals can withdraw funds from these accounts up to four times a year. There are some tax incentives but fewer than for the Cuenta 1. The CAVs have not been affected by the reform that took place in 2001. Colombia Individuals and their employers can make voluntary contributions to their individual pension fund accounts up to 30 percent of their salary. The tax treatment for these voluntary savings is the same as for mandatory savings. The funds can be cashed out after five years at any time with at least a six-month notice. The voluntary pension funds can be administered by life insurance companies and bank trusts as well as by the pension fund administrators. Costa Rica Individuals and their employers can make voluntary contributions to their individual pension fund accounts. Employer and employee contributions are tax deductible and are not subject to social charges of up to 10 percent of the employee's salary. Tax penalties apply if the accumulated balance is cashed out before retirement. Partial withdrawal cannot be more than 50 percent of the fund. Total withdrawal is only permitted after 66 months of contributions and payment of tax. (Table continues on the following page.) 36 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 2.4 (continued) Country Voluntary tax-favored savings vehicle El Salvador Individuals and their employers can make voluntary contri- butions to their individual pension fund accounts that cannot be cashed out until retirement. The tax treatment for these voluntary savings is the same as for mandatory savings. Contributions are only tax deductible up to 10 percent of salary. Mexico Individuals and their employers can make voluntary contributions to their individual pension fund accounts. The tax treatment for these voluntary savings is the same as for mandatory savings. The funds can be cashed out at any time with at least a six-month notice. Peru Individuals and their employers can make voluntary contributions to their individual pension fund accounts that cannot be cashed out until retirement. The tax treatment for these voluntary savings is the same as for mandatory savings. Individuals who have been affiliated with a plan for more than five years (or are older than 50) and their employers can also deposit additional voluntary contributions into their individual pension fund accounts that can be cashed out before retirement. Uruguay Individuals and their employers can make voluntary contri- butions to their individual pension fund accounts that cannot be cashed out until retirement. The tax treatment for these voluntary savings is the same as for mandatory savings. Source: Relevant pension legislation for each country. (inflation and recessions), and investment (low or negative returns) risks to retirement income. Although not yet optimal, the diversification of risk in these multipillar systems is widely regarded as an improvement over that which prevailed under the single-pillar, defined-benefit PAYG systems. The new multipillar approach to providing retirement income security was expected to have both direct and secondary benefits. The downsized public pillar would provide a more fiscally sustainable form of basic in- come protection against poverty in old age and would correct regressive transfers that prevailed under the single-pillar defined-benefit PAYG sys- tems, whereas the introduction of explicitly defined mandatory and vol- untary private pillars would have positive medium-term effects on the labor market (a more efficient allocation of labor and a greater coverage of formal income security) and the development of the financial sector. Combined, the multipillar approach was presented as a new package of social security policy that would protect old people through a better dis- STRUCTURAL REFORMS TO SOCIAL SECURITY IN LATIN AMERICA 37 tribution of risks to retirement income and the promotion of economic growth (World Bank 1994). In the chapters that follow we take stock of developments in each of these areas--fiscal sustainability of public pen- sion promises, development of capital markets and the financial sector, and improved equity and efficiency in the labor market leading to an ex- tension of coverage. We examine each of these promises to the extent that is possible, keeping in mind that a definitive evaluation cannot be made of what are still relatively young reformed pension systems, but exploiting all the tools available to provide guidance to policymakers that some may find valuable. Our focus is on the countries presented in table 2.1, which have instituted reforms; we pay less attention to those in table 2.2, whose reforms are very recent or have yet to be implemented. Notes 1. Along with old-age pension benefits, structural reforms to social security systems in the last 20 years in Latin America have dramatically affected the provi- sion of disability and life insurance and, in some cases, workplace injury insurance. 2. This section summarizes only the most salient aspects of the reformed pen- sion systems in countries where the new systems have been fully implemented. It is not intended as an exhaustive description of the multipillar model in each country. We have made every attempt to keep up to date with developments, but new re- form legislation is being debated and has even been passed in several countries in the region. For excellent, detailed reviews of each of the reformed systems, see Queisser (1998a) and Devesa-Carpio and Vidal-Meliá (2002). 3. These reforms, although passed into law, have yet to be implemented in Ecuador and Nicaragua. Several articles in Ecuador's reform are disputed as un- constitutional. 4. In this volume we categorize "pillars" by their objective rather than by whom they are administered (the public or private sector), by how benefits are structured (final salary benefit formula or defined contributions), or by their fi- nancing mechanism (PAYG or full funding). Thus we use the term "first pillar" or "pillar one" to refer to the part of a pension system intended to keep elderly peo- ple out of poverty; the term "second pillar" or "pillar two" to refer to the part in- tended to help individuals smooth consumption over their life cycle (i.e., to prevent a dramatic fall in income at the time of retirement); and the term "third pillar" or "pillar three" to refer to the institutions available on a voluntary basis for workers to increase their retirement income. We discuss the emerging distinction between pillar one and "pillar zero" in chapter 9. 5. Although Brazil introduced significant reforms to its retirement security regime for workers in the private sector, its reform does not qualify as "structural" among pension specialists because it did not introduce a system of mandated pri- vate saving. Brazil's reform is referred to as "parametric" because reformers ad- justed contribution and benefit parameters within a PAYG financing framework (although it required changes to the country's constitution). 6. We take up the discussion of pension programs designed to prevent poverty and the emerging distinction between the contributory pillar one and the noncon- tributory pillar zero in chapter 9. 7. In these four countries some sort of noncontributory benefit targeted to the elderly poor population also exists. The issue is taken up again in chapter 9. 38 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA 8. Peru is exceptional in the region for not providing a poverty prevention pension to the majority of workers who affiliated with the new system of private accounts. Although a first pillar minimum pension guarantee still exists for work- ers affiliated with the reformed PAYG system, and a guarantee similar to that in Chile was put in place in July 2002 for older affiliates of the private system, the ma- jority of workers covered by a formal retirement security system are not covered against poverty in old age. 9. The high rate of affiliation with the funded system is partly because work- ers who do not express a choice are assigned automatically to the funded system. 10. Administradoras de fondos de pensiones or AFPs in Chile, Peru, Colombia, Bolivia, and El Salvador; otherwise named in the remainder of countries that man- date private savings for retirement. 11. In Mexico and Costa Rica disability pensions are still paid by the public pillar. 12. Market valuation is not always possible because most securities, especially those issued by the private sector, are thinly traded. Regulators have developed mechanisms to proxy the value of less liquid securities. Argentina allows pension funds to value part of their government bond portfolio at book prices. Since 1998 a maximum of 30 percent of the total pension fund portfolio can be valued in this way. 13. It is interesting that while workers are fully exposed to investment and longevity risks over the accumulation stage, they are fully insured against the risks of disability and death. The premiums for these policies are paid by the pension fund administrators to private insurance companies, except in Mexico, where the Social Security Institute has retained the monopoly on these services. 14. The poverty prevention pillar, which can take several different institutional forms, including those described briefly here, is discussed at length in chapter 9. 3 The Fiscal Sustainability of Public Pension Promises in Latin America PENSION DEBT SERVICE IS OFTEN one of the largest items in government budgets. Governments' inability to meet growing pension liabilities-- implied in the benefit promises of single-pillar PAYG systems, and often unaccounted for on public sector balance sheets--can be a source of pol- icy risk to old-age income security and is usually the driving force (and a political selling point) of structural reforms (Holzmann, Palacios, and Zviniene 2001; Holzmann 1998). Just as with the pure PAYG institutions that were replaced, the fiscal sustainability of public pension promises after the introduction of the multipillar model with a large funded com- ponent can determine the credibility of the reformed pension system. The question therefore is this: have structural reforms made governments' re- maining public pension promises more fiscally sustainable? This chapter presents the results of simulation analysis of the likely medium- and long-term fiscal outcomes of structural reforms to retire- ment security systems in Latin America (as explained in Zviniene and Packard 2002). Although several studies conducted both prior to and since reforms have presented the simulated fiscal impact of the shift to multipillar systems with individual accounts, rarely do existing studies ex- tend beyond a single country case. Zviniene and Packard evaluated the likely fiscal impact of very different reforms in a group of diverse coun- tries, using a uniform set of indicators and applying a single generic simu- lation model, the World Bank's Pension Reform Options Simulation Toolkit (PROST).1 The results of PROST simulations are followed by a note of caution on just what simulations can and cannot show and how the simulated cost of reforms can diverge dramatically from actual transition costs. Moving beyond the fiscal impact of reform, we discuss the potential links between 39 40 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA structural reform and economic growth. Furthermore, we present a dis- cussion of wider macroeconomic risks that arise with structural reforms, and how these can determine the sustainability of long-term pension promises, presented in Fiess (2003). We find that although remaining public pension promises are fiscally more sustainable after structural reforms, the broader macroeconomic im- pact of making a portion of implicit pension liabilities explicit with the transition to private individual accounts is uncertain. Simulated Fiscal Impact of Structural Reforms Figure 3.1 shows the value of governments' implicit pension promises in each of the countries that undertook structural reforms to their retirement security systems. The figure also shows one relevant counterfactual--the value of this indicator had there been no reforms.2 Figure 3.1 Simulated Implicit Pension Debts (IPDs), with and without Structural Reforms (Percentage of GDP) IPD in 2001 IPD in 2020 IPD in 2030 IPD in 2050 400 as 350 debt 300 250 pension GDP of 200 implicit 150 percentage 100 50 Government's 0 Reform No Reform No Reform No Reform No Reform No Reform No Reform No Reform No reform reform reform reform reform reform reform reform Chile Peru Colombia Argentina Uruguay Mexico Bolivia EI Salvador Source: Zviniene and Packard (2002). THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 41 The value of pension promises is referred to as implicit pension debt (IPD) and is defined as the present value of the stream of future benefits that a public pension system will have to pay current participants (con- tributors, beneficiaries, and their survivors), according to the defined parameters of the system, to recognize their contributions up to the par- ticular year in question. There are several different concepts and methods for calculating the IPD. The simulations presented here employ a "practi- cal termination liability approach," described in detail in Holzmann, Pala- cios, and Zviniene (2001) and proposed as the best method of calculating the IPD for cross-country comparisons. It comes as little surprise that where reforms phased out earnings- related pensions from the public PAYG pillar--Chile, Mexico, Bolivia, and El Salvador--the simulated cost of governments' pension promises (as a percentage of GDP) falls rapidly (figure 3.1). However, even in countries where an earnings-related PAYG pillar was retained--Argentina, Colom- bia, and Peru--and where explicitly defined first-pillar benefits underpin pensions from a private second pillar--as in Argentina and Uruguay-- reforms are likely to slow the growth of public pension liabilities, as meas- ured by IPD. A decrease in governments' implicit pension promises is to be expected from reforms that partially privatize public pension systems. The simu- lated implicit pension debt falls in the wake of reforms as a portion of these obligations is converted into explicit debt or paid with transfers from the general budget. Thus, although changes in the implicit pension debt reveal the extent of reform and how countries chose to spread the costs of transition from one regime to the next, a better measure of fiscal sustainability is the rate at which the total public debt for pensions is accumulating after reforms, compared with the rate of total debt accumulation had there been no reforms. The total pension debt shown in table 3.1 and figure 3.2 is that financed by government borrowing, and includes (a) the current deficits of remaining PAYG systems (the difference between pension payments and contribution revenues), (b) payments to cover the minimum guaranteed pensions for workers contributing to private individual retirement ac- counts where such guarantees exist, (c) government contributions to either a PAYG regime or individual accounts in countries where these are made explicit in the law, and (d) payment of recognition bonds to honor work- ers' contributions to pre-reform systems. Even an analysis of governments' simulated total pension debts accu- mulated after 2001 and the rate of accumulation (figure 3.2) on the whole show a dramatic improvement in fiscal sustainability brought about by reforms. The simulations show substantial savings from the in- troduction of individual accounts and accompanying reforms. These sav- ings are most apparent in Bolivia, Chile, El Salvador, Peru, and, Uruguay. 42 Table 3.1 Current Pension Deficits (Benefit Expenditure Less Contribution Revenue) Financed by Government Transfers (Percent) Uruguay Argentina Mexico Bolivia Colombia Chile El Salvador Peru No No No No No No No No Reform reform Reform reform Reform reform Reform reform Reform reform Reform reform Reform reform Reform reform 2001 (4.0) (3.4) (2.5) (0.1) (0.5) 0.2 (3.5) 0.8 1.6 (0) (7.2) (0.1) (1.4) (0.8) (0.7) 0.7 2010 (2.6) (2.2) (2.2) 0.1 (0.6) 0 (2.2) 0 1.5 (0.4) (4.6) 0.2 (2.2) (1.0) (0.9) 0.7 2020 (2.1) (2.2) (2.3) (0.3) (0.7) (0.3) (2.1) (1.1) 1.0 (1.5) (3.4) (1.0) (3.2) (1.4) (0.9) 0.4 2030 (2.2) (3.4) (2.8) (1.3) (0.7) (0.8) (2.1) (3.0) (0.7) (3.3) (1.5) (2.7) (2.9) (2.1) (0.9) (0.1) 2040 (2.5) (5.0) (3.6) (2.8) (0.7) (1.5) (1.7) (5.3) (3.4) (5.5) (0.5) (3.9) (2.6) (3.1) (0.8) (1.2) 2050 (2.8) (6.6) (4.4) (4.3) (0.6) (2.3) (0.9) (8.5) (5.4) (7.6) (0.8) (4.0) (0.5) (4.1) (1.0) (2.3) Source: PROST simulations in Zviniene and Packard (2002). THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 43 Figure 3.2 Total Pension Debt (Explicit) Accumulated after 2001, with and without Structural Reforms (Percentage of GDP) 2010 2030 2050 debt, 400 350 explicit 300 GDP of 250 200 pension-related 150 percentage total as100 50 0 Reform No Reform No Reform No Reform No Reform No Reform No Reform No Reform No reform reform reform reform reform reform reform reform Government's Chile Peru Colombia Argentina Uruguay Mexico Bolivia El Salvador Source: Zviniene and Packard (2002). In Argentina, however, the simulations summarized here show a considerable increase in the federal government's total expenditure on pensions in the reform scenario.3 In addition to capturing the loss of con- tribution revenue from workers who switched to individual accounts, these projections capture the increase in PAYG deficits that arose from a policy of lowering employer contributions to the public pillar--a policy introduced after the 1994 reform in an attempt to increase compliance with the mandate for employers and workers to participate in the sys- tem.4 Another critical factor that caused Argentina's total spending on pen- sions to balloon since the reform in 1994 was the federal government's policy of accepting the liabilities of overly generous pension plans for civil servants at the provincial level. Provincial governments that agreed to close their pension regimes to new entrants and to force contributing civil servants to join the national system along with private sector work- ers (either the publicly administered, earnings-related PAYG branch or mandated private individual accounts) transferred the obligation of pay- ing the relatively generous benefits of retired provincial civil servants to 44 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA the federal system. The sudden increase in total pension liabilities combined with revenue reductions to the retirement security system stemming from lower contribution rates, widespread evasion, and a long recession aggravated the fiscal stance of Argentina's multipillar pension system. However, our simulations show only the effect of lower contri- bution rates. What Simulations Can and Cannot Show The results of our simulations presented in the previous section show that, with some notable exceptions, structural reforms with more modest pub- lic pension promises in Latin America are likely to have a beneficial impact on the fiscal sustainability of pension systems. As stressed by Holzmann, Palacios, and Zviniene (2001), however, the sort of cross-country simula- tion analysis presented here has to be interpreted very carefully, and can- not replace careful country-specific analysis of reforms. The indicators of fiscal sustainability presented above are simulations of the fiscal impact of reform laws and, thus (somewhat optimistically), they assume that reforms were implemented correctly and that the new systems are adequately administered. The simulations cannot capture probable administrative difficulties that could cause fiscal costs of reform to balloon unexpectedly. In background papers for this book Escobar (2003) and Fiess (2003) showed how the case of Bolivia is instructive in this regard: the transition proved more costly than initially anticipated. When the Bolivian reform was designed and implemented, insufficient at- tention was paid to the institutions that were to govern the transition from the old to the new system. Although a regulatory body was set up to govern the new private pension funds, the system transition itself was insufficiently regulated and thus invited fraudulent claims and a lax in- terpretation of the rules for the transition. This has contributed to higher than expected transition costs (see box 3.1). Similarly, Mesa-Lago (2000) pointed out that the initial projections of the transition costs of reform in Chile understated the true fiscal costs by more than half. Finally, the simulation results in this chapter are used solely to evalu- ate whether the sustainability of public pension promises has been im- proved by structural reforms in the region. This is to say that we are con- cerned with the reduction of "policy risks" to retirement income. Therefore our analysis is solely focused on whether the reductions of what were unsustainable PAYG pension promises are likely to reduce the risk of governments having to default on workers' public pension rights. The simulations cannot address wider macroeconomic concerns and possible risks that arise when governments convert implicit pension debts into explicit debts with structural reforms to the pension system (box 3.2). THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 45 Box 3.1 Bolivia's Pension Reform: A Transition Considerably More Costly Than Expected In 1996 Bolivia's pure PAYG retirement security system was insolvent and illiquid. A structural pension reform was implemented that termi- nated the old defined benefit system and introduced a new system based on defined contributions paid by the employee and, to a lesser extent, by the employer. As elsewhere in Latin America, the new system is adminis- tered by private fund managers, under supervision of a government regu- latory body. At the time of the reform it was estimated that the transition costs would decline steadily and disappear completely some time after 2037 (Von Gersdorff 1997), but with the benefit of hindsight observers agree that this projection was far too optimistic. In fact, the transition-related cash-flow gap has been steadily increasing from 4 percent of GDP in 1998 to 5 percent of GDP in 2002 (see figure 3B.1 below). The increase of Bolivia's pension-related deficit has been attributed to a series of factors (Revilla 2002; IMF 2003). The government has allowed the law to be loosely interpreted to permit a higher number of early retirees. Some groups that were not initially covered have managed to retire under the old system. The number of fraudulent claims has also been on the rise; estimates indicate that payments equaling half a point of GDP are fraudulent. Furthermore, the pension law introduced index- Figure 3B.1 Cash-Flow Gap in Bolivia as 4.0 deficit GDP of 2.0 percentage Pension-related 0 1998 1999 2000 2001 2002 Actual Projected in 1997 Source: Fiess (2003), based on Von Gersdorff (1997). (Box continues on the following page.) 46 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 3.1 (continued) ation linked to the exchange rate, which has proved very costly. The indexation mechanism was recently reversed. Finally, following social unrest the government introduced a minimum pension of B$850 per month in 2001--nearly twice the minimum salary. In many cases the new minimum pension substantially exceeds original entitlements. When the reform was designed and implemented, insufficient atten- tion was paid to the institutions that were to govern the transition from the old to the new system. Although a regulatory body was set up to gov- ern the new private pension funds, the system transition itself was insuf- ficiently regulated, and that invited fraudulent claims, a lax interpretation of the rules for transition workers, and higher than expected transition costs. Effects of Structural Pension Reform on Economic Growth Several studies have gone beyond fiscal impacts to address the wider macroeconomic effects of structural pension reform. As Barrs (2000) noted, economic growth is central to the viability of any type of pension system because pensions represent claims on the level of future (rather than current) output. Theory suggests that pension reform can lead to in- creased economic growth through three principal channels: (1) stimula- tion of savings and investment; (2) labor markets, by raising employment and labor productivity; and (3) capital market development leading to more efficient resource allocation and enhanced total factor productivity. Yet there is much disagreement over whether pension reform does indeed increase growth, both in theory and practice. Although a complete treat- ment of the wider macroeconomic effects of pension reform is beyond the scope of this book, in this section we summarize the key issues. Proponents of the view that structural pension reform increases eco- nomic growth argue that the effect occurs through various channels. First, mobilization of savings through second- and third-pillar contributions raises the aggregate savings rate, which leads to higher investment and output. Second, reductions in payroll taxes lead to increased employment and a shift of workers to the formal sector, raising both labor supply and labor productivity. Third, the savings channeled into pension funds through the second and third pillars stimulates capital market develop- ment and financial innovation, which leads to more efficient resource allocation and increased total factor productivity. Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth (World Bank 1994) [emphasis in original] typifies this perspective, THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 47 Box 3.2 Virtuous and Vicious Fiscal Circles: The Cases of Chile and Argentina The often-touted fiscal effects of structural pension reforms can depend in large measure on whether a country embarks on reform from a posi- tion of relative fiscal strength or weakness. Although any well-designed reform will yield fiscal benefits in the long term, countries that under- take reforms in a position of fiscal strength are better able to absorb short-term transition costs. Furthermore, because financial markets can react negatively to the conversion of implicit pension debt to explicit debt, a country's initial fiscal position plays a major role in determining whether structural reforms generate a virtuous or vicious circle of fiscal adjustment. The cases of Chile and Argentina provide contrasting ex- amples of the impact of pension reform in Latin America. In Chile, "authorities deliberately strengthened the fiscal stance for some years before beginning the pension reform" (Holzmann 1997, p. 173). Fiscal surpluses averaged more than 5 percent of GDP in the two years prior to the 1981 reform (figure 3B.2a), and contributed to ensuring that Chile's post-reform fiscal deficits were relatively mild and short-lived. Figure 3B.2a Chile Was Fiscally Strong prior to Reform; Argentina Was Not 6 5 4 (%GDP) 3 balance 2 1 budget 0 5 4 3 2 1 0 ­1 Overall ­2 Years prior to pension reform Chile Argentina Source: World Development Indicators (World Bank 2003). (Box continues on the following page.) 48 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 3.2 (continued) In contrast, Argentina did not substantially bolster its fiscal situation prior to the 1994 reform, despite enjoying several years of strong eco- nomic growth. The conversion of implicit debt to explicit debt that accompanied structural reform in Argentina revealed that the pre-reform fiscal stance "had been worse than shown by the published figures" (Perry and Servén 2003, p. 38). Further payroll tax deductions reduced revenues and worsened pension system deficits (Rofman 2000), con- tributing to growing fiscal deficits (see figure 3B.2b). To be fair, not all of the growing pension deficit was a result of the reform (Perry and Servén 2003), but Argentina's failure to take advantage of strong growth prior to the reform contributed to its post-reform fiscal deterioration. The cases of Chile and Argentina demonstrate why governments should develop a relatively strong fiscal position prior to undertaking structural reforms and they underscore the importance of reducing the implicit debt of unfunded PAYG systems prior to making the debt explicit by shifting to a funded second pillar (Holzmann 1998). Figure 3B.2b Pension System Deficits Contributed Signif- icantly to Deteriorating Fiscal Balance in Argentina 2 1 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 GDP ­1 of ­2 ­3 ­4 ­5 Percentage Consolidated public sector primary balance ­6 Social security balance ­7 Consolidated public sector overall balance ­8 Source: Perry and Servén (2003). THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 49 particularly with respect to the expected growth benefits of the second pillar: "A mandatory multipillar arrangement for old age security helps countries to . . . increase long-term saving, capital market deepening, and growth through the use of full funding and decentralized control in the sec- ond pillar" (pp. 22­23). Do these growth benefits materialize in practice? The most recent and comprehensive attempt to quantify the effect of structural pension reform on growth is a study of Chile by Corbo and Schmidt-Hebbel (2003). Con- trolling for other reforms, the authors used time-series regressions for the period 1981­2001 to estimate separately the impact of pension reform on the capital stock, labor supply, and total factor productivity. They then substituted these estimates in a Cobb-Douglas production function to estimate the overall impact of pension reform on economic growth. Corbo and Schmidt-Hebbel found that pension reform's impact on savings and investment, labor markets, and total factor productivity led to average annual economic growth of 0.49 percent, or almost one-tenth of Chile's average annualized growth of 4.63 percent over the period 1981­2001. The authors noted, however, that although they controlled for other reforms to the Chilean economy over the period, their estimate may capture some of the interactive effects of pension reform with other structural changes. They also noted that the effect of pension reform on growth is likely to decrease to zero over the long term as the system matures and the economy approaches steady-state growth. Although the Corbo and Schmidt-Hebbel study is econometrically sound, it is emblematic of an approach that relies on assumptions about the nature of pension reform and economic growth that are far from cer- tain. Regarding the effect of pension reform on savings, Barr (2000) argued, "There are not one, but three links in the argument that future output will be higher . . . : [that] funding leads to a higher rate of saving than PAYG; that higher saving is translated into more and better invest- ment; and that investment leads to an increase in output. None of the three links necessarily holds." Similarly, Easterly (2001) pointed out that econ- omists have not yet shaken off their adherence to Harrod-Domar growth models, which posit formulaic savings-investment-output relationships that fail to materialize in practice. With respect to labor market efficiency, establishing a link between pension benefits and contributions does appear to improve the incentives to join the formal sector, as we will argue in chapter 5. But as we also ar- gue there, these improved incentives are likely to have little impact on worker behavior in the absence of complementary labor market reforms and other favorable conditions, as the stagnant social security coverage rates in Latin America demonstrate. Finally, the evidence that pension reform contributes to capital market development is tenuous, as we explore in detail in the next chapter. The assumption of increased total factor productivity through pension funds' 50 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA more efficient resource allocation is even more dubious, given the restric- tions imposed on pension fund investment in equities in many countries and their heavy investment in sovereign debt. None of these reservations mean to say that structural pension reform cannot or does not contribute to economic growth--indeed it can, and country-specific analysis may find that it does. The link between pension reform and growth is not automatic, however, but rather is dependent on a number of associated linkages that may or may not materialize. And without denying the fundamental importance of growth, policymakers should remember that the function of a social security system is not to stimulate growth but to prevent poverty and smooth consumption in old age. Broader Risks and Macroeconomic Concerns Raised by Structural Reforms5 Several authors have argued that standard debt sustainability indicators should be enriched with a measure of implicit pension liabilities to pro- vide better performance indicators for fiscal sustainability and solvency (Holzmann, Palacios, and Zviniene 2001). Whereas previous studies (Feldstein and Seligman 1981; Moody's 1998) showed that markets and rating agencies take unfunded pension liabilities of corporations into ac- count when determining share prices and ratings, Truglia (2000, 2002) argued that the situation is entirely different for the impact of unfunded public pension liabilities on sovereign credit risk. Truglia (2000, 2002) stated that to date, net-present-value estimates of implicit pension liabilities have not influenced Moody's sovereign credit risk ratings.6 There are two reasons for this. First, net-present-value cal- culations are highly susceptible to sizable swings depending on relatively small changes in a number of parameters. Second, and more important, al- though net-present-value calculations of future pension liabilities provide a projection of a given scenario, they do not assign a probability that the projection will come true. Assessing fiscal solvency on the grounds of pro- jected implicit pension liabilities alone does not account for the fact that policymakers tend to change the parameters of the present pension system and hence the level of implicit pension liabilities before financing concerns become too pressing. Although a public pension promise is similar to a government bond in the sense that it represents a claim on future income, society treats both claims quite differently. Public pension promises are changed in ways that debt instruments would never be altered. Truglia (2002) pointed out that although no industrial country has defaulted on its debt since World War II, almost every industrial country has adjusted its pension system in ways that changed the original contract, that is, by increasing the retirement THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 51 age, changing the benefit formula, or both. The fact that pension reform is generally not referred to as "pension default" illustrates that society dif- ferentiates between changes in contractual terms of public pension and debt claims. However, the fact that risk-rating agencies do not account for implicit pension liabilities in their country risk ratings does not mean that pension reforms (and in particular the financing of the transition deficit) have no impact on country risk premiums. It is often argued that during the tran- sition period, when implicit debt is made explicit, the market perception of sovereign risk might rise as the observable debt burden increases. To our knowledge no empirical study has analyzed the impact of pension re- forms on country risk. In an attempt to better understand potential links between pension reforms and country risk in Latin America, Fiess (2003) drew from the literature on country risk to develop some guidelines for fu- ture research. The literature suggests a number of determinants for country risk, in- cluding measures of liquidity and solvency, macroeconomic fundamentals, and external shocks (see Edwards 1986; Haque et al. 1996; Barnes and Cline 1997; Eichengreen and Mody 1998; Kamin and von Kleist 1999; Min and Park 2000; Fiess 2003). Within this framework there are two main channels through which pension reforms might affect country risk. First, pension reform can change the level of implicit and explicit liabilities and, as such, potentially affect the perception of solvency. In most cases perception is reality. Sec- ond, it is argued (although not generally accepted) that pension reform positively affects economic growth. As we discussed in the previous sec- tion, the relationship between pension reform and growth stems from an assumed direct effect that reforms may have on growth through savings and capital accumulation, and via an indirect growth effect through the development of capital markets. If establishing a clear empirical link between pension reform and eco- nomic growth is hard, proving a relationship between implicit and explicit pension liabilities and country risk seems equally difficult. To do so one would have to disentangle at least two simultaneous effects that are not di- rectly observable and are likely to affect country risk in opposite direc- tions: (1) an implicit-to-explicit debt conversion is likely to increase coun- try risk if financial markets are myopic or suffer from fiscal illusion7 and if governments are liquidity constrained; and (2) if financial markets value implicit pension liabilities, a structural pension reform that manages to re- duce the level of implicit pension liabilities is likely to be rewarded with a discount on country risk as long-term solvency is improved. Figure 3.3 shows the Emerging Market Bond Index spread, a series of idiosyncratic country risks (Fiess 2003), the Institutional Investor's Coun- try Credit Rating Index, and the debt-to-GDP ratio of Mexico from 1994 to 2000. Mexico's pension reform was fully implemented in July 1997. 52 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 3.3 There Is No Indication That Pension Reform Increased Mexico's Country Risk 50 1,500 45 40 1,000 35 30 500 25 20 0 Jan-94Apr-94Jul-94Oct-94Jan-95Apr-95Jul-95Oct-95Jan-96Apr-96Jul-96Oct-96Jan-97Apr-97Jul-97Oct-97Jan-98Apr-98Jul-98Oct-98Jan-99Apr-99Jul-99Oct-99Jan-00 Debt/GDP Credit rating Pension reform EMBI Country risk EMBI (J. P. Morgan Emerging Market Bond Index) Source: Fiess 2003. Casual observation of the information provided in figure 3.3 suggests that Mexico's pension reform had no direct impact on country risk. However, little visible impact on the indicators in figure 3.3 does not imply that pension reform has no impact on country risk at all. Mexico had one of the lowest implicit pension debts (IPDs) in Latin America be- fore its reform (less than 30 percent of GDP), and the reform only reduced it marginally. As a result, the impact of the reform on country risk may have been less significant than in a country such as El Salvador, which drastically cut its IPD. In Argentina, where both explicit pension debt and (by many measures) implicit pension debt indicators worsened after the re- form, country risk deteriorated to the point where the government was forced to default on its debt. More generally, it is likely that a pension re- form will affect country risk through multiple and highly complex dy- namics, which might even cancel each other out. Understanding these changes is important for countries contemplating multipillar reforms. The first issue concerns financing the transition.8 In theory a govern- ment could pay off its total implicit debt by issuing checks to all transition workers and pensioners who have accrued rights under the old system. This would make all implicit debt immediately explicit at the time of the THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 53 reform. For budgetary reasons, reforming countries generally did not choose this option (James 1999) but rather adopted a mixture of instru- ments to finance the transition deficit to spread the fiscal costs of transi- tion over time.9 As a second issue, reforms can introduce new implicit liabilities. Measuring implicit pension liabilities is a difficult task, and comparing im- plicit pension liabilities pre- and post-reform is not straightforward be- cause counterfactual measures are not observable. Several reforming countries have introduced new implicit liabilities, such as minimum pen- sion guarantees, that could even have raised the level of implicit pension liabilities. In the case of Chile, Schmidt-Hebbel (1999) and Schreiber (2001) found that the costs of minimum pension guarantees are not negli- gible. Mexico's new pension system offers a "life-switch option" instead of recognition bonds to transition workers--that is, it allows transition workers to choose at the time they retire from the system the option (old or new) that will give them the highest level of benefits (Rodriguez 1999). The costs of this life-switch option are likely to be high if market returns are below expectations. As a third issue, investment rules can create captive finance for govern- ment debt. How private pension funds are regulated can also affect coun- try risk. In Latin America most pension funds' investment strategies are affected by quantitative regulations that limit the extent of investment in specific kinds of assets. These restrictions are more severe for equities and foreign securities than for fixed-income securities. Portfolio limits for pen- sion funds can produce a guaranteed market for government bonds and thus help smooth a debt-financed transition.10 Yermo (2002b) pointed out that although pension fund administrators in Latin America have been rel- atively efficient as financial risk managers, the new private second pillars have not been effectively insulated from political interference. Pension funds are effectively used by the governments as captive sources of finance. Government meddling in fund portfolios, as in Argentina in 2001 (see Rofman 2002) and in Bolivia in 2003 (see Escobar 2003), underscores this point. The fourth issue is that one must be able to account for idiosyncratic country risk. Researchers have to disentangle global risk from idiosyn- cratic risk because only the latter is relevant when evaluating the impact of pension reform on country risk. Sovereign bond spreads are often used as a measure of country risk. Bond spreads, however, are affected by both idiosyncratic and global factors. Global factors were primarily responsible for increasing country risk during the Asian (1997) and Russian (1998) crises. Because the crises coincided with the implementation of many pen- sion reforms in Latin America,11 attempts to isolate the impact of pension reform on country risk should be undertaken with some care. As a final issue, the impact of pension reform would have to be isolated from other structural reforms. Pension reforms generally were not carried 54 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA out in isolation. Corbo and Schmidt-Hebbel (2003) pointed out that the 1981 pension reform in Chile was part of a wider structural reform effort that included fiscal adjustment, labor market reform, financial liberaliza- tion, and capital market reforms. The complementarities of these reforms make it extremely difficult to properly isolate the impact of a specific re- form. It is likely that the combined impact on growth or capital market de- velopment was larger than the individual impact of any given reform. Conclusion Although the simulations presented in Zviniene and Packard (2002) showed that structural reforms in Latin America are likely to deliver sub- stantial reductions in public pension liabilities, fiscal sustainability is far from certain. Because of the different contractual nature of pension liabili- ties, a positive impact of pension reform on solvency perceptions is not as obvious as theoretical models claim. Furthermore, pension reforms can create new implicit and explicit liabilities. Although the benefits to eco- nomic growth from pension reform are important, the theoretical links be- tween structural reforms and growth are not straightforward, and empiri- cal evidence is still scarce. Finally, empirical evidence shows that pension reforms can produce severe cash-flow problems in excess of initially pro- jected transition costs and hence seriously constrain public sector liquidity. Notes 1. Even where the multipillar reform models chosen by policymakers in differ- ent countries are very similar, circumstances directly and indirectly related to formal retirement security differ widely, causing difficulties for any attempt at cross-coun- try comparisons. Thus the results of the simulations cannot be used to determine whether Chile's reform was fiscally "more successful" than Peru's or Mexico's, or whether the impact of reforms on equity in Argentina were "greater" than in Colombia. Furthermore, the authors pointed out that because a number of the assumptions imposed are likely to vary from one study to the next, the results can be used only as indicators of the order of magnitude of various statistics and can facilitate only rough comparisons among countries. So the sort of cross-country sim- ulation analysis presented in this chapter (and elsewhere; see Holzmann, Palacios, and Zviniene 2001) has to be interpreted very carefully and cannot replace country- specific analysis of reforms using tailor-made simulation models. 2. To project a "no-reform scenario"--a counterfactual to structural reforms--the current number of beneficiaries, wage distribution of contributors, and the like were used, and the parameters of the old single-pillar PAYG system were applied. 3. Readers should note that our data and assumptions for Argentina were taken prior to the 2001­02 crisis and the devaluation of the peso. An account of the crisis and its impact on the reformed pension system can be found in the Rofman (2002) (see appendix at the end of this volume). THE FISCAL SUSTAINABILITY OF PUBLIC PENSION PROMISES 55 4. The policy of lowering employer contributions was pursued to increase compliance. However, regulations in the product and factor markets that have little to do with the social security system keep the cost of compliance and formalization in Argentina very high. Thus, lowering employer contributions to the public pen- sion system had no substantial impact on lowering evasion and has only served to deepen pension deficits (Rofman 2002). 5. The remainder of this chapter is drawn directly from a background paper prepared by Norbert Fiess in the Office of the Chief Economist for Latin America and the Caribbean Region. The principal authors are grateful for this contribution. Readers can find a more detailed discussion of the issues raised here in the back- ground paper. 6. This does not imply that such will not be the case in the future (Fiess 2003). 7. If financial markets do not suffer from fiscal illusion, a trade-off between implicit and explicit pension liabilities should be of little consequence. 8. A more detailed discussion of transition cost financing can be found in Holzmann (1998), "Financing the Transition to Multipillar," available at http://www.worldbank.org/pensions. 9. There are many different ways to finance transition costs, and countries usually apply a mixture of instruments (James 1999). For example, transition costs can be financed (1) through a reduction of the value of IPD (e.g., by downsizing the old system through reducing benefits and increasing retirement age, or by retaining a public PAYG pillar in the new system), or (2) through special revenue sources (e.g., privatization revenues from public enterprises) or use of general taxation or borrowing (through fiscal adjustment or debt financing). 10. In the longer term, however, portfolio limits can undermine any benefits as- sociated with a fully funded pension system for aggregate savings, economic growth, and capital market development (Srinivas, Whitehouse, and Yermo 2000). 11. That is, when Costa Rica (1996), Uruguay (1996), Bolivia (1997), Mexico (1997), and El Salvador (1998) were conducting reforms. 4 The Financial Benefits of Pension Reform AN IMPORTANT JUSTIFICATION FOR PENSION reform in Latin America has been its expected benefits for capital markets. The growth of pension funds and other institutional investors can help make capital markets more resilient and dynamic. In turn, the development of capital markets can improve the efficiency of resource mobilization and investment in the economy. Deep and liquid domestic capital markets can also help curtail dependency on foreign capital and thus reduce the economy's vulnerabil- ity to external shocks. Levine and Zervos (1998), and Beck and Levine (2001) have found that capital market development has a positive impact on economic growth. The Latin American pension reforms have led to marked changes in the financial sectors of Latin American countries that have introduced manda- tory individual accounts. Foremost among these changes has been the appearance of a new market player--the special-purpose pension fund ad- ministrators who have captured all mandatory (second-pillar) and the bulk of tax-advantaged (third-pillar) pension savings. The Regulation of Mandatory Pension Funds The introduction of privately managed individual accounts created the need to establish regulations governing fund managers and the invest- ments they can make. Regulation of Pension Fund Managers Latin American pension fund administrators (AFPs) are independent legal entities whose exclusive purpose is the management of pension funds.1 The governance of pension fund administrators is subject to a variety of regulations that aim to protect the members from conflicts of interest. 57 58 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Such regulations include a prohibition on transactions between the administrator and its employees and the pension fund. Pension fund ad- ministrators are also banned from purchasing on their own behalf stocks that may be acquired by the fund. In Chile the pension fund administra- tors must have some independent directors whose duty is to guard the in- terest of affiliates. Chilean regulations also set forth a high principle of fiduciary responsibility: AFPs should ensure the adequate profitability and safety of the investment of the funds they manage. They are obliged to reimburse the pension fund for any direct damages they may cause, whether by omission or commission. Regulations also cover the role of pension fund administrators in cor- porate governance. Chilean pension fund administrators are required to attend the shareholder meetings of those companies in which they have acquired stocks for the pension fund, and they must vote in all agreements, including the election of board members. The AFPs cannot vote for board candidates who are related to the majority shareholders or to those who control the company. They are also typically required by the supervisory authority to file reports regarding events or transactions by security issuers that may harm pension fund investments. There are also some regulations that actually limit the extent of collu- sion in collective action by pension fund administrators. In Chile the su- pervisory authority has ruled that "it is entirely contrary to the spirit of the law (D.L. 3.500) for one or more funds to form an association or act in a block in order to exercise their shareholders' rights" (Iglesias 2000, p. 117). Nonetheless, an explicit authorization can be granted to AFPs to act jointly at board elections. In Chile there is also a prohibition against par- ticipating in "or having any bearing on the management of a company," which essentially restricts the influence of AFPs to their participation in shareholder meetings (Iglesias 2000).2 In Peru pension fund administra- tors are not required to attend or vote in shareholders' meetings, and they face the same prohibition as in Chile with respect to their involvement in the administration of the companies in which they invest. Rules governing disclosure to plan members, external audit, and re- porting to the supervisory authority are also applied widely and effectively in Latin American countries. The supervisors oversee the operations of both the administrators and the pension funds they manage. Potential ad- ministrators wishing to enter the market must apply for a license from the supervisory authority. The companies must comply with the minimum capital requirements established in the legislation. Pension fund managers in some countries (such as Argentina, Chile, Colombia, El Salvador, and Uruguay) must also guarantee a certain mini- mum return on the pension fund (usually relative to the industry average) and must maintain a capital reserve to meet any shortfalls in the fund's rate of return relative to the minimum. This reserve must be invested in the same way as the pension fund. THE FINANCIAL BENEFITS OF PENSION REFORM 59 Some countries have imposed limits on the share of the market that pension fund administrators may have. In Mexico, for example, the re- tirement fund administrators cannot control more than 17 percent of the pension fund market until 2002. Since that date they have been allowed to have up to 20 percent of the market. The law, however, does not specify whether the market share measure is assets under management or number of affiliates. Restrictions on Pension Fund Investment The investment of pension funds is subject to a comprehensive prudential regulatory framework. In each country that has reformed, all liquid fi- nancial assets bought by pension funds must be traded in secondary mar- kets and valued at market prices.3 For the less-liquid assets the supervisory authorities of some countries, such as Mexico, set a valuation mechanism based on historical prices and the valuation of related securities.4 All countries that permit investment in securities issued by private sec- tor companies and traded in regulated, secondary markets have also introduced new systems for risk rating (Colombia is the only exception). Investment limits also include limits by issuer and ownership concentra- tion. For example, a Chilean pension fund cannot own more than 7 per- cent or invest more than 5 percent of fund assets in any given company's stock. Other countries also impose limits on the percentage of a company stock that pension funds can hold (5 percent in Argentina and El Salvador, 10 percent in Colombia and Uruguay, and 15 percent in Peru). Possible conflicts of interest between pension fund managers and re- lated entities arising from the investment of pension funds are also strongly regulated. All countries set low limits on investment in securities of issuers related to the pension fund managers. In Chile and Mexico the limit is set at 5 percent of the pension fund assets. Pension funds may not be invested in assets issued or guaranteed by members (or relatives) of the governing body of the pension fund administrator, or by managers or owners of authorized entities. As shown in table 4.1, there are also limits by asset class. These limits are less justifiable from a prudential perspective except for the fact that capital markets may not be adequately regulated. Governments may also wish to avoid high-risk portfolios to the extent that they offer minimum pension guarantees.5 In Argentina, Bolivia, Colombia, Costa Rica, and Uruguay the regulatory framework does not distinguish between domestic securities issued in local and in foreign currencies. In Chile the limit on for- eign securities applies also to foreign currency­denominated securities issued by local entities. In Peru the limits on private sector securities apply equally to local and foreign currency­denominated assets. Pension funds were only allowed to invest in dollar-denominated government bonds (Brady bonds) in July 1998. In Mexico until December 2001, pension 60 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA funds could invest up to 10 percent of their assets in dollar-, euro-, and yen-denominated federal government and Central Bank securities. Since then, private securities have also become eligible. Portfolio limits have been relaxed in some countries, such as Chile, as the respective regulatory and supervisory frameworks were established or reformed to ensure a proper functioning of the capital markets. Hence, for example, equities investment was permitted in Chile in 1985, five years after the passing of the securities law. Investment in foreign securi- ties was first permitted five years later following legal reforms that, for example, permitted companies to issue American depository receipts (ADRs) for the first time.6 Other countries are also gradually liberalizing their investment regimen. Peru first permitted investment overseas in 2001. At the beginning of 2002 the Bolivian supervisor set out the regu- latory framework that permits foreign investment for the first time (the legislated limit was set at 50 percent at the time of the reform). Mexico recently eliminated a rule that required pension funds to invest at least 65 percent of their assets in financial instruments with a maturity of fewer than 182 days. The investment floors present in some countries are a greater source of distortions (see table 4.2). In Bolivia and Uruguay the goal of investment floors on government bonds was to ease the fiscal cost of the transition to a funded pension system. In Mexico the requirement to invest in infla- tion-indexed securities can also be justified as a measure to ensure a sta- ble real rate of return on the funds. Such conservative investment helps the government manage its contingent liability as a result of the retire- ment benefit guarantee offered to transition workers. In Costa Rica the floor is applied to mortgage securities, a decision that appears to be jus- tified by the government's desire to promote housing finance while offer- ing to pension funds an attractive long-term investment. Despite the pos- itive objectives of some of these floors, governments must take into account possible distortions to the diversification and performance of pension funds. In all countries, custody of pension assets must be carried out by enti- ties independent of the pension fund administrator. In Chile, until 1994 all assets were safeguarded by the Central Bank. As a result of the Capital Markets Law of 1994 private companies offering security deposit services can also act as custodians of pension funds. Rapid Growth in Pension Savings Thanks to its privileged position, the new pensions industry is beginning to dominate the financial system. As shown in table 4.3, in Chile, the ear- liest reformer, pension assets directed by the pension fund managers were more than 50 percent of GDP. Asset growth in other countries that have Table 4.1 Portfolio Ceilings by Main Asset Classes (December 2002) (Percent) Government Financial Corporate Investment Foreign securities institutions Stocks bonds funds securitiesa Argentina 80 30 70 40 30 20 Bolivia 100 30­50 20­40 30­45 5­15 10­50 Chileb 50 50 30 40 20 20 Colombia 80 42 30 30 5 10 Costa Rica 85c 100d 5 70e 10 0f El Salvador 50 40 20 20 20 0 Mexico 100 10a 0 100g 0 0 Peru 30 40h 35 40i 15 9 Uruguay 60 30 25j 25j 25 0 aMaximum of 250,000 in local currency (Mexican pesos) and US$25,000 in foreign currency plus the required amount for currency matching. bThis information refers to Fund C (the one with the average allocation to equities). The limit on foreign investment applies to the total of all funds administered by an AFP. cDeclining to 50 percent in 2009. dMaximum of 20 percent for public financial institutions. eThe 70 percent limit applies to AAA-rated bonds; 50 percent to AA-rated corporate bonds; 20 percent to A-rated corporate bonds. fForeign investment is currently not permitted by the supervisor, but the law set a ceiling of 25 percent, which can be increased to 50 percent if domestic returns are low. gNo limit (AAA-rated corporate bond); 35 percent (AA-rated corporate bonds); 5 percent (A-rated corporate bonds). hJoint limit between deposits, bonds, and promissory notes. iCorporate bonds of nonfinancial institutions. jJoint limit. 61 Source: Pension fund supervisors in the countries described. 62 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 4.2 Portfolio Floors by Country Country Description of regulations Bolivia The two pension funds together must invest a minimum of US$180 million annually between 1998 and 2013. Costa Rica At least 15 percent of the pension funds' assets must be invested in mortgage securities, with a minimum return no less than that of the mandatory complementary pension system. Mexico At least 51 percent of the pension funds' assets must be invested in inflation-indexed securities. Until December 2001 only federal government and central bank securities were eligible for investment under this rule. Since then, state and private securities that are indexed to inflation are also eligible. Uruguay Pension funds must invest between 40 percent and 60 percent of their assets in government securities. Source: Pension fund supervisors in the countries described. Table 4.3 Assets Held by Pension Funds Have Doubled as a Percentage of GDP (December 1998­December 2002) (Percent) Country 1998 1999 2000 2001 2002 Chile 40.3 53.3 59.8 55.0 55.8 Peru 2.5 4.1 5.4 6.6 8.1 Colombia 2.7 4.2 5.5 7.0 7.7 Argentina 3.3 5.9 7.1 7.4 11.3 Uruguay 1.3 2.8 3.9 6.1 9.3 Mexico 2.7 2.3 3.0 4.3 5.3 Bolivia 3.9 7.0 10.8 11.0 15.5 El Salvador 0.4 1.7 3.6 5.5 7.4 Costa Rica 0.0 0.0 0.0 0.1 0.9 Average 7.1 10.2 12.4 11.4 13.5 Note: Assets held by the Bolivian capitalization fund are not included. Source: AIOS 2002, Colombian Banking Superintendency. undergone pension reform has also been rapid. In Bolivia as in Chile the earnings-related PAYG pillar is being phased out and AFP-managed pension assets quadrupled as a share of GDP, from 3.9 percent in 1998 to 15.5 percent of GDP in 2002. In Latin America as a whole this ratio has doubled in just five years. In addition, in their role as providers of disability, survivors', and longevity insurance in the new systems, insurance companies have accu- mulated a significant amount of pension assets. However, because most THE FINANCIAL BENEFITS OF PENSION REFORM 63 systems are still primarily in their accumulation stage, the importance of insurance companies as financial market players is dwarfed by that of the pension funds. By December 2002 the assets held by pension funds were more than three times those of insurance companies. The dominance of pension funds in the domestic capital markets is demonstrated by the extent of capitalization of the various markets in which they invest. Throughout Latin America pension funds are becoming particularly important investors in government debt (see figure 4.1). Their presence in private sector securities markets is generally less significant, except in Chile and Peru. In Chile pension funds owned more than half of the total stock of mortgage and corporate bonds in December 2002. In Peru pension funds also owned more than half of the total stock of cor- porate bonds in circulation (see Yermo 2002a). The growth of pension funds is turning these new institutional investors into key players in the financial system, underscoring the importance of Figure 4.1 Pension Funds Are Major Investors in Government Debt 80 70 60 50 government total 40 of outstanding 30 debt 20 Percentage 10 0 Chile Peru Rica Bolivia Colombia Argentina Uruguay Mexico Salvador El Costa December 1999 December 2000 December 2001 December 2002 Source: AIOS (2002). 64 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 4.1 The Importance of Concomitant Reforms in the Financial System In addition to laying the basis for the new pension fund industry, Latin American governments have been active in reforming other aspects of the financial system. Three main reforms can be identified, some of which have been at least partly driven by the need to ensure the smooth func- tioning of the private pension system. Concomitant Reform 1: Modernization of Financial Market Infrastructure Key elements of the financial infrastructure, such as risk-rating, custodial, and brokerage services, and trading and settlement systems should be modernized before the introduction of pension funds. Although such functions are essential for every sector of the financial system, the reform of the pension systems has brought home the need for improving many of them. As mentioned above, the development of the risk-rating industry in Latin America is intrinsically related to the establishment of the pension fund industry. Risk rating has also been extended to all issuers of publicly traded instruments in Chile, not just those that receive investments from the pension funds. Another important pension reform­related improve- ment in the financial infrastructure is the modernization of trading sys- tems in stock exchanges. Pension fund portfolios are valued daily in Latin American countries. This has necessitated a revamping of the technology used by financial institutions to value their assets. Some hindrances to market trading remain in several countries, however. In Mexico, for exam- ple, trades carried out through the electronic system cannot be executed from outside the exchange. In all Latin American countries, depository and custodian services must be provided by a financial institution inde- pendent of the pension fund administrator. This regulation has helped de- velop the custodial services industry. Clearing and settlement systems are also still to be modernized in most Latin American countries. In Chile a public company was created in 1989 to deal with all clearing and settlement of securities transactions. The company is owned by the Santiago Stock Exchange and the main finan- cial institutions and intermediaries. Settlement of instruments issued by financial institutions takes place the same day as the transaction, those of fixed income securities the day after, and stocks trading two days after the transaction. The system contrasts with that in place in Peru, where settle- ment is still concentrated in one medium-size bank, which could represent a significant systemic risk. Concomitant Reform 2: Regulatory Reform Within the Financial Sector The enforcement of financial contracts through regulations and effective supervision is a key institutional feature that enables the development of THE FINANCIAL BENEFITS OF PENSION REFORM 65 Box 4.1 (continued) financial markets. La Porta et al. (1998) argued that the efficiency of the financial system is based on the extent to which contracts are defined-- and made more or less effective--by legal rights and enforcement mech- anisms. Levine et al. (2000) provided evidence that the quality of legal rights in financial systems can explain economic growth, whereas the rel- ative role of banks versus markets cannot. Reforms in securities markets have often been engineered with the goal of improving the functioning of private pension systems. For exam- ple, the 1994 reform to capital markets in Chile had as its main objective increasing the flexibility of the investments by pension funds and life in- surance companies. This reform also improved the regulation and super- vision of conflicts of interest (including insider trading). The reforms to the Capital Markets Law proposed in Peru in 2001 (and now in the process of being approved) also contemplated significant changes, such as the introduction of clear fiduciary responsibilities for asset managers. Some efforts have also been made to improve shareholders' rights. In a survey by La Porta et al., Chile got as high a score as Organisation for Economic Co-operation and Development member countries. Minority shareholder rights have been further strengthened in this country through the latest reform to the Capital Markets Law. In Peru amend- ments to that law and the law protecting the rights of minority share- holders were also proposed in 2001. Some of the measures proposed in Peru included permitting proxy voting by mail, stricter disclosure requirements for listed companies, and the promotion of independent directors. A new code of best practices has been proposed recently by the Mexi- can Stock Exchange to address some deficiencies in corporate gover- nance. Some weaknesses also exist in the Argentine Capital Markets Law. A project recently presented to the Argentine Congress is expected to im- prove corporate governance practices in accordance with international standards (including the introduction of a minimum number of inde- pendent directors). Reforms are still needed to strengthen creditors' rights in all Latin American countries. La Porta et al. showed some deficiencies in this area, particularly in Colombia, Mexico, and Peru, which had the lowest score in the region. In Mexico enforcement of creditors' rights still suffers from several deficiencies, especially in bankruptcy and collateral laws that weaken creditors and undermine market discipline. A bankrupt business can enter into reorganization without requiring creditors' consent. Se- cured creditors are not necessarily paid first; the claims of various social constituencies precede them. Some Latin American countries also have recently reformed the regu- lation and supervision of their insurance industries, but in general much (Box continues on the following page.) 66 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 4.1 (continued) work remains to be done. These reforms are of paramount importance given the role of the insurance industry as providers of disability and sur- vivor insurance and of retirement annuities. A starting point for many countries is to ensure compliance with international standards (in partic- ular those established by the International Association of Insurance Su- pervisors). Reforms to the banking system, although largely unrelated to the pen- sion system per se, have also been enacted throughout Latin America since the debt crisis of the early 1980s. In Chile in 1986 a new banking law was approved that required significant diversification, closer asset­ liability matching, and limits on related party transactions. Banks were also prohibited from holding equity, with a few exceptions. At the end of 1997 new amendments introduced to the banking law led to the adoption of the Basel recommendations on capital requirements and a new licens- ing process. The mutual fund sector is the laggard in the financial system, handi- capped by regulatory and supervisory deficiencies, such as insufficient control of conflicts of interest. In Mexico the establishment of the pri- vate pension industry has made more patent the deficiencies of the regu- lation and supervision of mutual funds. As a result, the government is coming to grips with the problem and is developing a new regulatory framework. This framework would include stricter disclosure rules and fiduciary standards in line with those applied to the pension fund ad- ministrators. Concomitant Reform 3: Tax Reform The tax treatments of different forms of savings and investment are key determinants of the evolution of any financial system. In many countries debt is preferable to equity as a source of financing because of its less onerous tax treatment. In Chile, where the benefits of pension reform are most pronounced, a significant tax reform also took place. In 1984 the tax rate for reinvested profits was reduced from 46 percent to 10 percent, and taxes for distributed profits of open corporations were reduced from 43.3 percent to 31.5 percent. Uthoff (1998) argued that the tax reform explains much of the increase in savings observed in Chile over the past two decades. The liberal Chilean tax reform contrasts with the Mexican tax structure. In Mexico capital gains tax must be paid on private sector securities but not on government securities. The difference in after-tax re- turns is sufficiently high to distort the investment strategies of pension funds and other institutional investors toward government securities. Source: Yermo (2002a). THE FINANCIAL BENEFITS OF PENSION REFORM 67 concomitant reforms in the financial system (box 4.1). Because of the high investment in government securities and the banking sector, how- ever, direct pension fund financing to the private sector through bonds and equities is still relatively low compared with bank credit. Even in Chile total direct investment in the nonfinancial private sector repre- sented less than 12 percent of GDP in December 2002. Bank credit to the private sector, on the other hand, was close to 67 percent of GDP at that time. As shown in table 4.4, direct pension fund investment in the private sec- tor is even lower in other Latin American countries (less than 20 percent of total pension fund assets), the only exception being Peru (46 percent of total assets). In Costa Rica, El Salvador, and Uruguay pension funds pro- vide little direct financing to the nonfinancial private sector (less than 5 percent of total assets). How the Pension Fund Industry Operates To understand the role of pension funds in capital market development it is important to consider not just the size of their portfolios but also how these portfolios are managed. Pension fund administrators can also play an important role in regulatory reform and in financial innovation. In this section we take a closer look at the governance of the pension fund industry. Table 4.4 Pension Funds Invest Mainly in Debt of Governments and Financial Institutions (December 2002) (Percentage of portfolio shares) Government Financial Corporate Investment Foreign Country securities institutions bonds Equities funds securities Other Argentina 76.7 2.6 1.1 6.5 1.8 8.9 2.4 Bolivia 69.1 14.7 13.4 0.0 0.0 1.3 1.5 Chile 30.0 34.2 7.2 9.9 2.5 16.2 0.1 Colombia 49.4 26.6 16.6 2.9 0.0 4.5 0.0 Costa Rica 90.1 5.3 4.6 0.0 0.0 0.0 0.0 El Salvador 84.7 14.4 0.5 0.5 0.0 0.0 0.0 Mexico 83.1 2.1 14.8 0.0 0.0 0.0 0.0 Peru 13.0 33.2 13.1 31.2 0.8 7.2 1.6 Uruguay 55.5 39.6 4.3 0.0 0.0 0.0 0.5 Note: Information for Colombia refers only to the mandatory pension fund system. Source: AIOS 2002, FIAP 2002 (data for Colombia). 68 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Decisionmaking in an Oligopolistic Industry Latin American pension fund administrators do not have to meet any con- tingent liabilities on the funds they manage, other than ensuring that the funds' performance lies within the stipulated bands. Their investment ob- jectives are therefore similar to those of mutual funds. In practice, how- ever, there are important differences between the investment practices of pension funds and the practices of mutual funds. The main reasons for these differences are, first, investment regulations, which tend to be more strict for pension funds. Second, Latin American pension fund adminis- trators have a captive market where individuals' contributions are re- tained until they retire. Third, the extent of switching between pension fund administrators is heavily regulated. Finally, in all countries except Chile individuals have no investment choice: under the guidance of regu- lators the administrators decide the asset allocation. Some of these regulations (single fund, restrictions on switching, investment rules, lack of individual choice), combined with economies of scale and the size of markets have created the perfect conditions for a highly concentrated industry where the only pressure to perform comes from other regulations (performance rules, fees regulations). The most ex- treme case of concentration is in Bolivia, where two administrators were assigned regional monopolies in the market. Even in the other countries the extent of concentration is very high (see table 4.5). The herding instinct among pension fund managers is particularly wor- rying in the context of an industry that is increasingly the dominant in- Table 4.5 For the Average Latin American Country, the Two Largest Institutions Control Two-Thirds of the Pension Funds (December 2002) Number of Market concentration Country administrators (two largest institutions) Argentina 12 42.7 Bolivia 2 100.0 Chile 7 55.0 Colombia 6 49.8 Costa Rica 9 70.7 El Salvador 3 99.7 Mexico 11 45.0 Peru 4 59.2 Uruguay 4 74.7 Average 6 66.3 Note: Assets held by the Bolivian capitalization fund are not included. Information for Colombia refers only to the mandatory pension fund system. Source: AIOS 2002, Colombian Banking Superintendency. THE FINANCIAL BENEFITS OF PENSION REFORM 69 vestor in bond markets. To the extent that a few pension fund managers who invest in a similar way dominate capital markets, it is unlikely that market liquidity will grow to the levels observed in Organisation for Eco- nomic Co-operation and Development (OECD) member countries. The increasing process of concentration in the pension fund management in- dustry, although efficient with respect to economies of scale in account management and record keeping, will only put investment decisions into even fewer hands. Reform of capital markets, such as the 2001 reform in Chile that in- troduced member choice over five pension funds of different risk­return characteristics, may help reverse this trend. By promoting individual choice and competition among different providers in the huge pension sav- ings market, the extent of homogeneity of portfolios and synchronization of trading decisions is likely to fall substantially. Liquidity, the key to vi- brant capital markets, is likely to rise. Dealing with the Transition Debt The pension reforms undergone by Latin American countries were envis- aged to end the fiscal imbalances that plagued their social security regimens. However, as discussed in chapter 3, the move from pure PAYG to funding raises in itself short-term fiscal pressures because pensions in payments and accrued rights must be financed from a lower level of mandatory contribu- tions. In some countries, such as Bolivia, Chile, El Salvador, and Mexico, governments can no longer rely at all on the mandatory contributions be- cause these are destined exclusively for individual funded accounts. Despite the radical and pioneering nature of its pension reform, Chile has been by far the most successful in managing the pension debt. As discussed in box 3.2, the government ran fiscal primary surpluses on the order of 5.5 percent of GDP prior to the reform to soften the impact of transition costs on government finances. No other Latin American country, however, has been able to effect such a huge fiscal contraction prior to reform. On the contrary, some countries, such as Argentina, embarked on reform with large fiscal imbal- ances and tensions on the exchange rate. The partial move to a funded system and the reduction in contribution rates only worsened further the finances of the social security system. The Argentina experience demonstrates that against a backdrop of macroeconomic instability the private pension system is not free from po- litical manipulation. By the end of 2001 nearly two thirds of its pension fund assets were invested in government securities either directly or indi- rectly (through bank trusts). The latter were not counted by the superin- tendency for purposes of meeting the ceiling on investment in government securities. This move, however, was not sufficient to soak up the debt cre- ated by the pension reform. The pension funds had accumulated assets 70 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA worth only 7.4 percent of GDP by December 2001, slightly over half the debt created by the loss in revenue to the social security system. The burden of the fiscal cost of the transition has forced many other governments in the region to impose quantitative investment restrictions that help channel pension funds toward government securities. Even in Chile pension funds were not allowed to invest in any asset class other than government bonds and banking instruments during the first four years of the system.7 Bolivia, Mexico, and Uruguay have imposed floors on government bond investments. Even in the absence of quantitative re- strictions, the instability created by a large transition debt is an obstacle to the deepening of financial markets. In the absence of a sustained fiscal ef- fort, therefore, transition costs can severely curtail the positive impact of pension funds on capital markets. Asset Management Practices, Financial Innovation, and Regulatory Reform Pension funds have complete freedom in their investments as long as they stay within the regulatory ceilings established by the industry supervisor. We can therefore assess asset management practices in Latin American countries by taking into account the constraints that they face. Even in countries with more liberal investment regimens, such as Colombia, Chile, or Peru, pension funds' equity and corporate bond portfolios are concen- trated in large companies. This is partly a consequence of the stage of de- velopment and small size of their economies, which result in a low number of firms whose securities are traded regularly in liquid markets. Regula- tions, however, also play a role because investments are restricted accord- ing to the risk rating of individual securities and their trading record in reg- ulated markets. Performance regulations, which require pension funds to obtain a rate of return within a band set as a percentage of the industry av- erage, may also explain the high degree of homogeneity in pension fund portfolios and the general conservatism in their investment strategies. On the other hand, pension funds have played a role in the process of modernization of financial markets experienced by countries such as Argentina, Chile, and Peru. Lefort and Walker (2000a) mentioned the marked improvement in professionalism in the investment decisionmaking process and their involvement in bringing about a more dynamic legal framework. For example, pension fund managers in Chile added flexibil- ity to foreign investment rules by obtaining permission to use currency for- ward contracts as hedging instruments. The pension fund administrators also played a central role in the 1989 establishment of the Electronic Stock Exchange, which competes directly with the Santiago Stock Exchange. The pension fund administrators had a strong interest in increasing com- petition in the market because they pay directly any transaction costs from investment activities. But the impact of pension funds on stock market THE FINANCIAL BENEFITS OF PENSION REFORM 71 development should not be overstated. Although pension funds in Latin America and the Caribbean can be large as a percentage of GDP, they are also small as equity holders, holding less than 10 percent of domestic eq- uity in all Latin American and Caribbean countries for which data are available (Catalán 2003). Role of Pension Funds in Corporate Governance A captive market of mandatory pension contributions and a highly con- centrated industry in which pension fund managers make all investment de- cisions may not be the ideal recipe for liquid capital markets. Pension funds cannot easily sell securities that are performing badly. If they do so they can turn prices against themselves, especially because other pension funds are likely to follow suit. On the other hand, pension funds can exert their power on capital markets indirectly by asking and voting for changes in corporate governance practices. This role, however, is not free of con- straints. In all Latin American countries there are limits on the percentage of the capitalization of a certain issuer that can be held by a pension fund. These ceilings range from 5 to 15 percent of total capitalization. It is largely because of these regulations that pension funds have only be- gun to play an active role as shareholders in Chile. The size of pension funds is such that they have become collectively the largest minority shareholder of many companies traded in the stock market.8 Iglesias-Palau (2000) iden- tified three main factors that explain increased shareholder activism by pen- sion funds: (1) the counterproductive effects of exit strategies, (2) the high sensitivity of pension fund managers to the public's reaction to bad invest- ments, and (3) the high concentration in ownership of Chilean corporations. Some examples of the corporate governance role of pension fund ad- ministrators in Chile are their voting for independent directors (a regula- tory requirement) and the pressure they exert to improve the transparency of company accounts. Independent directors have been particularly active in monitoring potential conflicts of interest between majority and minor- ity shareholders. Iglesias-Palau (2000) also argued that independent direc- tors have promoted the establishment of specialized committees, such as audit committees. The pension fund administrators are also required by the supervisor to file reports regarding events or transactions by security issuers that may have negative effects on pension fund investments. This whistle-blowing role is played effectively by the Chilean Association of Pension Funds' in- forming the authorities and the public in general about corporate gover- nance situations that are detrimental to pension fund performance. These regulations have led to better corporate governance despite the fact that pen- sion funds are only minority shareholders. According to Lefort and Walker (2000a), other investors often explain their ownership plans to pension fund administrators and consider their opinion as influential shareholders. 72 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Pension funds are helping create a new balance in the Chilean corpo- rate ownership structure. Public firms in Chile, as in other Latin American countries, are dominated by one large group or conglomerate, which of- ten has a single family owner at the top of the pyramid. Pension funds, to- gether with ADR holders, are the largest minority shareholders of Chilean firms. Unlike ADR holders, who are a diversified and unconnected group, Chilean pension funds have similar objectives and follow practically iden- tical investment strategies. Hence, they can present a united and powerful voice to defend minority shareholder rights. But given the small share of domestic equity in the hands of pension funds in the region, even in Chile, the impact of the funds on corporate governance should not be exagger- ated (Catalán 2003). Effects of Reforms on the Market for Government Debt Except in Chile and Peru, more than half of pension fund investments are directed toward government securities. In this section we take a closer look at the role of pension funds in the development of this market. Development of the Market for Public Sector Debt Historically one of the key deficiencies of Latin American governments has been their inability to raise long-term financing domestically and their consequent dependency on volatile foreign capital. The weakness of do- mestic government debt markets is itself largely a reflection of a lack of fis- cal rectitude that Latin American governments have only recently started to address. Chile stands out as having succeeded in the 1980s at avoiding the worst of the debt crisis and has since been hailed as a model of fiscal rectitude. El Salvador, Mexico, and Uruguay gained investment-grade rat- ings in the 1990s, and the prospects for government debt markets were promising. The collapse in liquidity in international markets in 2000­01, however, truncated the hopes of most Latin American governments. The Argentine crisis spread to Uruguay, which lost its investment-grade rating. International investors have stampeded from Colombia, which also lost investment-grade status. Of all Latin American countries, Chile has been the most successful in limiting its sensitivity to portfolio flows and in lengthening the maturity of government debt. It may be tempting to link these developments to the role of pension funds, but there are other, more important factors at play. The practical elimination of the government's and Central Bank's foreign debt, a unique case in Latin America, was a public objective achieved thanks to two decades of fiscal surpluses. Meanwhile, short- term borrowing by the private sector and portfolio inflows have been THE FINANCIAL BENEFITS OF PENSION REFORM 73 discouraged through punitive reserve requirements. Fiscal frugality has also made possible the successful development of a long-term public sec- tor bond market in Chile, together with two other factors: the develop- ment of an efficient indexation unit, the unidad de fomento (UF), and the government's and central bank's promotion of market liquidity through debt management. The UF, an inflation-indexed unit of measure for all financial transac- tions, was first introduced in 1974 and was adopted widely as the refer- ence index only in 1984. Currently the UF is used for pricing more than one-half of financial assets and practically all medium- and long-term fixed income securities and instruments of financial intermediation. Although other Latin American countries have introduced indexation mechanisms, none have been as successful as the UF (see box 4.2). The Chilean government's and central bank's debt management strate- gies have also been beneficial to the health of the bond market. By pro- moting liquidity in the market the indexation unit was made a more acceptable currency unit and has ensured an adequate supply of securities for institutional investors. Markets have rewarded these policies with the lowest spreads in Latin America. Chilean public sector bonds also have the longest average duration in Latin America. The bonds have a maturity of between 90 days and 20 years. Although mandatory indexation helped the Chilean bond market de- velop, it is certainly not a prerequisite for a healthy bond market. Other Latin American countries have recently succeeded in raising the maturity of their debt without requiring indexation of all fixed-income securities. In Colombia and Mexico the government is issuing inflation-indexed securi- ties that are attractive to domestic investors, and the average duration of government bonds is increasing. In Colombia the government started to issue inflation-indexed bonds in 1999. In 1997, 86 percent of government bonds had a maturity of less than five years. By March 2002, 22 percent of domestic treasury bonds were inflation-indexed. The share of government bonds with a maturity of between 5 and 10 years has also increased from 11.5 percent in 1997 to 46.7 percent in 2002. The Mexican government recently also started issuing inflation-in- dexed securities. Only 14 percent of government bonds outstanding at the end of 2000 were inflation-indexed. Meanwhile, the maturity of the debt is beginning to creep up slowly, from its low level during the 1994­95 cri- sis. The average maturity of government debt increased from less than one year at the end of 1994 to more than two and a half years by De- cember 2002. This trend is likely to progress because the government has stated its objective to lengthen the maturity of its debt. The government recently succeeded in introducing 10- and 20-year bonds. More than one- half of net public borrowing in 2001 was in fixed rate bonds with three- and five-year maturities. 74 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 4.2 Inflation-Indexed Securities in Latin America Chile was the second country in Latin America, after Brazil, to introduce widespread indexation of financial securities to a measure of the cost of living. Walker (1998) argued that indexation worked in Chile because (a) the unit has credibility, in the sense that it will not be manipulated by the authorities, and is based on the consumer price index that is computed by an independent entity, the National Institute of Statistics; (b) legislation requires that medium- and long-term credit be indexed to the unidad de fomento (UF), and calculation of assets and liabilities of insurance com- panies is in UF; (c) there exists a deep, liquid market for central bank in- dexed bonds, giving a risk-free rating used in other transactions; and (d) tax regulations are consistent with a generalized indexation of the econ- omy. In addition, the UF was successfully adopted in Chile because it is updated on a daily basis, functioning as a quasi-perfect indexation mech- anism. Indexation in bank intermediation and government funding permitted the practical elimination of money illusion and hence created the condi- tions for macroeconomic stability. The inflation rate has been falling gradually since the early 1980s. In 2000 it was already below 4 percent. Indexation also permitted the stabilization of real interest rates and the targeting of the real exchange rate around what were deemed equilibrium values. These developments, coupled with the government's tight fiscal policy stance, created the conditions for a healthy government bond mar- ket. This market has since avoided much of the pain endured by other countries, such as Argentina, which went down the alternative route-- dollarization--in their attempts at stabilization. Of all the other Latin American governments only Brazil has had as high a degree of indexation of financial securities as Chile. Unlike the situation in Chile, however, there was no single reference unit, and ad- justments often took place on an arbitrary and irregular basis. Moreover, indexation spread to labor markets, where it sustained ever-higher infla- tionary pressures. The introduction of a new currency, the real, pegged to the dollar succeeded in bringing down inflation. At the same time, the proportion of indexed debt in total debt decreased from about 70 percent at the beginning of 1994 to less than 30 percent by 2001. The governments of Colombia, Mexico, and Peru have also recently begun to issue inflation-indexed securities. In the other Latin American countries, such as Argentina, Bolivia, El Salvador, and Uruguay, govern- ments rely mainly on dollar-denominated securities. There has been much discussion about the benefits and drawbacks of both indexation and dol- larization. Indexation can sustain and even augment inflationary expec- tations when it spreads to labor markets. Dollarization, on the other hand, exposes the government to significant currency risk because tax revenues from the nontradable sector are linked to the domestic currency. THE FINANCIAL BENEFITS OF PENSION REFORM 75 Box 4.2 (continued) This may explain why foreign investors generally have been unwilling to provide long-term financing denominated in U.S. dollars. Source: Yermo (2002a). Effects of Pension Funds on the Market for Government Bonds There is little doubt that pension reform has contributed to the improved health of the Chilean public bond market. First, it may have encouraged fiscal restraint by the government. Second, it has provided "recognition" bonds with a long maturity. These bonds, issued by the government to compensate those who accumulated pension rights under the old system, became transferable in 1994 and have been traded in exchanges. The pen- sion reform, therefore, contributed directly to the financial depth of the economy. Moreover, these bonds have relatively long duration (they are zero coupon bonds) and are therefore a stable source of funds for the Chilean government. Finally, the bonds have also helped the development of a benchmark yield curve that can be used for pricing private sector securities. On the other hand, the role of pension funds in the development of the public sector bond market is less clear. Pension funds have certainly been a significant and reliable source of financing for the Chilean government and the central bank. In August 2002 more than 21 percent of the mixed fund portfolio consisted of central bank bonds with an average duration of three years and eight months. Nearly 6 percent of the same portfolio was invested in recognition bonds with an average duration of four years and seven months. The Chilean experience contrasts with that of Peru, where the govern- ment has relied mainly on external financing. Only recently have pension funds been allowed to invest in these issues, so their contribution to the de- velopment of the market has been minimal. Unlike Chile, El Salvador and Peru have not yet permitted trading in secondary markets of recognition bonds. Given the lack of long-term government securities issued in do- mestic currency, such bonds would be a highly attractive investment for institutional investors such as pension funds. They would also assist in the construction of a yield curve that can be the basis for developing the cor- porate bond market. As mentioned earlier, the role of pension funds in Mexico has been conditioned by investment regulations (pension funds are required to in- vest at least 51 percent of their assets in inflation-indexed securities). In 76 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA December 2002 the Specialized Retirement Fund Investment Societies (Sociedades de Inversión Especializada de Fondos para el Retiro) invested more than 70 percent of their assets in inflation-indexed government bonds. On the other hand, their contribution to the increase in maturities has been minimal because until December 2001, regulations impeded them from investing more than 65 percent in instruments with maturities longer than 183 days. The average maturity of the pension fund portfo- lios during the first two years of the system was in fact only 238 days (Rubalcava and Gutierrez 2000), well below the average maturity of Mexican government debt over that period. The experience in other countries has been largely disappointing. Where pension funds have contributed to providing long-term funds, it has often been the result of government regulations or political pressures. In Argentina pension funds could invest up to 30 percent of their assets in an "investment account" where government bonds, mainly dollar-linked, were held up to maturity. After the 2001 crisis the government bonds held by pension funds were transformed into illiquid long-term loans to the government. In Bolivia the requirement to buy US$180 million worth of government securities per year has turned pension funds into the largest holders of these securities in the space of a few years. The pension funds must buy dollar-denominated government bonds that must be held to ma- turity (15 years) and that pay an 8 percent coupon. Sixty percent of the pension funds' assets were invested in these bonds in December 2001. In addition, in the secondary market pension funds buy government bonds that have maturities between one and three years. The government is also currently facing fiscal pressures and is considering replacing the dollar- denominated government bonds held by the pension funds by domestic currency­denominated debt. Effects of Reforms on the Banking System Except in Chile, bank loans are still the main form of external financing of the nonfinancial private sector. In Chile the stock market overtook bank credit as the main source of external funds in the late 1980s. Despite the slower growth of bank credit relative to the stock market, Chile stands out among Latin American countries for its high credit-to-GDP ratio--about 67 percent in December 2002, more than double the average for Latin American countries. Pension funds can affect the evolution of the banking system through their investment strategies. In Latin America pension funds invest a large portion of their assets in financial instruments issued by banks. Only in Argentina, Costa Rica, and Mexico do pension funds allocate less than 14 percent of their assets to bank instruments. Pension funds therefore appear to play a complementary role to the banking sector. THE FINANCIAL BENEFITS OF PENSION REFORM 77 Role of Pension Funds in the Development of the Market for Banking Sector Securities Pension funds have played a central role in the growth of the mortgage bond market in Chile, by far the most developed in Latin America. Pen- sion funds provide housing finance indirectly through two main types of investment: letras hipotecarias (mortgage bonds) and real estate invest- ment funds. The letras are by far the most important type, making up more than 13 percent of the pension funds' portfolio. Pension funds own more than one-half of this market. The letras are mortgage bonds backed by a portfolio of real estate and guaranteed by the commercial banks that issue these instruments. They are traded in exchanges and are thus eligible for investment by pension funds. They can have a maturity of 8, 12, or 20 years. The letras were introduced in 1977 and experienced rapid growth until the 1982­83 financial crisis. The loans financed through the letras have financed mainly the middle- and high-income residential sector.9 As with government bonds, the reason for the long maturity of letras has much less to do with pension fund investment per se than with the availability of price indexation for these securities and increased macro- economic stability. Indexation to the UF itself has been possible thanks to the availability of a liquid market for UF-indexed government and Central Bank bonds of long maturity. These bonds provide the risk-free bench- mark for pricing and, therefore, trading mortgage bonds. Pension reform had a defining impact on this market. Pension funds, together with life insurance companies, have been the main investors in mortgage-backed securities since the early 1980s. Pension funds had 12 percent of their assets invested in letras in August 2002, with an average duration of four years and seven months. By providing medium- and long- term funding for house purchases, pension funds played a central role in the expansion of real estate investment that took place in the second half of the 1980s. The only other country in Latin America that has seen significant growth in housing-related securities is Peru.10 Leasing bonds account for more than 10 percent of the pension fund portfolios. The market is domi- nated by subordinated and leasing bonds that are issued by financial insti- tutions. These bonds have replaced government debt as the reference benchmark because that market is small and illiquid. An increasing portion of private sector bonds is indexed to the valor adquisitivo constante (VAC, constant purchasing power), the currency unit that is linked to the con- sumer price level. As of September 2001 nearly one-fifth of all private bonds were denominated in VAC; the rest were mainly denominated in dol- lars. Peruvian bonds have also increased in maturity over the last decade. The share of bonds with a maturity longer than five years increased from 9 percent in December 1998 to 37 percent in September 2001. 78 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Pension funds, however, cannot be held responsible for these changes because they invest largely in dollar-indexed bonds. In 1998, only 10 per- cent of total private bonds held by them were denominated in VAC. The limited interest in indexed bonds may be a sign of lack of credibility in the indexation unit in the context of a highly dollarized economy. Unlike in Chile, the mortgage bond market in Peru (Títulos de Crédito Hipotecario Negociable) has not benefited from pension fund investment. These bonds have existed for many years but pension funds have hardly invested in them (less than 0.1 percent of the portfolio). This apathy de- rives from the low capitalization of the market. The main obstacle to its development is the inability of banks to issue such instruments directly without the authorization of the borrower. Pension funds in other Latin American countries invest a smaller per- centage of their assets in securities issued by banks. Most of the assets in- vested in the financial sector in other countries go to time deposits and liq- uid bank instruments. In Colombia, for example, pension funds invest more than one-quarter of their assets in the financial sector, but less than 1 percent is invested in mortgage or leasing bonds. Impact on Bank Efficiency and Stability Pension funds can help reduce the cost of issuing securities and hence re- duce the market power of banks. As a result of this competitive pressure, net interest margins11 may decrease. At the same time, pension funds can contribute to sustainable growth in bank credit. By helping promote cap- ital markets, and in particular the stock market, pension funds stimulate information disclosure and monitoring and hence may reduce the credit risk borne by the banking sector. Pension funds may also be attracted by long-term deposits, helping reduce term transformation risk in the bank- ing system. In addition to their impact on bank efficiency and the growth of credit, pension funds can affect the maturity structure of bank loans. Depending on whether complementary or substitutive effects dominate, the average maturity of loans may increase or decrease. Levine (1997) argued that stock markets and banks tend to be complements rather than substitutes in emerging economies, and thus the maturity of loans would normally in- crease. Impavido, Musalem, and Tressel (2001a) confirmed the complementary nature of the relationship between banks and institutional investors for a broad sample of countries, including four Latin American countries (Argentina, Brazil, Chile, and Mexico). Both bank profitability and the ma- turity of loans increase as pension fund and insurance company activity in- creases. They find that these increases result largely from a reduction in credit risk. Differences in bank efficiency across countries are also related to the level of development of the pension fund and insurance company THE FINANCIAL BENEFITS OF PENSION REFORM 79 Figure 4.2 Interest Rate Spreads Have Declined in Peru and Bolivia Since the Reforms (1993­2002) Percentage point difference between lending and deposit rates 60 50 40 30 20 10 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Bolivia Chile Colombia Mexico Peru Source: IFS (International Monetary Fund) 2003. sectors. The positive correlation is larger at low initial levels of develop- ment, and it decreases as pension funds and insurance companies develop. This trend is shown in figure 4.2, which plots the interest spread (lend- ing rates minus deposit rates) in selected Latin American countries over the 1990s. The spread is lowest in Chile, the country with the most developed pension fund and insurance company sectors. But over the 1990s the spread did not decline much. This stability is related to the higher degree of competition faced by the banks in providing financing sources, which has led them to concentrate in alternative markets (such as personal bank- ing or small- to medium-size firms), which are associated with higher costs. In other Latin American countries, such as Peru, the competition provided by pension funds and insurance companies is currently sufficient to stimulate a reduction in bank spreads but low enough to allow banks to maintain their financing of larger companies. Another indicator of improved efficiency is the reduction in operating expenses. Chilean banks have operating costs that are less than 3 percent 80 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA of total assets, less than one-half the level in other Latin American countries. Pension funds may have contributed to the increased competi- tion in the banking system during the late 1980s and early 1990s by pro- viding an alternative form of financing. However, there is no systematic evidence that the low interest rate spreads in Chile are related to the role of pension funds. Effects of Reform on the Market for Private Sector Securities In this section we look deeper into the operation of pension funds in Latin America to better assess their role in the development of the market for private sector securities. Pension Fund Investment and Stock Market Liquidity Only in Argentina, Chile, and Peru have pension funds been able to invest significantly in the stock market. Colombia is the only other Latin American country where such investments are permitted, but investment in stocks has been minimal (less than 3 percent in December 2002). Most of the existing evidence on the impact of pension funds on stock market development is from Chile. The growth in the capitalization of the Chilean stock market after 1984 coincided with heavy investment by pen- sion funds in shares. There is a high correlation between the amount of equities held by pension fund assets and the increase in the ratio of stock market capitalization to GDP, which has reached levels similar to those of the most developed OECD countries (greater than 100 percent). The evi- dence on market liquidity is also supportive of a causal link with pension fund investments. Iglesias (1998) provided evidence that transaction costs in securities markets fell in Chile after pension funds started investing in private sector securities. Fees charged by the Santiago Stock Exchange for market transactions dropped from 0.500 and 0.015 percent in 1985 to 0.120 and 0.000 percent in 1994. Holzmann (1997) identified a positive correlation between the growth of pension fund assets and monthly traded values in Chile, and Lefort and Walker (2000a) found corroborating evi- dence showing that the growth in pension investments has contributed to the growth in traded volumes in Chile since 1985. Despite these positive findings the liquidity of the Chilean stock mar- ket is well below what would be expected from a country with one of the highest ratios of stock market capitalization to GDP and well-developed pension fund and insurance company sectors. The growth in pension fund investments and the rapidly growing allocation to domestic equities since 1985 have not been sufficient to raise turnover ratios (value traded as a percentage of market capitalization) to the levels observed in THE FINANCIAL BENEFITS OF PENSION REFORM 81 Figure 4.3 Stock Market Turnover Ratios in Selected Latin American Countries (1990­2001) Ratio of traded volume to market capitalization 1.2 1.0 0.8 0.6 0.4 0.2 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Argentina Brazil Chile Colombia Mexico Peru Uruguay Source: World Bank Financial Structure Database. countries like Brazil or Mexico, let alone in developed countries (see figure 4.3). In fact, this measure of liquidity is below the development threshold level of 15 percent proposed by Demirguç-Kunt and Levine (1999) for the 1990s. The low liquidity of the Chilean and most other Latin American stock markets can be partly explained by the high degree of ownership concentra- tion and deficiencies in disclosure standards and in protection of the rights of minority shareholders that have been addressed only recently. Nonetheless, in other countries in the region, such as Brazil and Mexico, these deficien- cies have been at least partly offset by foreign investors actively trading in local stocks (though mainly through ADRs). Foreign investors have helped make Brazil and Mexico the two most liquid markets in the region, ac- counting for more than 90 percent of all Latin American equity trading. In Chile capital controls and an onerous tax treatment of foreign port- folio investment have prevented foreign investors from playing an active role in its markets. Despite their size, local pension funds have not been 82 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA able to sustain levels of liquidity as high as foreign investors have in Brazil and Mexico. The high degree of synchronization in the choice and timing of stock purchases and the rapid accumulation in pension fund assets have contributed to creating a market where "buy and hold" is the only viable investment strategy. Chilean pension funds have no large counterpart that can buy their stocks when they are ready to sell. The positive liquidity ef- fect of pension fund investment is largely a result of the monthly flow of mandatory contributions rather than the daily trading activities of the pension funds. This situation may change with the liberalization of the capital account approved at the end of 2001. The Central Bank has since eliminated all ad- ministrative barriers that regulated capital flows and ADR issues. The Chilean Ministry of Finance eliminated the tax on short-term purchases and authorized short sales. Although the liberalization of the capital account may have a positive impact on stock market development, it may not be enough. In fact, high exposure to foreign capital could actually make local mar- kets less liquid and more volatile during a crisis than in more closed economies. The Asian and Argentine crises have dried up liquidity in most Latin American markets. The most prominent case is Argentina, where some of the companies with the largest capitalization have been delisted after being acquired by foreign players (for example, the oil company YPF was sold to Repsol). Even in Chile the presence of a large institutional in- vestor industry did not prevent the fall of traded volumes of 55 percent ob- served between 1995 and 2000. In Peru pension funds invest heavily in equity and the capital account is open, but just one Peruvian stock, gold producer Cia. de Minas Buenaven- tura, is liquid enough to attract foreign investment. The prevalence of work- ers' shares may deter foreign investors because of their attendant dividend and liquidation rights and the complexities and uncertainties associated with them.12 At the same time the investment strategies of Peruvian pension funds do not seem to serve as a stimulant to market liquidity. Indeed, Lefort and Walker (2000a) did not find any causal relationship between pension fund investment in equities and market liquidity between 1993 and 1999. Investment regulations (particularly those rules that limit investment to securities with the highest risk rating and those that limit the portion of cap- italization that pension funds may own) are also not conducive to stock mar- ket liquidity. Additionally, they reinforce another negative aspect of Latin American stock markets: the high concentration of both capitalization and liquidity in a handful of stocks. The 10 most-traded companies in each country account for more than two-thirds of the local stock market's trad- ing in all Latin American countries except Brazil (40 percent). In Chile and Argentina three companies account for almost 50 percent of current capital- ization and turnover. In Mexico Telmex accounts for one-quarter of stock market capitalization and between 20 percent and 40 percent of daily trading. THE FINANCIAL BENEFITS OF PENSION REFORM 83 Relaxing these investment rules within a prudential regulatory frame- work would go a long way toward improving the diversification of Latin American stock markets. Pension funds in Chile have steadily increased the number of issuers in their portfolios, thereby contributing to the broadening of the market. On the other hand, the continuing high con- centration and herding of the industry can hardly be expected to bring about drastic improvements in liquidity. Only in Chile, where workers can now choose among five different funds, is liquidity likely to increase sig- nificantly. Diversity in options and preferences is an essential aspect of liquid markets. Development of the Corporate Bond Market Unlike the mortgage bond market, the corporate bond market has experi- enced very limited growth in Chile. Longer-term debt has become more dominant, but this seems to have much more to do with the availability of indexation than with the growth in pension fund assets. The development of the corporate bond market would also be unthinkable without the tight information disclosure standards introduced by the 1980 Securities Law. Nonetheless, pension funds have an important presence in this market and have been investing in longer-term maturities. In August 2002, 5 percent of their portfolios was invested in corporate bonds with an average dura- tion of five years. Chilean pension funds have also recently been permitted to invest in infrastructure bonds. These bonds are backed by insurance companies that guarantee repayment of the principal. The companies that issue these bonds are also guaranteed a minimum revenue by the state. Such invest- ment is still very small, less than 1 percent of pension fund portfolios. The availability of leasing and subordinated bonds in Peru has helped develop a local corporate bond market. However, these bonds cannot serve as a perfect substitute for treasury bills and bonds because they may have a significant credit risk. Without a reliable yield curve, trading corporate bonds in secondary markets exposes pension funds to liquid- ity risk. Pension funds have been avid buyers of corporate bonds, but they do not trade them regularly in secondary markets. The contribution of pension funds to the development of this market is also limited by risk-rating requirements. Pension funds are also relatively important in- vestors in corporate bonds in Bolivia, Colombia, and Mexico. In Mex- ico, pension funds started to invest significantly in corporate bonds only during the past couple of years. So far their investment has been in short maturities, so their contribution to the breadth of the market has been very limited. The corporate bond market has shown limited development in Argentina, where the bonds must be held until maturity (about two years) because they do not have a secondary market. Pension funds currently 84 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA hold around 1 percent of their assets in corporate bonds. In Costa Rica, El Salvador, and Uruguay pension fund investment in corporate bonds has also been very subdued. Venture Capital and Real Estate Funds Pension funds are not permitted to invest directly in venture capital or real estate. In Argentina, Chile, and Peru pension funds can invest in these assets indirectly, by purchasing shares of investment and mutual funds that themselves own these assets (ceilings at 30 percent and 20 percent of the fund, respectively). Chilean pension funds invest less than 2 percent of their assets in venture capital funds (Enterprise Development Funds). These entities, which take the form of closed-ended mutual funds, were in- troduced in 1989 and pension funds quickly became their main investors. On the other hand, the real estate fund sector has hardly taken off. An important obstacle to further investment in these funds is the body of investment regulations that limit the portion of the value of the securi- ties issued by one such fund that can be owned by a pension fund. Given the small size of the private equity industry in relation to the pension funds, such limits are highly constraining. Pension funds are also discour- aged from investing in such funds by performance rules because the valu- ation of venture capital funds is subject to much uncertainty. In Argentina pension funds used to invest in this asset class before the crisis, but since then most of these investments have been sold. In Peru pension funds can invest in real estate funds but venture capital funds are not functioning yet. Investment in real estate funds is also very small, rep- resenting less than 1 percent of pension fund assets. In other countries in- vestment in venture capital and real estate funds is not permitted. Indeed, in Mexico and Uruguay pension funds are barred from investing in any type of mutual funds. It may be expected that pension funds' role as providers of finance for small- and medium-size enterprises and infrastructure projects will in- crease in the future. In a region where small businesses account for the bulk of sales and employment, and where there is need for massive infra- structural investment, such a development would be highly welcome. Governments can play an important role by ensuring that investment and performance regulations do not impede pension funds from investing suf- ficiently in private equity. Conclusion In this chapter, which is based on Yermo (2002a), we have shown that so- cial security reform in Latin America has been accompanied by a set of corollary reforms that have had salubrious effects on capital markets. THE FINANCIAL BENEFITS OF PENSION REFORM 85 Capital market development in some Latin American countries has been driven largely by the state-sponsored modernization of the financial sector infrastructure, tax and bankruptcy reform, and the regulatory structure developed by the authorities for the pension funds and other financial in- stitutions. Possibly the most important development in capital markets in the re- gion during the 1990s has been the introduction of a new type of financial institution, the pension fund administrator, whose function is to invest pension contributions in financial assets. The reforms can therefore be credited with setting up a new financial industry that, at least in terms of institutional oversight, has been a role model for other financial institu- tions in the region. Although pension fund regulators in Latin America have erred on the side of caution, there is little doubt that the new systems have achieved some of the highest standards in the region in asset valua- tion, risk rating, and disclosure. By subjecting the new financial intermediaries to high regulatory and supervisory standards, pension reform has also made a major contribution to the rapid modernization of the financial market infrastructure observed in the region over the last few years (especially custodial and risk-rating services). Furthermore, it has forced an adjustment of standards and prac- tices in other financial institutions and in the capital markets. Trans- parency and integrity in financial markets have been dramatically im- proved as a result. In principle these improvements could have taken place independently of the pension reform. However, the mandatory nature of the funded pension systems provided the political justification for these much-needed developments. There are some caveats to this generally positive assessment: 1. In terms of investment and performance objectives, pension funds are hardly different from mutual funds. Hence, it is possible that similar benefits could have been obtained had the pension fund regulatory frame- work also been applied to the mutual fund industry. An important ques- tion is whether the pension reforms necessitated establishing a new finan- cial intermediary. It may be argued that if stricter regulatory and supervisory standards had been applied to existing financial institutions such as banks, insurance companies, and mutual funds, and these institu- tions had been allowed to manage the pension savings, the financial de- velopment indicators would have similarly improved. The liberalization of the market for voluntary pension savings in Chile (see chapter 7) may pro- vide some ground for testing this hypothesis. 2. Although the private financial and nonfinancial sectors in Latin America have benefited from the growth of the pension industry, the main beneficiary has been the government debt market. Pension funds in some countries have participated actively in securing bank loans by investing in mortgage bonds (Chile) and leasing bonds (Peru). Pension fund investment 86 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA has also contributed to the growth of corporate debt and equities markets in these countries. Only in Chile, however, is there some evidence of a pos- itive, causal relationship between pension fund investment and stock mar- ket liquidity. Similarly, Lefort and Walker (2000a) found evidence only in Chile that pension funds contributed to a lowering of the cost of capital to firms. In other Latin American countries investment in bank instruments has been limited largely to time deposits and other short-term securities. Investment in nonfinancial private sector securities has been muted (pen- sion funds invest less than one-fifth of their assets in such instruments), partly as a result of investment and performance regulations. Regulations have also constrained the role of pension funds in the development of de- rivatives instruments. 3. Contrary to the situation in some OECD countries, where pension funds have been an independent driving force behind important financial innovations, the role of pension funds in the development of capital mar- kets in Latin American countries is largely determined by government in- structions that touch every aspect of their operations, from the amount of contributions that the industry receives to the investment of pension as- sets. The industry's structure and regulations reinforce the pension funds' preference for "buy and hold" investment strategies that are not conducive to market liquidity. 4. Although the illiquidity of pension investments coupled with the conservatism in investment strategies have brought stability to these mar- kets, much of this stability is artificial. It is driven at least in part by port- folio rules that force pension funds to hold mainly domestic assets and in some countries oblige them to invest a minimum percentage of their assets in government bonds. 5. Macroeconomic stability is essential for capital markets to reap the benefits from pension fund investment, regardless of whether such invest- ment is forced. The government debt market is most developed in Chile, where the government pursued fiscal consolidation prior to the reform and sustained this effort over later years. The pension reform also con- tributed directly to financial deepening because the transition debt was turned into tradable government securities. In Argentina, on the other hand, any positive short-term effect of pension fund investment on capital market development has been obliterated by the economic crisis. Notes 1. The pension funds are pools of assets owned by the members (affiliated workers) of the pension fund and legally separated from the fund administrators. The only exception to this legal norm is Mexico, where the pension fund is itself an independent legal entity containing a board of directors. 2. Pension funds are also subject to ownership concentration limits as dis- cussed later in this chapter. THE FINANCIAL BENEFITS OF PENSION REFORM 87 3. An important exception is Argentina, where up to 30 percent of pension fund assets can be invested in government bonds held to maturity in an "investment account." The assets are therefore priced at book value. Bolivia also experienced a serious blow to the transparency of the system when the collective fund assets (which finance the BONOSOL program) were mixed with those of the individual accounts. The collective fund is invested in nonlisted companies and is subject to discretionary valuation. 4. Such a method was originally designed with a view to ensuring the compa- rability of pension fund portfolios and permitting adequate monitoring by the reg- ulator, the National Retirement Savings Commission (Mexico). It is now expected that insurance companies and mutual funds will be required to use the same valu- ation method. 5. Concerns over the fiscal cost of pension guarantees can explain why Mex- ico does not allow investment in equities. The guarantee is equal to the salary- linked defined benefit under the previous social security regime for all workers who contributed to it. In other Latin American countries the state's liability in the funded system is limited to a minimum income guarantee (a flat benefit, at a level somewhat below the minimum wage), as discussed in chapter 9. 6. ADRs are certificates issued by banks in the United States that represent shares or bonds issued by a foreign company or a foreign subsidiary of a U.S. com- pany. ADRs are backed by securities in custody in the country where the firm that issues the securities is based. ADRs can be converted at any time into the underly- ing securities. They serve as a signaling device and can therefore contribute indi- rectly to the liquidity of the local stock market. 7. Chile first liberalized the investment regimen for pension funds in 1985 by permitting investment in equities and corporate bonds. 8. Because pension funds in the region often concentrate their equity holdings in a small number of blue-chip firms, they can exert major influence on these com- panies' corporate governance. In Chile, for example, AFPs have become the most important minority shareholder in most of the country's listed corporations. As of December 2000 equity investments by pension funds accounted for more than 7 percent of total market capitalization. 9. Lower-income households are provided with very generous subsidies to buy their first home. Only a small loan is necessary to be able to access these gov- ernmental subsidies (Rojas 1999). 10. Pension funds are contributing to the financing of the housing market through other means in the other Latin American countries. In Mexico workers must make a separate contribution to the National Housing Fund for Workers In- stitute, a state-sponsored body that provides housing loans. In Uruguay the pension funds have a large part of their deposits in the Banco Hipotecario, the national mortgage bank. 11. The net interest margin is equal to total interest revenues minus total inter- est expenditures divided by the value of assets. This measure was proposed by Demirguç-Kunt and Levine (1999). 12. Workers' shares are nonvoting shares created during the military govern- ment in 1970 as a form of profit sharing and popular capitalism. 5 Social Gains from Pension Reforms in Latin America SINGLE-PILLAR PUBLIC PENSION SYSTEMS in developing countries, particu- larly in Latin America, tend to generate regressive transfers from poorer workers to the relatively small number of higher-income workers covered by the systems. Simulations show that equity can be increased by moving from a purely PAYG system to a multipillar system with a large funded component by reducing this regressive character of transfers. However, the large proportion of the workforce employed in informal activities in developing countries hinders realization of an equitable pension system, even if a multipillar system is put in place. Informal workers will remain largely uncovered under the reformed system but nevertheless share the tax burden of financing minimum guarantees to the covered few. There- fore, increasing coverage is crucial not only to help households manage risks but also to generate equitable outcomes. Economic theory would predict that, by reducing both the actual and the perceived tax on labor by establishing individual retirement savings ac- counts, pension reform will increase formalization of the labor force and its by-product, pension system coverage. Although there is no clear evidence of increased coverage in Latin American countries that have implemented pension reforms, Packard (2001) found that when controlling for other ex- planatory variables, there is a positive incentive effect of individual ac- counts on pension contributions. Nonetheless, pension coverage remains low and inequitably shared among income groups in Latin America, and this presents challenges even for countries that have implemented reform. Inequitable Effects of Social Security and Progress with Reforms Although presented as a model of solidarity when they were introduced decades ago, single-pillar social security systems in developing countries 89 90 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA that define benefits on a PAYG basis can be regressive in a number of ways. First, pension benefits are based on earnings rather than on need and are often calculated to favor better-educated workers with steeper age/earnings profiles. Second, contributions from poorer workers with higher average mortality often subsidize benefits paid to longer-lived, higher-income workers. Third, and related to the previous point, poorer workers tend to begin working and contributing earlier than do those who are better off: workers in higher-paying jobs requiring more education tend to join the labor force later; because they start working sooner, the poor contribute longer during their active lives and receive a relatively shorter stream of benefits in retirement. Fourth, various exemptions, such as earlier retirement ages for select groups of workers including teachers and police, and lower contribution rates for civil servants often redistrib- ute income from poorer to wealthier groups. By replacing what were often regressive, single-pillar PAYG systems that frequently paid overly generous pensions to a privileged few with sys- tems that diversify the risks to retirement income across multiple pillars, reforms were expected to introduce a more equitable system. Reformed systems would offer minimum income guarantees while they freed public resources for better-targeted forms of social assistance. The question is, have multipillar systems corrected the regressive impact of single-pillar public PAYG systems? Very little evidence exists one way or another, largely because re- forming governments are still paying transition costs and because the final impact of structural reforms on income inequality cannot be pre- cisely measured until large segments of the population begin to retire with pensions financed primarily from individual retirement accounts. That said, there are important equity-related implications of the reforms that can be examined using simulation tools such as the World Bank's PROST.1 The introduction of multipillar systems is likely to have a substantial impact on distribution between beneficiaries of different income levels. Figure 5.1 shows the simulated effect of reform on the internal rates of return earned by poorer versus wealthier workers covered by formal retirement security systems. The figure shows the percentage-point dif- ference between the internal rate of return earned by a representative wealthier-than-average worker and that earned by a poorer-than-average worker of either gender in each country. Both reform and no-reform scenarios are depicted, showing gains in equity under reforms in all cases. Single-pillar PAYG systems in Latin America were notoriously regres- sive, conferring substantially higher returns on wealthier workers. In every country (for which data were available for PROST simulations) pension reforms that introduced multipillar systems lowered the regressive impact of single-pillar PAYG systems. In Chile and Argentina our simulations show that reforms even reversed regressive returns, increasing returns SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 91 Figure 5.1 Structural Reforms Are Likely to Improve Equity by Lowering Regressive Transfers and Returns 4 Reform No reform return system 3 wealthier of 2 rates security between 1 internal 0 in retirement difference -1 workers national -2 from poorer -3 Men Men Men Men Men Men Men Men Percentage-point and earned Women Women Women Women Women Women Women Women Chile Peru Colombia Argentina Uruguay Mexico Bolivia El Salvador Source: Zviniene and Packard (2002). earned by poorer men and women covered by the systems relative to the return earned by wealthier workers of either gender. Reforms have also had a notable impact on distribution of returns be- tween genders. Figure 5.2 shows the percentage-point differential between the internal rates of return earned by men and women at average income levels (of covered workers) in each country. In Chile and Colombia re- forms marginally increased the returns earned by women relative to those earned by men. Cox-Edwards (2000) claimed that the impact of reforms in Chile favoring women can be attributed to the minimum pension guar- antee underpinning individual private savings in that country. In Peru reforms actually reversed distribution of returns from men to women. However, in every other country where the new retirement security model was adopted, women earn lower returns from the new systems relative to the returns earned by men. That said, James, Cox-Edwards, and Wong (2003) used survey data and simulation techniques to show that poor women in Argentina, Chile, and Mexico have gained from reforms, receiving higher pension benefits than they would have received under the single-pillar PAYG systems. These gains are attributed to a better-targeted first pillar. 92 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 5.2 Structural Reforms Make the New Systems More Gender Neutral, but Women's Average Benefits Can Be Significantly Lowered by the Use of Gender- Specific Mortality Tables 4 and from Reforms No reforms 3 men earned system 2 between return 1 of security 0 difference rates retirement -1 internal in national-2 Percentage-point women -3 Chile Peru Colombia Argentina Uruguay Mexico Bolivia El Salvador Source: Zviniene and Packard (2002). However, in Bolivia, Mexico, and El Salvador our simulations show that reforms replaced systems that subsidized benefits to women with sys- tems that favor men. This shift may partially reflect the relatively greater importance of labor market participation and regular contribution in de- termining pension benefits under the new retirement security systems that are based on defined contributions (James, Cox-Edwards, and Wong 2003). The negative impact on the returns to women may reflect their rel- atively fewer years of employment and active contribution to the pension system. Men with fewer years of employment and those who work with fewer years of contributions would suffer a similar fall in returns with the introduction of individual accounts. The internal rates of return calculated with PROST that are shown in Figure 5.2 assume that participating men and women have the same work and contribution histories. There is a more important factor increasing the gender-specific difference in returns between the two systems: the use of gen- der-specific mortality tables in calculating retirement annuities. Because women have relatively longer life expectancy at retirement, annuity providers using gender-specific mortality tables will calculate annuity SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 93 payments that are significantly lower than payments made to men retiring with similar levels of accumulated savings. Policies requiring a married male affiliate to retire with joint annuities that will cover his female spouse can improve the retirement security of sur- viving widows. Mandates requiring that private annuity providers use uni- sex mortality tables can correct disparities in returns from systems based primarily on individual retirement accounts, but they are highly controver- sial because they hinder the functioning of insurance markets. Attempts to meet equity objectives through further mandates in the operations of the private pillar--requiring insurers to use unisex mortality tables, in particu- lar--can be detrimental to the development of private insurance markets and are better pursued through whatever first-pillar arrangements coun- tries have put in place. There is evidence to suggest that the new first-pillar arrangements are succeeding in lowering the vulnerability of elderly women (James, Cox-Edwards, and Wong 2003). Equity Implications of a Large Informal Sector The equity issues examined in the sections above focus on the universe of workers actually covered by formal retirement security systems. However, the low rates of regular contribution among workers in Latin America, where many (if not most) workers are employed informally or self- employed, add another dimension to the inequitable impact that formal retirement security systems can have. Because the majority of workers will not receive any benefits, the deficits of unbalanced, single-pillar public pension systems (where contribution revenue does not cover benefit expenditure) that are financed from current and future tax revenues can represent a transfer from uncovered workers to those covered by the systems--a relative minority of workers already benefiting from more stable, better-paying forms of employment. But even in the new multipillar pension systems with a large funded pillar, individ- ual savings are almost always underpinned by some sort of public guaran- tee--often a minimum pension guarantee under which access to the mini- mum benefit is conditioned on a history of contributions to the system. Figure 5.3 shows the marginal contribution to inequality of income (as measured by the Gini index) from public and publicly mandated pension systems (and primary employment, for comparison) in selected Latin American countries. Although any extrapolations have to be made with caution, the inequitable impact of pension benefits is notably greater in countries with unreformed purely PAYG systems (Brazil, Paraguay, Venezuela, and at the time Bolivia, although Uruguay is an exception) than in countries that introduced the multipillar model (Chile; Argentina's reform in 1994 was too recent for any impact to be apparent in the data shown). As explained above, however, even reformed pension systems 94 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 5.3 Pension Income Increases Inequity Relative to Earned Income from Labor 1.4 Pensions Primary labor 1.2 1.0 elasticity 0.8 0.6 income Gini0.4 0.2 0 Argentina Bolivia Brazil Chile Paraguay Uruguay Venezuela Note: Data from 1995. Gini elasticity less than 1 indicates redistributive impact. Source: Wodon (2000), using data from 1995. with large funded pillars will contribute to inequitable outcomes when a substantial portion of the labor force is not covered by the system. Whether a pure PAYG system or a multipillar system with funded in- dividual accounts and a minimum guarantee, pension systems that condi- tion eligibility on a history of explicit contributions but pay benefits that are guaranteed by government transfers can redistribute income from all current and future taxpayers to those who have accumulated rights. This can lead to inequitable outcome in countries where most workers are not covered by the pension system. Even where the contribution and benefit parameters of a pension system are set to be "self-financing," government (society) still pays for shortfalls between benefits and contributions during economic downturns and for indexation to protect the real value of bene- fits during periods of inflation. Thus all current and future taxpayers "contribute" in one way or an- other to maintain the number and value of pension benefits paid to a relatively smaller group of "covered" workers. This makes increasing SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 95 Box 5.1 Defining the Coverage Problem: What Is It and Why Do We Care? At the turn of the 21st century fewer than 15 percent of the world's 6 bil- lion people had access to a formal system of retirement income support (Holzmann, Packard, and Cuesta 2000). The majority of this population that goes without formal coverage lives and works in developing coun- tries. To the extent that households in these countries can still rely on tra- ditional arrangements to provide income security in old age, there may be little cause for worry. However, with the rapid aging of populations, urbanization, and economic development, many governments are in- creasingly concerned that their elderly people will have reduced access to traditional safety nets. Extending the coverage of social insurance there- fore has become a policy priority. Retirement pensions are typically the largest component of the set of public interventions that make up a social insurance system. Low cover- age by the formal old-age pensions system, therefore, usually mirrors low coverage by other forms of social insurance, such as income support dur- ing unemployment, disability, access to family allowances, and in many cases access to public health care. Low social insurance coverage presents several problems (Barr 1998, 2000). First and foremost is the problem for the individual and by exten- sion the household. Workers who do not contribute to formal social in- surance, either by choice or because of market or institutional barriers, are not accumulating rights toward the receipt of benefits should they be- come unemployed or disabled or when they lose their ability to work in old age. Nor are their dependents covered should the workers suffer un- timely death. Household members may find it difficult to cope with losses resulting from these risks and can be forced into poverty when losses from an adverse shock are great. Second is the problem for society. An individual's failure to save or insure imposes an externality. If he or she chooses to make no provi- sions for the risks to income, the costs of that decision fall on others. In countries where a significant number of workers fail to insure, govern- ments face a "Samaritan's dilemma" in that politicians cannot credibly refuse to come to the aid of a large number of people who suffer a loss, and the burden of these losses can mount rapidly on current and future taxpayers. Finally, low levels of coverage pose a problem for the social insur- ance institutions themselves. Low coverage can weaken a traditional PAYG pension system if not enough active workers and employers contribute to finance the benefits of the inactive retired, disabled, or un- employed population. If a substantial share of the population is not contributing, the system cannot pool risks efficiently and can quickly (Box continues on the following page.) 96 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 5.1 (continued) lose financial viability. Similarly, where social insurance includes indi- vidual retirement accounts, the savings that can arise from scale in fund management are difficult to obtain when a large number of workers do not participate; persistently high administration costs eat into the sav- ings of those who do participate. The growing "informal," unregulated sector in many developing countries is important in the analysis of coverage to the extent that infor- mal employment opportunities lift the constraint on choices by allowing individuals to avoid government mandates to pool risks or save in the for- mal retirement security system. We define "coverage" both as a "stock" and a "flow" concept. The stock of the population that is covered includes all those people of retire- ment age and older who are receiving a formal retirement pension. The flow are those individuals of working age who are members of the work- force and currently are accumulating rights toward a retirement pension, either by contributing within the parameters of a PAYG benefit formula or by regularly depositing savings into a private individual retirement ac- count. Thus, the "coverage gap" also has stock and flow dimensions. The stock consists of the current mass of elderly people (of most concern, eld- erly poor people and those living close to the poverty line) with no formal income protection. The flow consists of the likely stream of current active workers who would fall into the former category year after year. Evi- dence shows, for example, that Peru's coverage gap is particularly wide. Source: Packard (2002). coverage of formal retirement security arrangements (and indeed the wider social insurance system) critically important for individuals and households that need instruments to manage the risks to their income that arise in old age (see box 5.1) and to eliminate institutional determinants of inequality. Has Reform Increased Coverage? Low rates of coverage of the working population under pure PAYG systems were a strong motivating factor for pension reform (box 5.2). By tightening links between contributions and benefits (second pillar) and cutting the pure-tax component of payroll deductions, the reformed systems were ex- pected to eliminate a substantial labor market distortion arising from pure defined-benefit PAYG regimens, lower incentives to evade mandated con- tributions, and encourage greater formalization of the workforce. In most developing countries only a small share of workers sell their la- bor in a regulated "formal" sector that is subject to a mandated minimum SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 97 Box 5.2 Was Increasing Coverage an Objective of Pension Reform? Pension reform literature and policy discourse have been consistently crit- ical of the low coverage levels of purely public PAYG pension systems and the distortions they introduce to labor markets, particularly in develop- ing countries. Because pension benefits in defined-benefit PAYG systems often have little relationship to mandatory contributions, both workers and firms can view contributions to these systems as a tax rather than as savings. In developing countries with dual labor markets, this perceived tax creates incentives for evasion, thereby reducing participation in the pension system and lowering coverage levels. Furthermore, the high pay- roll tax rates required in the formal sector to keep the system solvent re- strain labor demand while the incentives for early retirement in many PAYG systems reduce labor supply. Increasing coverage by reducing flight to the informal sector was con- sidered more than just a fortunate consequence of introducing "multipil- lar" pension systems with a large privately funded component. Increasing coverage has been presented as a core objective of the multipillar model. In 1994 the World Bank's Averting the Old Age Crisis presented clear evidence of the impending insolvency, inequitable benefits, and unac- ceptably low coverage rates of purely PAYG systems in the developing world. Among the "main aims of any structural reform of a pensions system" is "to increase the incentives to participate" (Devesa-Carpio and Vidal-Meliá 2002, p. 9) and stem the flow of workers to the informal sector. The result is greater coverage and a more efficient labor market. Indeed, reductions in "effective tax rates, evasion and labor market dis- tortions" are presented among the objectives and principal benefits of the multipillar reform model (World Bank 1994, p. 22). It has been extensively argued that reforming from a single-pillar, PAYG, defined benefit system to a multipillar system with a fully funded, defined contribution pillar achieves this increase in coverage by reestab- lishing the broken link between contributions and benefits. The link is particularly crucial in reducing evasion in developing countries: "When escape to the large informal sector and other means of evasion are easy, it is more important than ever to . . . link benefits closely to [payroll-based pension] taxes" (World Bank 1994, p. 320). If workers' pension benefits depend on the contributions they make during their working lives, then there will be "positive effects on workers' incentives to participate in the formal sector" (Mitchell 1998, p. 15). In addition to its salutary effect on coverage, the multipillar approach also "is designed to reduce labor market distortions" caused by public PAYG systems (James 1997, p. 10). On the demand side, broadening the tax base, shifting the tax burden to the worker, or both reduces payroll (Box continues on the following page.) 98 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 5.2 (continued) taxes paid by firms, thus increasing "employers' ability to hire and keep their employees" (Mitchell 1998, p. 15). On the supply side, linking benefits to contributions not only induces workers to leave the informal sector, but also "encourage[s] people to remain in work longer" by re- moving the incentives to early retirement present in many PAYG systems (Disney and Whitehouse 1999, p. 30). In short, literature on pension reform clearly states that increasing coverage is both an objective and a predicted result of implementing a multipillar system with a large pri- vately funded component. Source: Contributed by Todd Pugatch. wage and is covered by a social security system. The remainder work in an unregulated, uncovered "informal" sector where wages are solely deter- mined by the market and where both workers and employers escape the mandate to contribute to social security.2 A country's social security insti- tutions can determine the allocation of labor between the sectors because social security contributions are one of the main components of nonwage labor costs. Thus "informalizing" production allows firms to reduce their costs. In Latin America the costs imposed by social security are estimated to be as high as 20 percent of the operating expenses of small firms (Tokman and Martinez 1999). Theory suggests that at the margin a higher contribution rate for social security distorts labor allocation if workers do not consider their contri- butions "appropriable" in the future at the market rate of interest (Corsetti 1994; Schmidt-Hebbel 1998a). When the link between man- dated contributions and perceived benefits is ambiguous, social security acts simply as a tax on labor (Atkinson and Stiglitz 1980; Summers 1989). In the case of public pensions where the payoff to workers' "investment" in the system lies far in the future, this perceived tax can be even more onerous if discount rates are high and access to credit is constrained (Samwick 1997; James 1999). In addition, in many developing countries public institutions like social security lack credibility (i.e., workers may strongly believe that they will receive no pension at all), thereby further in- creasing the perceived tax burden of current contributions (James 1996). Several authors have shown that the extent of distortion to the labor market is independent of whether a country opts for a purely public PAYG system or for private individual retirement accounts (Diamond 1998; Barr 1998; Thompson 1998; Barr 2000). Corsetti (1994) found that to the extent that workers link current contributions to future pension benefits at the margin, individual retirement accounts do not necessarily produce fewer labor market distortions that determine the size of the informal sector than SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 99 does a public PAYG system. In fact, when contributions and benefits are ac- tuarially linked there may be more income incentives to work in the formal sector under a PAYG regimen than in a fully funded system. Orszag and Stiglitz (2001) and Barr (2000) presented similar arguments. Corsetti (1994) acknowledged that although the link between contributions and future ben- efits is unambiguous in a system of individual retirement accounts, such an actuarial balance must be carefully built into the design of the benefit for- mula for a PAYG system. James (1997) stressed this point, showing that rarely do PAYG formulas clearly link benefits to contributions, and even when they do, the balance is frequently upset by demographic and political pressures, especially in developing countries. Rather than enter into this de- bate, in this section we focus on studies that attempt to measure the impact of Latin American pension reforms that were expected to lower labor mar- ket distortions and improve workers' incentives to contribute. Have the in- centives to participate in formal retirement security systems been improved by reforms? And has the share of the working population covered against the loss of earnings ability in old age increased? Attempts to arrive at a common, cross-country indicator for coverage of social security systems have been confounded by differences in legisla- tion and institutional structures. Coverage may be determined by citizen- ship, residence, or income status, or it may be restricted to workers who make contributions for a minimum number of years (Palacios 1996; Pala- cios and Pallares-Miralles 2000). In Latin America and the Caribbean re- gion 25 countries legally require that all salaried workers contribute to be covered.3 Despite legislation, however, numerous opportunities for unreg- ulated employment and limited enforcement capacity allow large segments of the working population to escape these mandates. Figure 5.4 tracks the evolution in the most commonly used indicator of coverage--the share of the economically active population accruing rights by contributing to the national social security system.4 In several cases the time series shown in Figure 5.4 match the 1995 cross-section data on contributors in the labor force reported in Palacios and Pallares-Miralles (2000) for selected Latin American countries and years during the 1990s. Coverage is higher (between 30 percent and 60 percent of the economically active population (EAP)) in relatively affluent countries like Argentina, Chile, Colombia, Mexico, and Uruguay5 and lower (between 10 percent and 20 percent) in poorer countries like Bolivia, El Salvador, and Nicaragua. It is interesting to note the relative stability in the share of the economically active population that contributed to social security throughout the long recessions of the 1980s. Another trend that is evident is the rise in contributors with economic growth in the early 1990s in all but three of the countries shown. The most dramatic change in labor force participation in the retirement security system has been in Chile, where after the introduction of manda- tory individual retirement accounts in 1981, and until only recently, the 100 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 5.4 Has Participation Increased? There Is No Clear Pattern in Data on Contribution to National Pen- sion Systems (Percentage of EAP) 70.0 system 60.0 50.0 security 40.0 retirement 30.0 to 20.0 10.0 Contributions 0.0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 Argentina Bolivia Brazil Colombia Chile Costa Rica Ecuador El Salvador Mexico Nicaragua Uruguay Source: National pension agencies. share of contributors to the retirement security system (the statistic pre- sented shows all branches, including the closed PAYG plan in the separate regimens for police and the military) climbs steadily, but only to reach a level similar to that in 1980. Also noteworthy, however, is the dive in the share of contributors in Argentina since 1989, five years prior to the in- troduction of individual accounts in that country. Almost a decade after reforms--two decades in the case of Chile--the evidence of a change in the levels of coverage (proxied by contributors in the workforce) that is attributable to the introduction of individual retire- ment accounts has been mixed. It is difficult to discern a pattern from the data on contributors in figure 5.5. Several studies found that the share of the Chilean workforce covered by the reformed system has increased SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 101 (Corsetti and Schmidt-Hebbel 1994; Schmidt-Hebbel 1998b; Edwards and Cox-Edwards 2000a; Corbo and Schmidt-Hebbel 2003). Others claimed that there has not been an improvement in incentives and that there has even been a fall in the share of workers who contribute (Cortazar 1997; Arenas de Mesa 2000; Arenas de Mesa and Sanchez 2001; Mesa- Lago 2001). Palacios and Pallares-Miralles (2000) showed that in a large and varied sample of countries the share of contributors to a formal pen- sion system in the labor force is almost entirely explained by income per capita and varies little by the type of pension system in place.6 Until very recently, however, most studies relied on simulations, casual observation of data on labor force participation, or single-variable analy- sis. Furthermore, most analysis of participation in the pension systems fails to control for macroeconomic conditions or for policy variables un- related to social security that affect the labor market. Despite the time that has passed since reforms were implemented in several countries, the ex- pected improvement in incentives attributable to the introduction of indi- vidual retirement accounts has not been rigorously tested. In a background paper for this book, Packard (2001) employed the model presented by Edwards and Cox-Edwards (2000a) to estimate the impact of a transition from a purely public PAYG system to one with pri- vately managed individual accounts on the share of the workforce that contributes to retirement security systems. The aim was to identify the in- centive effect on participation expected by the proponents of reform by controlling for development, cyclical, and country-specific features. The results of panel analysis on 18 countries observed from 1980 to 1999 in- dicate that, after controlling for level of development and for the impact of the economic cycle, introducing individual retirement accounts has a positive but small incentive effect. These results are reported in table 5.1. The results also suggest that the extent of private provision influences the level of participation: the greater the share of mandated payroll con- tributions that accumulates as private savings, the larger the portion of the workforce that contributes after reforms.7 However, this increase in contributors may only occur gradually after individual accounts are intro- duced, as employers and workers overcome uncertainty and become fa- miliar with the set of new institutions that reforms put in place.8 It should be pointed out that the results do not rule out improved incentives from simply aligning contributions and benefits within a PAYG system (as with reforms in Brazil and the establishment of "notional" accounts in several Eastern European countries), without having to incur the costs of intro- ducing individual accounts. In accompanying background papers, Packard (2002), using household data from Chile, and Barr and Packard (2002), using data from Peru, showed that the contribution density (the share of their working lives in- dividuals have contributed to a formal pension system) of workers who joined the labor market after the introduction of individual accounts is 102 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 5.1 Evidence That Pension Reforms Positively Affect Incentives Pooled Random Fixed Independent variables OLS effects effects Income per capita 0.035 0.124 0.094 (0.016)** (0.025)*** (0.033)*** Life expectancy 0.011 0.009 0.007 (0.002)*** (0.003)*** (0.003)** Change in unemployment 0.428 0.246 0.263 (0.203)** (0.072)*** (0.070)*** Female labor supply 0.473 0.059 0.085 (0.144)*** (0.265) (0.315) Total payroll tax 0.36 0.388 1.189 (0.107)*** (0.217)* (0.355)*** Portion of payroll tax going 0.272 0.057 0.153 to individual account (0.110)** (0.064) (0.076)** Years since reform 0.012 0.025 0.025 (0.024) (0.009)*** (0.008)*** Years since reform2 0.002 0.004 0.004 (0.004) (0.001)*** (0.001)*** Years since reform3 0.00009 0.0001 0.0002 (0.0001) (0.000)*** (0.000)*** Year dummies included? yes yes yes Observationsa 287 287 287 Number of countriesa 18 18 18 R-squared 0.53 0.29 0.31 (within) (within) F test of joint significance of fixed effects, H0: OLS accepted F(17, 241) = 131.50 P > F = 0.0000 Hausman specification test, H0: difference in random and fixed effects not systematic 2 (28) 27.68 P 2 0.4816 Notes: Contributors to national pension system as share of EAP, various estima- tions. OLS ordinary least squares. Standard errors in parentheses. Specification tests indicate that the fixed effects estimator in the last column is preferred. a Colombia removed from the sample; 20 observations dropped. * Significant at 10 percent level. ** Significant at 5 percent level. *** Significant at 1 percent level. Source: Packard (2001). SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 103 significantly greater than that of workers who contributed to the pension system prior to reforms. Although the reported analysis found evidence of an improvement in incentives attributable to the shift to individual retirement accounts, other factors directly and indirectly linked to the reformed pension sys- tems are found to determine rates of participation among workers. In another background paper for this inquiry, Valdés-Prieto (2002c) showed that a broader set of policies have a significant and substantial impact on coverage. Changes in labor and social legislation that impose costs on par- ticipation in the formal labor market or confer benefits to covered workers affect the overall coverage of the pension system. Using time-series analy- sis of the labor force in Chile that contributed to the pension system from 1990 to 2001, Valdés-Prieto presented evidence that the share of con- tributing workers in the labor force falls with increases in the minimum wage. This effect did not result from workers being excluded from formal employment--as has been argued in the past--but because increases in the minimum wage also increase the minimum that workers are mandated to save in the AFP system. This increases the amount that those who con- tribute only the minimum are forced to save. Furthermore, increases in the flat commissions charged by the AFPs also lower the number of workers who contribute. Finally, participation in the pension system increases with increases in the subsidy the government pays to cover the health expenses of workers who contribute to Chile's public health system. Conclusion Despite the positive impact of pension reforms on workers' incentive to seek coverage, the share of workers who contribute to the formal retire- ment security systems is still low relative to OECD countries, even in the wealthier countries in Latin America that offer individual accounts. Totals reach barely above 65 percent in a few cases. That finding warns strongly against complacency. The relatively small share of contributors in the la- bor force indicates that the wedge created by the payroll tax to public PAYG systems prior to reform was just one of many possible factors that still lead certain groups of workers to turn away from government-mandated retirement income protection. In analyzing coverage outcomes--that is, the share of the new elderly population who receive pension benefits each year--Rofman (2002) found that coverage is falling in Argentina. Fewer elderly people among every new cohort in Argentina receive pension benefits. Furthermore, whether one considers participation rates or the incidence of pension ben- efits, coverage indicators show a regressive pattern (see table 5.2), and in- dicate that more needs to be done to increase access to at least minimum levels of protection against the risks to income that accompany old age. 104 Table 5.2 Participation Rates and Incidence of Benefits Show a Regressive Pattern of Coverage by Formal Pension Systems a. Contributors as a percentage of population aged 15 to 65 years Income KEEPING Group Argentina Bolivia Chile Colombia Costa Rica El Salvador Nicaragua Peru Venezuela Total 36.0 10.3 62.7 22.3 22.9 25.5 11.0 13.0 33.1 Quintile 1 18.7 0.6 43.4 9.8 8.9 3.6 -- 1.6 14.4 THE Quintile 2 14.8 2.5 57.4 10.3 25.2 7.3 -- 6.1 28.8 Quintile 3 34.7 9.2 64.3 16.1 25.5 18.2 -- 10.6 32.9 PROMISE Quintile 4 47.4 14.6 67.9 26.0 27.0 34.6 -- 17.0 37.7 Quintile 5 58.4 25.8 72.0 42.1 23.2 49.5 -- 25.1 44.3 b. Pension recipients as a percentage of population older than 65 years of age OF Income Dominican SOCIAL Group Argentina Bolivia Chile Colombia Costa Rica Republic El Salvador Peru Venezuela National 66.2 11.9 41.4 15.2 33.2 14.6 8.8 23.5 9.4 SECURITY Quintile 1 54.1 0.3 15.0 0.2 25.5 10.7 1.6 8.2 2.2 Quintile 2 66.5 6.3 41.6 12.4 31.5 10.8 9.3 17.5 6.1 Quintile 3 70.4 13.3 50.2 15.8 41.0 17.1 11.1 26.2 9.7 Quintile 4 70.0 28.9 57.4 22.8 48.3 19.8 11.4 34.4 15.8 IN Quintile 5 71.2 29.0 55.0 32.5 35.0 19.2 20.6 41.8 16.1 LATIN -- Not available. Source: National household surveys (Argentina, EPH 2002; Bolivia, ECH 2000; Chile, CASEN 2000; Colombia, ENH 1999; Costa Rica, EHPM AMERICA 2000; Dominican Republic, ENFT 2000; El Salvador, EHPM 1998; Mexico, ENEU 2001; Nicaragua, ENMV 1999; Peru, ENAHO 2000; Venezuela, EHM 2000). SOCIAL GAINS FROM PENSION REFORMS IN LATIN AMERICA 105 Notes 1. For the assumptions employed in the simulations presented here, see the technical annex at the end of this volume. 2. Recent evidence (Maloney 2003) shows that factors favoring the nontrad- able sector, such as recessions and currency appreciations, increased the size of the informal sector in the 1990s. 3. Only 13 countries require self-employed workers to pool risks or save along with the rest of the working population, whereas 10 countries invite self-employed workers to participate on a voluntary basis. Self-employed people are required to contribute in Argentina, Brazil, Cuba, Honduras, Uruguay, and Venezuela. Almost all the countries in the Caribbean region require that self-employed workers par- ticipate in the national social security system, with the exception of Antigua and Barbuda. Participation is voluntary in Bolivia, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Mexico, Nicaragua, Panama, and Peru (Mesa-Lago 2000). 4. This indicator is second best because it does not capture coverage of public (or publicly mandated) disability and survivor insurance, which often extends for a determined period after a worker stops contributing toward a retirement pension. The share of contributors in the workforce at any given point in time also excludes workers who may have contributed in the past and acquired some rights, and it fails to take account of rights of dependent spouses and children in the workforce. Although working spouses and children may not contribute themselves, they are likely to be covered for survivor and health risks through the contributions of the head of household. 5. With the option to switch between the defined contribution AFP and the defined-benefit PAYG system every three years, the data from Colombia may count contributors twice and overstate the rate of coverage, thus data should be inter- preted with caution. 6. Although the correlation presented by Palacios and Pallares-Miralles (2000) is indeed high, one would expect analysis of a large sample of diverse coun- tries to show that per capita income explains most of the variation in almost any indicator of interest, whether it is contributors to a pension system, education, longevity, or household ownership of consumer durable goods. 7. An important caveat to this result should be made with respect to Ar- gentina. After the 1994 reform workers could choose between a reformed PAYG system and private accounts for the earnings-related portion of their pensions. To increase the participation in individual accounts, the government assigned workers who did not make an explicit choice to the private branch of the system by default. In a background paper for this book Rofman (2002) found that as many as 2.8 mil- lion affiliates to the private branch of the system never explicitly chose the private option. Although this could affect the results cited above, the dependent variable in the panel analysis is the share of pension-system contributors among workers, not the share of affiliates. Furthermore, each reforming country was dropped from the panel one at a time to see if the positive incentive results were particular to any single reform. The results withstood this experiment. 8. Packard (2001) found evidence of a J-curve effect in the share of workers who contribute after the introduction of individual accounts, but the effect was very weak and not robust to small changes in the sample. II Analytical: How Governments Can Help Individuals Deal with Retirement 6 How Individuals View Social Security CHAPTERS 3, 4, AND 5 PRESENTED EVIDENCE showing that countries in Latin America that adopted the multipillar model--distinguished best by the presence of mandatory savings accounts--experienced improvements in the economic sustainability and socioeconomic orientation of social security systems and have increased financial sector depth. These outcomes were often the stated goals of the reform, and the evidence shows that countries have been successful to varying degrees. But it is fair to ask if these objectives should be ends in themselves or simply means to improving the welfare of citizens. Having a fiscally healthy government, a well-functioning labor market, and a growing financial sector are important because they can make individuals and their families better off. But the main question is, Are people in the reforming countries better off because of the adoption of multipillar systems of so- cial security? This is the question addressed by part 2 of this volume. This chapter presents a simple analytical framework centered on the behavior of an individual--a rational "representative agent"--to help in answering this question. The theory yields important clues for the role of government in a setting where individuals make decisions under risk; the implications are similar to those drawn by Barr (2000; see box 6.1). Chapter 7 provides evidence that Latin American workers may not judge the new system quite as kindly as do fiscal and financial specialists. Chapter 8 provides both an argument that Latin American workers deserve to be trusted more in deciding how best to save for their old age and some empirical clues to how governments can best help workers address the risks associated with aging. It is important to stress here that the framework used is not novel; it is borrowed from the economics of insurance literature, building on the sem- inal work of Ehrlich and Becker (1972, 2000).1 The novelty lies only in ex- tracting the implications for public policy. 109 110 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 6.1 The Welfare State as a Piggy Bank to Reduce Old-Age Income Insecurity Barr (2000) proposed a simple and elegant rationale for the role of government in a real-world scenario of risk and imperfect information. Using this simple Fisher framework of individual choice, Barr derived the principles for realistic expectations from and effective design of pension systems. Contrasting these principles with thinking inspired by the Chilean reforms, Barr listed 10 myths concerning the macroeconomics of pensions, the design of pension systems, and the role of government: 1. Funding resolves adverse demographics. 2. The only way to pre-fund is through pension accumulations. 3. There is a direct link between funding and growth. 4. Funding reduces public pension spending. 5. Paying off debt is always good policy. 6. Funded schemes have better labor market incentive effects. 7. Funded pensions diversify risk. 8. Increased choice is welfare improving. 9. Funding does better if real returns exceed real wage growth. 10. Private pensions get government out of the pensions business. It would be fair to say that chapters 3, 4, and 5 provide considerable evidence from Latin America during the last two decades that supports the argument that the elements of Barr's list are genuine myths, not self-evident truths. This volume also takes seriously Barr's advice that analysis of pensions "needs to draw on several perspectives, including microeconomics, macroeconomics, financial economics, and an understanding of the theory of social insurance" (2001). Chapters 3, 4, and 5 address the first three aspects; this chapter applies simple principles of insurance to the economics of pensions. In applying the basic principles of insurance to the problem of social security, this volume characterizes the challenges associated with pension design as the matching of instruments to the two fundamental losses associated with aging in a world of increasing prosperity: the loss of the ability to earn in old age and the loss associated with poverty during old age. Changing demographics imply that the former is an increasingly frequent loss, and the greater availability of saving instruments for old age imply that the latter is an increasingly rare loss. The economics of insurance suggest that frequent losses cannot easily be insured against (or only at prohibitively high premiums) but that rare, idiosyncratic losses can be pooled. The preferred risk management strategy for frequent losses is a combination of saving and measures to reduce risk. HOW INDIVIDUALS VIEW SOCIAL SECURITY 111 Box 6.1 (continued) Our approach differs from that of Barr in one respect. Whereas Barr treated the poverty relief role (the "Robin Hood" function) of government as distinct from its role in redistribution over the life cycle (the "piggy bank" function), we subsume the responsibility of old-age poverty relief within the set of decisions associated with the life cycle. As a result of this we arrive at a "size of government" that is somewhat different: government­defined benefit pensions serve only a poverty-relief function. But expanding the insurance function of government to cover poverty relief actually strengthens Barr's argument that the "welfare state is here to stay" (2001, p. 1) because the insurance function of governments will not diminish as economies grow, and may indeed increase. Source: Barr (2000, 2001). The Advantage of Taking the Individual's Viewpoint The attractiveness of an approach that is centered on the individual is not just that it refocuses attention on what really matters in designing public policy--that is, the welfare of individuals and their families. Another attractive feature is that it uses a well-developed literature on the economics of insurance to draw implications more carefully rather than attempting to build a conceptual framework from scratch. There are two main payoffs from the improved analytical rigor that comes from this overtly incremental approach to conceptualization of old-age income security. First, it brings the appropriate role of government into sharper relief and, second, it reliably points at directions in which policymakers should take their countries to provide better income security in old age. "Better" here is understood to mean sustainable, secure, and sufficient support for more people. The next section in this chapter provides the reasons why saving and not pooling should be the main instrument for dealing with the loss of earning capacity during old age, that is, smoothing consumption over the life cycle. Using the same approach we discuss the rationale for having a pure pooled plan as well--to deal (largely) with the risk of poverty in old age. Readers will see the potential payoff in policy formulation: it provides some guidance on the relative sizes or "weights" of the pooling and savings components in the multipillar system of any country, depending on meas- urable parameters such as life expectancy at retirement and the incidence of old-age poverty. Unlike much of the pension literature that assumes at the outset that in- dividuals do not behave rationally (e.g., are reluctant to prepare for old age), 112 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA we begin with assumed rational individuals who maximize lifetime utility for themselves and their families. We then extend the analysis to deal with a commonly assumed market failure--that individuals may not save "enough" for old age. Using a paper commissioned for this book (Valdés-Prieto 2002b), we discuss the basis for benevolent policy--a justification for a government- imposed mandate on individuals to save for old age. The analysis yields use- ful guidance on the relative sizes of the mandatory and voluntary pillars of the savings component. Finally, we examine another widely assumed ration- ale for a government-imposed mandate to save for retirement--that savings are somehow "underproduced" at an aggregate level when the benefits of saving for financial sector development and growth are considered. Saving as the Mainstay of Old-Age Income Security The principles of insurance are clear on the matter of income security in old age. In the face of a possible loss, a "comprehensive insurance" approach would suggest that individuals can insure against the loss, take steps to lower the likelihood that the loss will occur, or do nothing and simply "take their lumps" (see, e.g., Ehrlich and Becker 1972, 2000; and box 6.2). The purchase of insurance transfers income from "good" to "bad" times to lower the size of losses in the latter. Individuals can insure themselves in two ways: through mechanisms that pool the risk of the loss occurring among those who are exposed to this risk, or through individual savings or "self-insurance." The cost of pooling risk (equal to the insurance premium, assuming an actuarially fair price) is set according to the probability that the loss will occur. Thus, as the probability of a given loss rises, the cost of pooling to insure against that loss will also increase. In contrast, the im- plicit or shadow price of saving--the other way to transfer income from good times to bad--does not vary with the probability of the loss. As the likelihood of the loss increases, the price of pooling risk relative to the price of saving will rise. Rational individuals confronted by actuarially fair prices will then prefer saving to pooling to insure against the loss. The loss in question here arises from the inability to earn an income because of the body's natural deterioration in old age. As life expectancies rise as a result of improvements in basic hygiene, medicine, and education, the probability that most people will face a period of life in which they will need to consume but be unable to work can also rise. As such a state of the world becomes more likely, rational individuals should increasingly turn to saving to smooth consumption over their lifetimes and step up "self- protection" efforts to lower the impact of this loss of earning capability. If individual preferences are not argument enough, the same logic can be applied at the aggregate level to support a transition to individual savings accounts. Improvements in longevity increase the share of the population that faces a relatively predictable loss. Pooling risk by defining HOW INDIVIDUALS VIEW SOCIAL SECURITY 113 Box 6.2 A Theory of "Comprehensive Insurance" In the Ehrlich and Becker (1972, 2000) characterization there are two states of the world: bad and good. Faced with risk a person may purchase insurance that involves paying a premium. The individual also spends resources on self-insurance (transferring income from good to bad states by themselves) and self-protection (lowering the probability of the bad state). The individual chooses the levels of market insurance and self- insurance where the price of market insurance (the premium) equals the shadow price of self-insurance. Resources spent on self-protection are optimized where the marginal gain from reducing the probability of loss equals the marginal loss from having to pay for self-protection. The framework has several implications for pension policy. First, market insurance and self-insurance are substitutes. As the price of market insurance rises relative to self-insurance, the person will prefer the latter as a way to transfer resources from the "good" state of the world (working age) to the "bad" one (old age). Second, the individual will prefer market insurance to self-insurance for insuring relatively rare losses because the shadow price of self-insurance does not fall as the probability of the loss decreases, but the price of market insurance does. Third, although imperfect information or weak institutions may result in a thin or missing market for insuring against some losses, and govern- ments would feel justified in stepping in to address these "market fail- ures," the basic principles of insurance still apply: where the losses are relatively rare the preferred mode of intervention should involve pooling of risks; where losses are frequent the emphasis should be on providing or facilitating instruments for self-insurance or saving. Finally, although the possibility of adverse selection necessitates a mandate to participate in a public plan that involves pooling (namely, that mimics a market insurance plan), the rationale for mandatory participation in a private savings system is much weaker; that is, government-provided instruments for pooling imply a strong insurance-related rationale for mandating par- ticipation, but the rationale for coerced participation in a self-insurance scheme is based on weaker reasoning (e.g., moral hazard in the presence of a public pooled plan and political economy reasons in the presence of an existing--and inappropriately designed--social security system). Source: Ehrlich and Becker (1972, 2000); de Ferranti et al. (2000); Packard (2002). retirement benefits financed on a PAYG basis will become more expensive relative to individual savings as the number of old people rises relative to the number of young people. Although advances in health can postpone the loss of earnings ability, in most countries the legal retirement age has not risen commensurately. Little wonder then that social security systems 114 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA that relied on pooling to deal with this increasingly frequent loss ran into trouble unless they altered the very definition of the risk in question to keep it relatively rare, for example, by raising the age at which retirement benefits commence as retirees' life expectancy increased. The main point is that if the loss of earnings ability while living is wide- spread--as it is in much of Latin America--simple economics of insurance dictate that saving, not pooling, is the appropriate insurance mechanism to smooth consumption across these "states of the world." All this is not to say that there is no room for a pooled scheme. Pooling is the appropri- ate instrument for the rarer losses associated with disability and untimely death--in the terminology of Ehrlich and Becker (1972, 2000), "market insurance." So the reliance on saving for old-age income security to smooth consumption does not imply that there is no room for pooling instruments such as disability and survivors' insurance and annuities. For simplicity, however, we refer to the consumption-smoothing part of a pension system as the savings component. In a useful survey of the gains from pension reform around the world, Lindbeck and Persson (2003) identified the "individualization" of social security as a prominent feature of reform efforts, reflected in a shift to individual accounts, either notional or real. But that survey ended by em- phasizing that these "reforms do not diminish the need for basic, or guar- anteed, pensions. Quite the contrary; growing reliance on quasi-actuarial and actuarially fair systems, which in themselves do not encompass any systematic intra-generational redistributive elements, makes it even more imperative to maintain a safety net to prevent poverty in old age" (p. 109). We discuss this issue next. Pooling to Insure Against the Risk of Old-Age Poverty In contrast to the incidence of old age, and assuming rising incomes, the prospect of poverty during one's retirement years will become relatively rare in Latin America over time. For this reason the cost of insuring against the risk of poverty will become relatively low, providing justification for defined benefit programs financed either through payroll or general taxes that pool the risk of old-age poverty among taxpayers. In the absence of sufficiently flexible statutory benefit entitlement ages, the extent of pooling in defined benefit plans increases over time as popu- lations age. As the cost of pooling increases, governments are forced to raise taxes to finance these systems. Such costs may be bearable for a pure pooling arrangement designed to prevent or alleviate indigence in old age. On the other hand, the growing cost of pooling arrangements, which are geared to addressing the consumption-smoothing motive of individuals, can overwhelm government finances. Defined contribution regimens (whether PAYG or funded) introduce an automatic rebalancing mechanism HOW INDIVIDUALS VIEW SOCIAL SECURITY 115 because benefits depend on the life expectancy of a given generation. The growing preference for defined contribution plans among employer- provided pension systems in OECD countries and the shift to notional defined contribution systems in countries such as Italy, Latvia, Poland, and Sweden are evidence of the increasing relative costs of pooling for old-age income security. The comprehensive insurance approach has immediate and obvious implications for the size of the first pillar (the pooling component, which has a poverty-prevention objective) as compared with the second and third pillars (the savings component, which has a consumption-smoothing, income-replacement objective) in dealing with risks related to old age, namely that it should be much smaller. Under reasonable assumptions the role of government in providing an instrument to pool against the risk of old-age poverty also emerges as important (asymmetric information), as does the need for mandating participation (adverse selection). But the analysis does not shed nearly as much light on the relative importance of the mandatory versus voluntary savings components, namely, the weights of the second and third pillars. It is to this issue that we turn next. Justifying Mandatory Saving--Individual Welfare Most fields within economics that seek to justify a role for government begin with the premise of individual rationality. The case for government intervention usually hinges on some failure of market mechanisms to induce individuals and firms to choose optimal levels of consumption and production. In the case of underconsumption or underproduction of something good--for example, vaccines against tuberculosis or automobiles that run on solar energy--the usual solution proposed is to subsidize its production or consumption. This "rule" arises from the principle that quantity re- strictions are more distortionary in terms of unintended effects than inter- ventions to manipulate prices. When an activity is mandated by societies (e.g., compulsory enrollment of children in basic education), the govern- ment is generally expected to provide this service at a lower price than could be asked by the market. Somewhat in contrast, pension policy discourses usually presume irrationality on the part of individuals. Where this is not assumed, exter- nal benefits to saving for old age are often asserted. And more often than not the prescription is to rely on mandated minimum savings levels instead of, say, providing fiscal incentives to induce individuals to save more than they would without such inducements. The result has to be that many peo- ple are coerced into saving more than they should (the distortion discussed above), given their tastes for current consumption or the needs of people at their stage in life. Young people and poor people are often the main 116 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA targets of such paternalistic policies, and they are most likely to end up worse off than had there been no interventions. We started this chapter with the assumption of rationality and arrived at the conclusion that under most circumstances, the mainstay of old-age income support should be saving. Now we examine the possible rationale for mandating this saving, relying on a paper commissioned for this book.2 Valdés-Prieto (2002c) examined the five most frequently used arguments for justifying a policy that mandates savings for retirement: myopia, moral hazard caused by the first pillar, incentives for intergener- ational abuse, adverse selection in annuity markets, and "improvidence" or a systematic mistake in assessing the length and cost of old age until it is too late for the individual to correct. After reviewing the evidence and internal consistency of these theories, only improvidence survives critical evaluation as a plausible basis for benevolent policy (see table 6.1). How- ever, as the gains from eliminating improvidence are bounded--after all, the case is that voluntary savings are low, not zero--this implies a limit on the social costs of the mandate that should be tolerated. Even pro-mandate or interventionist interpretations based on individual behavior imply a second pillar that is relatively small, with savings only enough to allow purchases of annuities that yield the same level of bene- fits as first-pillar pensions. The main rationale for the large second pillars that we observe in the countries that have adopted multipillar reforms must then be the existence of oversized first pillars. Presumably this is because people were accustomed to high replacement ratios in the old PAYG systems, and it was difficult for governments to lower this ratio suddenly. If this is the case, then a gradual decline in the size of the second pillar is all the more sensible because it will enhance the equity of both mandatory components--the defined benefit first pillar and the defined contribution second pillar. The implications for the third pillar are ambiguous. Given rational individual behavior, large first and second pillars imply little room for voluntary saving. The distortion in old pooling-dominated systems was a lack of emphasis on saving; the distortion in reformed systems may be a large second pillar at the expense of the third pillar. Justifying Mandatory Saving-- Economy-Wide Externalities The case for mandating saving for old age--that is, for setting up a second pillar--has also been made in a more circular manner than individual myopia or improvidence. Mandatory savings programs are believed to increase the economy-wide savings rate, thus increasing access to credit for firms that would otherwise be starved of capital. It is argued that a second pillar fosters capital market development and thereby provides better financial Table 6.1 Justifying a Government-Imposed Mandate to Save for Old Age--Microeconomic Reasons HOW Rationale Definition Evidence Consistency checks Implication INDIVIDUALS Myopia with Broadly defined as high No evidence of psychic Paternalism against Myopia, poor self- respect to preference for consump- externalities (that myopia inconsistent control, and presence of saving for the tion now relative to society would be made with democracy; people psychic externalities use- future or lack later; also includes better off if people are have voluntary instru- ful for studying individ- VIEW of self-control self-control failure protected from their ments (e.g., mortgages) ual behavior, but fall far own weaknesses) impor- to overcome the prob- short of justifying a SOCIAL tant enough to justify lem of self-control government second- or large second pillar third-pillar intervention Moral hazard The presence of a pooling Second pillar is usually Given presence of first Introducing a second SECURITY arising from plan or safety net far larger than this moral pillar, introduction of pillar to reduce moral the first pillar (e.g., first-pillar pension) hazard would justify-- second pillar makes hazard is not Pareto- implies perverse accumulation should poor workers worse off desirable because it incentive not to save not be greater than what makes poor people worse enough is needed to afford a off, and may even be pension of 1 or 2 times Pareto-worsening; third- the minimum salary pillar interventions likely to help the wealthy Intergenera- PAYG systems allow Timing of introduction of Assumes median voter is Not convincing enough tional abuse massive transfers from such systems--removal "egotistic" or selfish, to be taken as the in pooled future generations of of wealth requirements but same voter makes explanation for second- plans workers, but mandatory for voting--such that unselfish decisions about pillar pensions savings do not median voter was environment and public 117 manual worker support for basic research (Table continues on the following page.) 118 Table 6.1 (continued) Rationale Definition Evidence Consistency checks Implication Adverse If insurance provider has No evidence that workers Even if true, the interven- Probably not adequate KEEPING selection in less information than have more information tion should be limited to justification for annuity worker, only the most on their own longevity the annuity market, mandating saving; if markets long-lived people than do specialized rather than mandating true, choice between purchase insurance in insurance providers saving; implies no annuity and programmed THE voluntary markets government intervention withdrawal should not PROMISE because good and bad at accumulation stage, be permitted risks are all charged the only at dispersion stage same premium OF Improvidence Systematic mistake in Evidence raises doubts The best solution is to Educating workers about or systematic assessing the length and about individuals' provide information on old-age needs is best SOCIAL miscalculation cost of old age until ability to distribute life- correct length of policy; third-pillar insti- worker is too old to time wealth over time retirement tutions better suited for SECURITY rectify this mistake at because of psychological this because their suc- modest cost constraints cess depends on attract- ing voluntary saving Source: Authors' interpretation of Valdés-Prieto (2002b). IN LATIN AMERICA HOW INDIVIDUALS VIEW SOCIAL SECURITY 119 instruments for saving. The argument here is that having a second pillar stimulates the growth of the third pillar by hastening the growth of "institutional capital." Table 6.2 summarizes the different macroeconomic reasons for setting up a second pillar. These arguments again imply that an individual left to himself or herself would underproduce savings for old age, not relative to what would be optimal for the specific individual--the rationale discussed in the previous section--but compared with what would be optimal for the economy as a whole. This is again a market failure and it makes a case for government intervention. Paradoxically, this argument is generally made by people who distrust governments and ostensibly trust markets. And the case made is for quantity controls, not for price subsidization. This issue cannot be resolved except by appeal to data. Schmidt-Hebbel (1997) found some evidence that part of the increase in national savings observed in Chile can be traced to the pension reform, both directly through limited crowding out of voluntary savings and indirectly through capital market development and higher productivity growth. Most of the increase in savings, however, results from an increase in public savings (that were not fully offset by private dissaving) and by other structural changes, such as a comprehensive tax reform, that triggered a sharp increase in corporate savings rates. For three of the Latin American countries that have adopted multipillar reforms (Argentina, Chile, and Peru), Walker and Lefort (2002) examined whether there is evidence of links between mandated saving and capital market development. They found that some of the key improvements in the capital market infrastructure, such as the development of depositary and risk-rating services, have been linked intrinsically to pension reform. In addition, the regulatory oversight of the financial system, and in par- ticular of the new pension fund industry, has been modernized to increase the security of retirement savings. Pension funds can also contribute to capital market development by improving the liquidity of securities markets and lowering the cost of capital for firms. As shown in chapter 4, this evidence is clearest in Chile, where pension funds have become the largest minority shareholder of many companies traded in the stock market. Chilean pension funds also appear to be playing an increasingly important role in corporate governance by voting for independent directors and demanding greater transparency in company accounts. In other countries that permit pension fund investment in corporate securities, such as Argentina and Peru, too little time has passed since the inception of the system to extract conclusive evidence. The main recipients of the new retirement savings, however, have been governments and financial institutions. Pension fund administrators and especially life insurance companies have been avid buyers of medium- and long-term government bonds in the region and have contributed to re- ducing the dependency on foreign capital. The greater degree of market Table 6.2 Justifying a Government-Imposed Mandate to Save for Old Age--Macroeconomic Reasons 120 Rationale Definition Evidence Consistency checks Implication Savings The mandate increases Some evidence of increase in Sound fiscal management and Transition debt needs to be national savings because of savings as a result of pension inflation protection mecha- financed explicitly, and man- KEEPING limited crowding-out by vol- reform in Chile, but main nisms are central to promot- date needed to finance this untary dissaving in countries cause is government savings, ing savings; Chile is the debt, leaving little ground to with credit-constrained not mandate. exception in Latin America. increase savings. THE households. Capital mar- The mandate creates a pool of Some evidence of greater liq- The mandated savings are man- Capital market development PROMISE ket develop- long-term savings that is uidity of stock market in aged by a handful of pension can be achieved without ment channeled into the capital Chile as a result of pension fund managers who invest in mandating savings, the markets and contributes to fund investment; the mod- more or less the same way; United Kingdom and the OF their development. ernization of capital markets such concentration of deci- United States being the two SOCIAL is driven mainly by regula- sionmaking is not conducive prime examples. tory reform, not pension to dynamic markets and is funds. prone to political capture. SECURITY Economic Funds are channeled into pro- Most funds are channeled into Investment regulations deter- Mandated savings call for growth ductive investments; there is government debt and mine asset allocation; regula- strict prudential regulations a more efficient allocation of deposits, so there is little left tions limit investment in proj- that constrain innovative IN resources in the economy. for direct financing for the ects without history and investment. LATIN corporate sector. securities of lower credit rating. Labor market Social security contributions The new funded system is per- Lower contribution rates Labor market efficiency is best efficiency are no longer perceived as a ceived as more risky than increase participation and served by voluntary savings. AMERICA tax. alternative investments, so the improve compliance in tax element is not eliminated. mandatory pension systems. Source: Authors' compilation. HOW INDIVIDUALS VIEW SOCIAL SECURITY 121 discipline over government debt has not always been forthcoming (Ar- gentina is the prime example), but pension reform has been at least a cata- lyst for concomitant reforms aimed at improving the management of the government debt. The health of the banks' balance sheets has also improved with the influx of new capital. Both pension funds and life insurance companies have allocated a significant portion of their assets to deposits, thereby contributing to raising lending rates and lengthening the maturity of loans. In Chile pension funds and life insurance companies have also played a key role in the development of the mortgage debt market, thus contributing to development of the real estate market. In general the evidence supporting capital market development is clear- est in the case of Chile. This is certainly a result of the longer history of the system, which permits more accurate empirical estimations. At the same time some developments in Chile, such as growth of the mortgage debt market or the increasing maturity of bank loans, would have been impos- sible had inflation protection mechanisms not existed (fixed income mar- kets are indexed to a measure of the cost of living) and had there been no serious and credible policy of fiscal discipline, as argued in chapter 4. Conclusion The relevant point is that rare and idiosyncratic losses can be pooled or insured against, and frequent or systemic losses should be saved for, or self- insured against. With life more frequently extending beyond the point where earning ability declines and with more systemic loss resulting from the welcome development of increased longevity, the mainstay of old-age income security becomes self-insurance or savings. The role of government is to ensure that inter-temporal contracts are honored--hence the impor- tance of regulation of private long-term savings plans. By the same token, however, there are two risks that are rare and idiosyncratic and that justify pooling, whether facilitated by markets or by governments. The first risk is old-age poverty. This loss is difficult for markets to insure against because of moral hazard, potential adverse se- lection, and the social nature of the definition of poverty. Hence there is a clear argument for a mandated scheme--"first-pillar" pensions in World Bank (1994) terminology, "first-tier" pensions in Barr's (2000) words, and so on. The second risk is outliving one's savings because of unexpected longevity, which is rare by definition. So the retiree buys an annuity, which allows the transformation of the stocks of saving into re- tirement income flows. But while there may be a problem of adverse se- lection (where only those expecting to live longer buy annuities, and the others opt for withdrawing their pension savings in lump sums), there is no problem of moral hazard here, and the definition is technical rather 122 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA than social. This consumption-smoothing instrument is something the private sector can provide. There is, of course, the problem of adverse se- lection that may require help from the government in the form of a man- date to participate in annuity plans or may be addressed privately through employer-sponsored retirement plans. Here again the role of government is to ensure that inter-temporal contracts are honored and to offer inflation-indexed savings instruments to the annuity providers because some of the losses may result from inflation. Savings plans that are not mandatory but are tax advantaged are called "third-pillar" (World Bank 1994) or "third-tier" (Barr 2000) pensions. If the savings plans are mandatory, they are called "second-pillar" or "second-tier" plans, and the role of government in ensuring reasonable management and investment fees is (arguably) greater. But although these two components can be broadly characterized as savings, both have critical components that involve the pooling of risk in the form of annuity schemes. This is why in chapter 1 (see table 1.1) we argued that the main distinction among the three components of the multipillar system is not whether they are purely saving or self-insurance schemes, or purely pooling or market- type insurance. Rather, the distinction lies in the principal objective for the program and the role of government in ensuring that the objective is achieved: · In the first pillar the government defines the benefits to facilitate the sharing of losses associated with old-age poverty. · In the second pillar the government defines the contributions to ensure adequate consumption smoothing over the life cycle. · In the third pillar the government defines the incentives to encourage the maintenance of a reasonable standard of living during old age. To conclude, in a world of uncertainty and imperfect information one may summarize the interactions among workers, markets, and the gov- ernment as follows: Individuals save for old age and often are encouraged to do so by preferential tax treatment for long-term saving. This personal saving is the mainstay of old-age income security. Workers also buy in- surance from governments in case they do not save enough to stay out of poverty, or if they have bad luck with their savings. Upon reaching old age--defined broadly as losing the option of labor earnings--people have a lump sum of savings, but they face another uncertainty: they do not know how to spread this money over their remaining lives. So they insure against this lack of knowledge by converting the lump sum into a flow that ensures they will have retirement income even if they live longer than expected. Had they relied on individual rather than group solutions to address the problem of unexpected longevity, many would outlive their savings and others would leave behind larger bequests or estates than desired. HOW INDIVIDUALS VIEW SOCIAL SECURITY 123 Notes 1. Gill and Ilahi (2000) employed the Ehrlich-Becker concept of "comprehensive insurance" to address the potential role of government in augmenting individual efforts to lower the probability of losses (self-protection), to pool risks (market insurance), and to undertake precautionary savings (self-insurance). De Ferranti et al. (2000) used this extended framework to structure a discussion of economic risk management in Latin America and the Caribbean region. Packard (2002) used the comprehensive insurance framework to analyze the losses associated with old age. 2. See Valdés-Prieto (2002b). This reliance notwithstanding, the views expressed in this chapter are not necessarily those of the author of this background paper. 7 How Well Has the "Savings" Component Performed from the Individual's Perspective? EARLIER CHAPTERS HAVE TAKEN STOCK OF structural reforms of retirement security systems in Latin America largely from the perspective of the policymaker, assessing the new multipillar pension systems in terms of their impact on laudable socioeconomic objectives. There is evidence that pension reforms in the region have increased the fiscal sustainability of remaining public pension promises; contributed to the development of the financial sector and to the deepening of capital markets; corrected what were regressive institutions; and even removed distortions in the labor market to improve workers' incentives to participate in formal pension systems. However, stalled progress with one of these objectives, increasing coverage, is cause for concern among the region's governments. The share of the workforce that contributes to a formal pension system remains low. Rates of worker participation by level of household income even show a regressive pattern. In several Latin American countries the share of the el- derly population receiving pension benefits is falling. For at least some in- dividuals the new funded, privately managed individual savings pillars are not as attractive as they are made out to be. Furthermore, few workers have found it worthwhile to make voluntary contributions to their individual retirement accounts even in those coun- tries where these are relatively liquid, impose no additional commissions, and offer attractive tax benefits (at least relative to savings in other financial instruments).1 This chapter, which is based on Yermo (2002b), explores why this is the case, evaluating how well individuals have fared under the savings component with respect to their consumption-smoothing objectives. We find that the earnings ceilings used to calculate mandatory contri- butions are relatively high, which leaves little space for voluntary con- tributions. Mandatory contributions may also place a heavy burden on 125 126 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA young and poor workers who have pressing consumption needs. We also evaluate some aspects of the performance of the new funded pillar that may make them less attractive than those of other formal and informal in- struments available to secure adequate retirement income. These factors may also help explain why many workers in Latin America still choose to ignore formal pension systems despite reforms. Risks to Old-Age Income Security Structural reforms of Latin America's purely public pension systems were intended to have three main benefits for individuals. First, the introduc- tion of a fully funded system based on a defined contribution formula would grant individuals full legal ownership over a financial asset--the accumulated fund in their individual account--that could not be appro- priated by third parties, even in the case of personal bankruptcy.2 Second, the partial replacement of state administration of a PAYG system with private management of pension funds would insulate workers' pension savings from political manipulation and ensure higher efficiency and professionalism. Third, the shift from a defined benefit formula to a defined contribution plan would encourage each age cohort to procure enough resources for their own retirement. Overall these improvements should have permitted more efficient consumption-smoothing.3 One can evaluate whether reforms delivered these benefits by referring to the four main risks that individuals are exposed to in retirement income security systems (see box 7.1), as well as the administration costs involved in establishing and operating different pension arrangements. The first two improvements should have translated into lower policy and agency risks for individuals. The last improvement should have led to better man- agement of investment and longevity risk. Investment risk tends to vary significantly across countries, depending on the extent of development of the domestic financial markets (cost, volatil- ity), the legal protection of property rights and contracts, and access to for- eign assets. The management of investment risk in a funded system depends primarily on the importance (relative weight) of the system in the provision of total retirement income. All individuals have a minimum level of income that they wish to attain in old age. Reaching this target with a high degree of certainty requires significant investment in risk-free, long-term assets. From the government's perspective providing such assets is indispensable to permit efficient consumption smoothing by households. In countries where macroeconomic management and fiscal prudence are the norms, such an as- set would normally be an inflation-indexed long-term government bond. PAYG defined benefit systems can also offer implicit rates of return that are inflation protected, but they are relatively inflexible with respect to changes in life expectancy. On the other hand, recent reforms of public pension HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 127 Box 7.1 The Four Fundamental Risks in Retirement Income Security Systems Investment Risk Investment risk arises from the variation in account balances and portfo- lio values as a result of inflation and changes in the prices of assets held by a pension plan. In defined contribution plans this risk is borne by the individual. In defined benefit plans the risk is borne by the plan sponsor (the government or employer). Normally one can trade off investment risk against the expected reward from the investment. The more risk averse is an individual the less risky will be his or her optimal investment portfolio. Longevity Risk Longevity risk refers to the uncertainty surrounding the length of retire- ment--or the time between retirement and death of the retiree or survivor. This risk is in principle borne by the plan sponsor in defined benefit plans, and shared between the retiree and the annuity provider in defined contribution plans. Policy Risk Policy risk arises from interference by policymakers in the operation of a pension system. Intervention can range from arbitrary changes in plan rules (e.g., benefits, tax treatment) to more direct intrusion in the opera- tion of the pension fund (e.g., strict investment rules that do not permit adequate diversification of investment risk. Agency Risk Agency risk refers to risks arising from private management of pension plans. The most serious forms of agency risk include the misappropria- tion of assets and outright fraud. More commonly, agency risk surfaces in circumstances where there is a conflict of interest, such as when a pen- sion fund manager engages in an investment transaction with a related party. A weaker but no less harmful form of agency risk is negligence or ignorance on the part of the pension provider. Sources: Yermo (2002b); Barr (2000). systems introduced in Italy, Latvia, Poland, and Sweden, which are based on notional defined contribution (NDC) formulas, do not suffer from this problem. NDCs are comparable to a funded system in which the assets re- semble a portfolio of inflation-indexed government bonds, but there are some important differences (described in box 7.2). In particular, it may be 128 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 7.2 Comparing Notional Defined Contribution Systems and Pension Funds Invested in Inflation- and Wage-Indexed Government Bonds Notional defined contribution (NDC) systems are publicly managed indi- vidual account-based retirement systems that are run on a PAYG basis, like publicly managed, defined-benefit plans. Their investment and longevity risk management properties, however, are closer to those of the funded, defined contribution systems introduced in Latin America than to those of defined-benefit plans. Individual accounts are credited annually with a re- turn that is normally linked to some macroeconomic income measure such as the wage base or output, and accumulated funds are converted into income streams only near retirement. There is an automatic adjustment to increasing life expectancy. Later generations pay higher premiums than earlier generations for the same degree of protection against longevity risk. NDCs can be thought of as an investment in short-term, wage-indexed government bonds. In the absence of PAYG pension systems, such bonds, if they existed, would have valuable risk management properties, espe- cially later in life when a greater proportion of a household's wealth is invested in financial assets. NDCs also offer some protection against inflation. The actual extent of inflation protection will depend on the in- come measure used to calculate the return and the correlation between this measure and inflation. In the European countries that have introduced NDCs a measure of the wage base is used, which in normal circumstances would offer a high degree of inflation protection. An inflation-indexed bond, on the other hand, offers full protection against inflation. One disadvantage of NDCs with respect to long-term government bonds is that the return on both current and past contributions is deter- mined annually, whereas a long-term government bond offers a fixed coupon payment until the maturity date. There is therefore less interest rate uncertainty with a long-term government bond. NDCs, however, can have return-smoothing mechanisms to mitigate the inherent volatility of the indexing factor. NDCs may also be subject to more policy risk than the explicit debt is- sued by the government to meet pension liabilities. Explicit pension debt is subject to the scrutiny of foreigners as well as the local population. It may therefore be more costly for governments to default on explicit debt than on implicit, NDC debt. The main advantage of notional defined contribution systems is that they are cheaper to run. The difference in administrative cost is of a high order of magnitude because under an NDC system there is no need to as- sign fund management to financial institutions. The cost of an NDC is not much higher than that of a PAYG defined-benefit plan. The extra cost arises from the need to maintain a detailed record of individual accounts and additional disclosure to participants. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 129 argued that the political risk of the implicit government debt of an NDC is greater than that of the explicit debt of a funded system. Where inflation- indexed bonds or similar assets are not available domestically, pension funds may find them by investing abroad. In practice the optimal investment portfolio in the savings component of a social security system will vary significantly among individuals, depending on the rate at which they discount the future, their degree of risk aversion (both of which may be age dependent), the extent to which they face unexpected shocks to their wealth and income (and hence their need for liquidity), the correlation of these shocks with their investment portfolio, their desire to pass on some of their assets to their descendants, and their access to defined benefit pension plans and other statutory pool- ing instruments (Campbell et al. 1999). In general, however, inflation- indexed bonds, NDCs, and similar assets offering inflation protection are attractive investment instruments for long-term income smoothing, espe- cially as individuals approach retirement. Housing also tends to offer good protection against inflation over the life cycle of an individual (at least in urban areas), but house prices can experience much volatility over short periods. Education and children are also traditional forms of saving for retirement used around the world. The management of longevity risk in funded systems depends primar- ily on the type of plan chosen. Defined benefit plans, whether funded or PAYG, privately or publicly managed, do not provide automatic adjust- ment for increasing life expectancy and hence are less suitable for man- aging intergenerational longevity risk. Given the inflexibility of statutory retirement ages, defined contribution plans are better endowed for this purpose. In a defined contribution system the cost of increases in life expectancy of earlier generations cannot be easily shifted to later gener- ations as it can in a defined benefit formula.4 Instead, every generation must either save more or retire later to maintain its standard of living in old age. The introduction of a defined contribution system, however, does not guarantee the efficient management of intragenerational longevity risk.5 Financial instruments such as annuities that offer protection against such risk lose their attractiveness as a result of adverse selection and high com- missions. Workers also need flexibility in the purchase of annuities to ensure adequate smoothing of investment risk. In principle, policy risk will affect private plans less than publicly managed ones. Unwelcome forms of government intervention, however, can affect private plans as well. For example, governments can impose quantitative investment regulations and performance rules whose conse- quences may not always be desirable for individuals. Agency risk is specific to privately managed systems. Governments play a central role by regulating such plans to ensure that they are managed in the best interest of plan members and other beneficiaries. 130 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA In addition to their risk management properties, funded systems differ in administrative costs. These costs can be made considerably higher if a new infrastructure is created to manage individual accounts. More generally, however, commissions can remain high as a result of marketing expenses incurred as private providers fight for market share in captive markets, such as those of mandatory funded programs. Performance of the New Funded Pillars in the Accumulation Stage In a multipillar pension system with a prominent funded savings component structured on a defined contribution basis, affiliates participate in an "accumulation stage" (when contributions accumulate as savings) and a drawdown or "distribution stage" (at retirement age when workers receive their pension benefit either as a lump sum, programmed withdrawal or as an annuity, depending on pension laws). The following sections focus on the performance of the newly funded pillars in the accumulation stage, and the distribution stage. Returns from Investment High But Volatile Table 7.1 shows the annual real returns obtained by the pension fund industry from the time when private pension systems were established until December 2002. The rate of return is calculated net of any asset Table 7.1 Gross, Real Returns to Pension Funds Have Been High (December 1994­December 2002) Country Real return (percent per annum) Argentina 10.4 Bolivia 17.1 Chile 10.3 Colombia 9.9 Costa Rica 7.0 El Salvador 10.9 Mexico 10.4 Peru 6.6 Uruguay 15.0 Note: Real returns are annualized cumulative values. Returns for Chile are for Fondo 1 (in 2001) and Fondo C (in 2002). Colombian average pension fund return is measured from inception until December 2000. For Mexico returns are net of asset management fees. Source: AIOS 2002, Colombian Banking Superintendency. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 131 management fees (only permitted in Mexico), but is not adjusted for salary- or contribution-based charges (the only form of charge permitted in countries other than Bolivia, the Dominican Republic, and Mexico). These commissions do not have an impact on the accumulated fund because they are paid on top of the mandatory contribution that goes into the individual account. To calculate net investment returns one would need to calculate the equivalent fee as a percentage of assets and subtract it from the gross return. Using a 40-year contribution horizon to translate contribution-based fees into asset-based fees, the net real return in Chile between 1994 and 2002 would be 9.1 percent, compared with the gross return of 10.3 percent. The highest real return by December 2002 was obtained by the pension fund industry in Bolivia, a 17.1 percent annual average return in real terms. The lowest was Peru's 6.6 percent return rate. Overall, real gross returns of Latin American systems appear attractive. The pattern of pension fund returns has been largely determined by investment regulations that either prohibit or substantially limit investment in foreign securities, as well as by floors on investment in domestic govern- ment bonds.6 As a result of these regulations, pension fund portfolios in all Latin American countries (with the exception of Peru) are concentrated in domestic government bonds, deposits, and other instruments issued by financial institutions (e.g., mortgage-backed securities). As shown in figure 7.1 these instruments account for more than 70 percent of all pension fund assets in all countries except Peru. Because investment regulations are the main determinants of investment performance, there is a strong element of policy risk in the new savings systems. In some countries, such as Chile, where fiscal rectitude and macroeconomic stability are well established, individuals may be content if their pension fund portfolios are heavily invested in domestic government debt. Indeed, as argued above, such investment makes much sense for a large part of an individual's income, especially when PAYG pensions are no longer available. Generally, investment restrictions on domestic assets have not hindered pension fund performance relative to other instruments available in the domestic market. Interest-bearing assets have been attractive investments in an environment of high interest rates designed to rein in inflation and stabilize the currency. This was the case in Chile in the early 1980s. The his- torical performance of Chilean pension funds has been largely driven by the extraordinary rates of return during the early years. These rates were the result of high bond returns as interest rates fell in the early 1980s. In at least one case, however, the strict investment regulations have exposed workers to the vagaries of a local bond market whose recent high yields basically reflected a high default risk (and hence a high premium over the yields of government bonds of the United States and countries in the Euro zone). This risk ended up materializing in Argentina in late 2001 132 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.1 Pension Funds Invest Mainly in Government Bonds and Instruments Issued by Financial Institu- tions (December 2002) (Percent) 100 80 60 40 20 0 Chile Peru Rica Bolivia ArgentinaUruguay Mexico Salvador El Costa Government securities Financial institution securities/deposits Corporate bonds Equities Investment funds Foreign securities Other Source: AIOS (2002), Colombian Banking Superintendency. (see box 7.3). The Argentine experience raises serious concerns about the ability of governments to ensure the performance and indeed the sustain- ability of mandatory funded plans that replace PAYG systems--even only partly--when the right macroeconomic conditions are not in place. To the extent that the transition costs imposed by pension reform over- whelm government finances, mandating saving that ends up invested largely in government bonds exposes workers to policy risk similar to that of the old system. Nonetheless, it may be argued that even under impru- dent governments, workers are better protected against political intrusion by investing in explicit government bonds rather than implicit pension debt. Defaulting on explicit government debt has repercussions on the country's access to foreign capital. Capital inflows may dry up, leading to further deterioration in economic conditions. As discussed in chapter 3, defaulting on implicit government debt, such as PAYG debt, often has less HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 133 Box 7.3 Argentina's System in Crisis: Do Private Accounts Protect Workers from Policy Risk? It is often claimed that mandating individual retirement accounts admin- istered by private dedicated providers gives workers' retirement pensions a greater degree of protection against political interference than they would have under a purely public PAYG regimen. However, the degree of protection against policy risk offered by privatizing a large portion of mandated pensions can be exaggerated. The recent economic and political crisis in Argentina illustrates how any government-organized retirement security system--whether directly administered or simply mandated-- can fall prey to politicians. Argentina's private pension system was vulnerable even before the cri- sis. Since the start of the system in 1994, nearly 50 percent of the privately managed assets were invested in government bonds. This concentration of portfolios, usually explained and excused by the lack of alternative in- struments and the high-risk premiums paid on government bonds, left the pension funds dangerously exposed to the government's fiscal problems. In the midst of a long economic recession a fiscal crisis flared in early 2001. The authorities pursued a number of "urgent" policies to cope with the pressures of an overvalued currency, mounting debt, and disintegrat- ing investor confidence. The government's attempts to cope have had a direct impact on the pension system by altering the value of current benefits, contribution rates, and investment rules for the private fund managers. Although some of the system's parameters did in fact need adjusting (see Rofman 2000), the actions taken by the authorities were shortsighted, hasty, lacked adequate analysis, and circumvented critical political institutions. Most of the policies implemented would have normally required changes in legislation but were approved by decree because of "exceptional circumstances," or were enacted by regulatory authorities using procedural technicalities to bypass the legislative process. The government's expediency is an important factor in analyzing recent events because the lack of political support in Argentina's congress for many of the measures taken has deeply wounded the system's credi- bility in the public eye and greatly increased the likelihood of a backlash against the structural reforms of 1994. In a desperate effort to cut government spending, the authorities first went after pensions paid by the public branch of the system. In July 2001 monthly benefits greater than 500 pesos paid by the PAYG branch were reduced by 13 percent. This cut affected about 15 percent of beneficiar- ies. By the end of the year the government had lowered the maximum pension paid by the PAYG system from $3,100 to $2,400 a month. During the first quarter of 2002, as the government decided to abandon a failed currency board, there was no attempt to index pensions to mounting (Box continues on the following page.) 134 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 7.3 (continued) inflation. Accumulated inflation of about 40 percent in 2002 has not been compensated, thereby reducing the real value of benefits for many retirees with no other source of income. The private system, however, was not spared. In the closing months of 2001 the government executed a swap of bonds held by domestic in- vestors, mostly banks, insurance companies, and pension funds in the private second pillar. In exchange for the old bonds (which had a market value and were regularly traded on the local stock market and exchanges abroad), new instruments backed by a "guaranteed loan" were issued. These new instruments had a lower interest rate and no secondary market, which made their valuation subjective. Although the swap was "voluntary," strictly speaking, the government exerted strong political pressure to accept the swap on the private fund managers both directly and indirectly through the industry regulator. Furthermore, in the fol- lowing weeks the government enacted a deposits swap by issuing a decree ordering that all pension assets invested in certificates of deposit be applied to buy treasury bills directly from the government. In the wake of Argentina's chaotic peso devaluation, in March 2002 the government decided to convert the instruments backed with guaran- teed loans--still denominated in U.S. dollars--to pesos, at an exchange rate of 1.4 pesos per dollar. These new loans were indexed to inflation and receive an annual interest rate of between 3.0 percent and 5.5 per- cent. The first scheduled payment of interest, in April 2002, was made by the industry regulator applying the 1.4 exchange rate. This conversion re- sulted in a significant increase in the real value of assets, if compared with inflation (by 40 percent) or average wages (82 percent). Although the government has not defaulted since it began repayments on the swap, in several cases the pension funds have legally challenged the conversion and refused to receive interest payments on the new instruments. The govern- ment continues to deposit these payments in custodian accounts. The government's heavy hand in setting private pension portfolios (in addition to cuts in the rate of contribution to the private system restored in 2002, and a temporary ban on new annuities that restricted new re- tirees from the private system to scheduled withdrawals) will have a long-lasting effect on the system's credibility. The increasing concentra- tion of investments in government debt has raised the share of bonds in combined pension portfolios to 78 percent. This concentration would be dangerous in normal times but, considering that the Argentine govern- ment has defaulted on part of its debt, it becomes a major concern. The authorities have declared their intention to honor the guaranteed bonds and other papers, and although the government has met its obligations to the funded system to date, the instability of the general economic situation raises the risk of further default. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 135 Box 7.3 (continued) The most likely outcome of the events of 2001 and 2002 is reduced confidence in any form of mandated retirement security provision. Ar- gentina's mixed pension system suffers from weak public confidence in institutions in general and, after the crisis, financial institutions in partic- ular. The pension system did not attract workers' interest even prior to the crisis. In the years leading up to the crisis, among the minority of the labor force that participated in the system more than 75 percent of new participants failed to make an explicit choice to join either the funded or the PAYG branch, and had to be assigned through a default process. However, recent events and public debate about the funds losing most of their assets in the devaluation are likely to lower confidence further. This is likely to keep the number of participants in the system low, and the general lack of support could increase political support for new reforms that would deeply damage the efficiency of the system. Source: Rofman (2002). dramatic repercussions on investors' perceptions of the fiscal stance of a government. Diversification into domestic equities, however, has not significantly helped improve the returns of pension funds, and when it has done so, as in Chile, it has made the pension funds a relatively risky investment. Adjusting for volatility--proxied by the standard deviation of returns-- and comparing with alternative instruments provides an idea of the real value of pension funds to individuals. When adjusted for risk the performance of the Peruvian pension funds--those most exposed to domestic equities-- lags far behind that of other countries. As shown in figure 7.2 the pension fund industry return was the lowest of the group (0.5 percent per month on average), but the standard deviation was one of the highest in Latin America, at 1.2. Hence the ratio of return to standard deviation (i.e., the return per unit of risk) was by far the lowest in the region. Given the earlier discussion of sovereign risk, this measure of investment risk must be considered along with some important caveats (see box 7.4). Although high pension fund returns help generate higher pensions (as long as contribution periods are also long), their volatility will lead to sig- nificant differences in pension benefits across cohorts. An example, based on Chile's historical returns, will help show the impact of return volatility on the accumulated fund. Figure 7.3 shows the pension fund average cumulative annual return for 20 different cohorts. Each cohort represented in the figure, except 1981, starts contributing at the beginning of the year. The 1981 cohort starts contributing in July, the month the system was launched. As of 136 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.2 The Standard Trade-Off between Risk and Return Has Not Materialized in Latin America (Returns from Inception to December 2000) 1.4 1.2 Argentina 1.0 month) Uruguay Chile per 0.8 (% Mexico 0.6 return Peru 0.4 Average 0.2 0.0 0 0.5 1 1.5 2 2.5 Standard deviation (% per month) Notes: Colombia does not report monthly returns. Returns for Chile are for Fondo 1. Source: Pension fund supervisors. December 2000, cohorts that started contributing in the 1980s earned cumulative returns that range from a minimum of 8.7 percent (1987 cohort) to a maximum of 10.9 percent (1981 cohort), with an average of 9.4 percent. Cohorts of the 1990s, on the other hand, have earned cumu- lative returns that range from a minimum of 4.1 percent (1995 cohort) to a maximum of 10.2 percent (1999 cohort), with an average of 7.0 per- cent. Such differences in returns will generate a gap between the average accumulated balance of 1980s and 1990s cohorts on the order of 35.0 percent. Some participants will find the volatility of pension fund returns exces- sive, especially at a time when alternative--in principle more liquid and safer--domestic investments such as bank deposits were also yielding high returns. As shown in figure 7.4, Chilean pension fund real returns before fees have certainly been higher on average than real yields on bank HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 137 Box 7.4 Points to Keep in Mind When Comparing Risks in Pension Fund Portfolios in Latin America In a pension plan based on defined contributions in individual retirement accounts, the accumulated savings of affiliated workers earn variable re- turns from investment in domestic and international capital markets. The skill of AFPs at maximizing returns and minimizing investment risk will have a direct impact on the retirement benefit affiliates can expect to earn from their savings. A first-glance comparison of the risk of AFP portfolios shows that Peru's AFP investments are relatively more risky than those of other mandatory private pension systems in the region (when the measure of risk used is the average statistical volatility or standard deviation of the combined investment instruments held by the AFP industry). However, some important caveats have to be made with respect to this measure of risk. For Peru the high standard deviation in the price of AFP investment securities can be explained in large part by the relatively greater alloca- tion of pension funds in private equities compared with that of other countries, whereas funds invested in government securities constitute a relatively small share of Peru's AFP portfolios. This is true because until only very recently the Peruvian government had a policy of not issuing domestic debt securities to maintain fiscal discipline--most funding was obtained in the external dollar market. Hence the level of treasury secu- rities in Peru's domestic capital market is extremely small compared with most other countries in the region, although holdings of private equities (which are typically more volatile) are higher. Government (treasury) securities in most countries in the region, with Chile being one of the few exceptions, generally have short to medium maturities (typically between 30 days and three years). These relatively short maturities mean that, as fixed income instruments, their duration and price volatility are low. In other countries where AFPs hold extremely large shares of government securities in their investment portfolios, such portfolios will reflect less volatility and presumably less risk than portfolios such as those held by Peruvian AFPs, which rely pri- marily on private sector securities, including significant shares of cor- porate stocks. Finance textbooks assume that government securities are risk free. However, the experience of Argentina as well as the precarious economic condition of many countries in the region demonstrate that this is hardly the case. Government securities, particularly in countries with chronic fis- cal deficits, are far from risk free. Their price may not be very volatile, but if governments default on their debts, as happened in Argentina, the price volatility essentially becomes 100 percent because at the default event the market price for the defaulting government's bonds falls to zero. (Box continues on the following page.) 138 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 7.4 (continued) Statistical series on portfolio volatility seldom show these drastic events because governments default in crisis situations. Thus, the holding of large shares of government securities in AFP port- folios in Bolivia, Colombia, El Salvador, Mexico, or Uruguay may not re- flect a low-risk strategy at all. The absence of default-induced price volatility data for these securities in historical statistical series of AFP portfolio performance does not mean that the potential for default is not present, as recent events in Argentina poignantly demonstrate. In fact, re- cent expert views on this matter propose that government securities should be assigned risk ratings according to the fiscal and macroeco- nomic situation of each country. If this were done portfolio risk would be measured not only by historically observed volatility trends but also by the risk categorizations or default probabilities of each security in the AFP's (and other institutions') portfolios. When the proper methodology for realistically rating government securities is fully developed, it may very well be that the investment portfolios of Peruvian AFPs are much less risky because they have low holdings of government debt. Source: Lasaga and Pollner (2003). deposits. But during the 1990s when domestic equities investment accounted for about one-quarter of the total portfolio, the difference decreased significantly (6.6 percent real yield on deposits, 9.8 percent real return on pension funds), and may not have compensated for their much greater volatility (1.1 against 8.5 percent). The rather dismal performance of Peruvian pension funds compared with bank deposits should caution other countries against a rushed liberalization of pension fund investment in domestic equities. The Latin American equity markets went through an unstable period during the 1990s and underperformed bonds in practically every country in the region. As shown in figure 7.5, in Peru, the country with the highest exposure to domestic eq- uities, U.S. dollar­denominated Peruvian government bonds (Brady bonds) offered higher and more stable yields between 1993 and 2000. Pension funds, however, were only allowed to invest in these instruments after 1999. For Latin American countries the only effective way to improve diversification, lower aggregate portfolio risk, and possibly even enhance returns is by investing in foreign securities. Currently, however, only Argentina, Bolivia, Chile, Mexico,7 and Peru permit investment in foreign assets. Chile had the highest limit of 13 percent at the end of 2000. Early in 2002 the legislated limit was raised to 30 percent, but the actual limit was set at 20 percent. In Argentina there was a low investment in foreign securities relative to the limit until 2001, which is largely explained by the HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 139 Figure 7.3 In Chile Intercohort Differences in Returns Have Been Large Average AFP cumulative return (percent) 1982 1984 return, 1986 1988 cumulative 1990 AFP 1992 1981­2000 1994 average 1996 Chile: 1998 2000 0 2 4 6 8 10 12 Percent Source: Superintendent of AFPs. government's efforts at forcing pension funds to buy government bonds. Since the devaluation foreign investment in the pension funds' portfolio has more than doubled and is now close to the ceiling. In Peru foreign investment was not permitted until 2000 and the Central Bank sets an actual ceiling on the share of permitted foreign investment to prevent volatility in the foreign exchange market. This explains why investment abroad remains somewhat below the limit. In Bolivia the limit on invest- ment in foreign securities was set by the legislation at the time of the reform at 50 percent of the pension fund's portfolio, the highest in Latin America. The Bolivian supervisor, however, only established the regulatory framework for foreign investment at the beginning of 2002. The pension funds, therefore, are only starting to increase the diversification of their portfolios overseas. With the exception of Bolivia, the demand for greater investment in foreign securities is demonstrated by the fact that AFP investments alloca- tions are close to their permissible ceiling (see table 7.2). At the same time the limited gains from diversification into domestic equities are 140 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.4 Returns on Deposits in Chile Have Been Lower but More Stable Than Pension Fund Returns 3,500 3,000 2,500 2,000 return 1,500 Percent1,000 500 000 ­500 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Deposit rate 90­365 days (UF adjusted) Pension fund return (UF adjusted) Source: Central Bank of Chile; Superintendent of AFPs. demonstrated by the large gap between the actual and permitted invest- ment in this class of securities. The gains from diversification into foreign equities are understand- able given the limited liquidity and high volatility of stock markets in the region. Using time-series data extending back to 1976, Srinivas and Yermo (2000) found that equity investors in Argentina, Chile, Mexico, and Peru would have achieved much higher risk-adjusted returns by in- vesting a large portion of their assets in foreign benchmarks such as the Standard & Poor's index (for the United States) and the Morgan Stanley Capital International Europe, Australia, Far East index (for non-U.S. eq- uity investments). In some cases (e.g., Peru since 1990, Argentina 1976­90) domestic investors would have done best by investing their whole equity portfolios in foreign equity. In all other cases investors would have benefited by investing at least half their equity portfolios in foreign equity. International diversification of bond portfolios is also valuable for investors in countries where government debt has a high default risk. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 141 Figure 7.5 Brady Bonds Would Have Been a Better Investment Than Domestic Equities for Peruvian Pension Funds December 1993 = 100 650 550 450 350 250 150 50 Dec-93Jun-94Dec-94Jun-95Dec-95Jun-96Dec-96Jun-97Dec-97Jun-98Dec-98Jun-99Dec-99Jun-00Dec-00 EMBI + US$ in local currency IFCG Notes: The Brady bond index is the Peruvian index from the JP Morgan Emerging Market Bond Database. The domestic equity index is the International Finance Corporation's Global (IFCG) Index. Source: JP Morgan; International Finance Corporation. Dollar-linked bonds issued by local governments may offer protection from devaluation and high returns to pension funds in the short term, but these returns are often the consequence of premiums demanded by in- vestors to hold bonds of a high default risk. On the other hand, investment in foreign government bonds is less valuable for pension funds in countries (e.g., Chile) that have a liquid market of indexed fixed income instruments and where government debt is rated highly.8 The experience of Argentina shows how even relatively well-diversified domestic portfolios cannot guarantee a decent performance when macro- economic conditions--economic growth and government finances in the case of Argentina--are in a dire state. In such countries the return on domestic equities is closely linked to that of fixed income securities. Only by investing abroad can pension funds reduce risk by diversifying away from domestic currencies, investing in more stable equities markets, and offloading domestic government debt with a poor credit rating. The Table 7.2 Asset Allocation and Portfolio Limits (December 2002) Domestic equities Foreign securities Actual Actual investment Limit Difference investment Limit Difference Argentina 6.5 35.0 28.5 8.9 10.0 1.1 Bolivia 0.0 0.0 0.0 1.3 50.0 48.7 Chile 9.9 37.0 27.1 16.2 20.0 3.8 Colombia 2.3 30.0 27.7 0.0 0.0 0.0 Costa Rica 0.0 0.0 0.0 0.0 0.0 0.0 El Salvador 0.5 5.0 4.5 0.0 0.0 0.0 Mexico 0.0 0.0 0.0 0.0 0.0 0.0 Peru 31.2 35.0 3.8 7.2 10.0 2.8 Uruguay 0.0 0.0 0.0 0.0 0.0 0.0 Note: Data for Colombia are for December 2000. Source: AIOS 2002, Colombian Banking Superintendency. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 143 Argentine experience, however, clearly shows also that mandatory savings systems can easily fall prey to desperate governments irrespective of the investment regimen in place. In Argentina the AFJPs were allowed to invest abroad (up to 10 percent of their portfolio), but the government still found ways to elevate their exposure to increasingly risky government bonds. The Cost to Affiliates of the New Savings Component In addition to the relatively high volatility of investment returns and the exposure of the new funded pillars to substantial policy risk, affiliates of Latin America's new funded pillars with individual accounts face steep commissions that reduce their accumulated balances drastically relative to what they could have been had these commissions been invested in the re- tirement accounts.9 As discussed by Valdés-Prieto (1998) the relatively high commissions can be traced partly to strategic competition between the pension fund administrators that resulted in high costs. Pension fund administrators have relied heavily on sales forces to attract new affiliates, which has pushed up the cost of their services. In Chile, for example, marketing and sales costs accounted for more than one-half of operational expenses in the late 1990s (SAFP 1998). The resulting high cost­high price equilibrium also depends critically on a low price elasticity of demand. Mastrángelo (1999) showed that the marketing elasticity of demand is 18.5 times greater than the price elasticity. Similar results have been found in Argentina. Furthermore, workers do not appear to react to differences in returns, although this could be the result of the limited variability in returns across AFPs. Pension fund administrators, therefore, face a strong incentive to spend on marketing and sales forces to attract new affiliates. In addition to marketing costs, the fees can be explained by the set-up costs of the new industry that in some cases have fallen largely on earlier cohorts. There are two main measures of charges that may be used: the reduc- tion in yield and the charge ratio. The reduction in yield shows the effect of charges on the rate of return, given a set of assumptions about the rate of return, the time profile of contributions, and the term of the plan. The charge ratio, on the other hand, is calculated by dividing the accumulated balance that the charges by themselves would have generated had they been invested by the sum of the accumulated balance resulting from the ac- tual contributions and the charges (had they been invested). One can also calculate the cumulative charge ratio at retirement, which results from in- cluding in the calculation all contributions and fees paid over the career of a worker. As discussed by Whitehouse (2001), the use of the charge ratio can provide a misleading picture of the cost of a private pension system when commissions are set as a percentage of the accumulated fund. On the other hand, the charge ratio is a more useful and appropriate measure of 144 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA the cost efficiency of the system in Latin American countries where commissions are charged mainly only on contributions or salaries. The commission structure is regulated in all Latin American countries. Commissions to cover administrative costs (account and asset management expenses) can be set as a percentage of salary or contribution in all Latin American countries. Additional, fixed commissions are permitted in Chile, in Mexico, and (for one pension fund) in Uruguay.10 In Bolivia and Mexico pension fund managers can also set commissions as a percentage of returns, and in Costa Rica the fund managers can set fees on the performance. The commissions must be the same for all affiliates of a given pension fund ad- ministrator. The only exception to this rule is loyalty discounts (reductions in fees for remaining with the same administrator), which are permitted in Argentina, El Salvador, Mexico, Peru, and Uruguay. Most countries also do not permit fees to be charged on voluntary contributions. Some coun- tries directly limit commission levels through explicit caps. In Bolivia the caps on both the contribution- and the asset-based charges were set during the tender process through which the fund managers were selected. Colombia and El Salvador set a ceiling on the sum of commissions and insurance premiums. In Costa Rica, there is a cap of 8 percent on fees set on the per- formance of the pension fund. A ceiling has also been set on contribution- based charges, which were first allowed in 2003. Variable commissions (those calculated as a percentage of contribution/ salary) are nonetheless the most important component of the total cost in most Latin American countries (Bolivia, Costa Rica, and Mexico are the ex- ceptions). These commissions vary significantly across countries, as shown in table 7.3. For a worker of average income the lowest variable commission was Bolivia's, at 0.5 percent of the affiliate's salary in December 2002. The highest charge was Peru's, at 2.27 percent. This comparison, however, ignores the additional asset-based fee in Bolivia of approximately 0.2 per- cent, which is equivalent to approximately 0.5 percent of the average salary (over a 40-year contribution period). The total fee-to-salary ratio in Bolivia is therefore closer to 1.0 percent, which brings it closer to other Latin Amer- ican countries but still far below the Peruvian level. The ratio of variable commissions to total contributions (excluding insurance premiums) in December 2002 is shown in the last column of table 7.3. For countries in which commissions are set only as a percentage of the worker's contributions or salary, this measure is equivalent to the charge ratio over a one-year period.11 In Argentina the charge ratio in 2002 was 36.19 percent, by far the highest of any Latin American country. The large gap in charge ratios between Argentina and the other countries is the result of the decision made in December 2001 to halve the manda- tory contribution to the funded pillar. Given the significant fixed costs in account and asset management, lower contribution rates are translated into higher charge ratios. For the same reason, the number of contributors in the country will affect the charge ratio. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 145 Table 7.3 Workers Still Pay High Commissions in Some Countries (December 2002) (Percent) Administration Contribution to fee/salary fund/salary Fee/total contribution (item a) (item b) (item c a/(a b)) Argentina 1.56 2.75 36.19 Bolivia 0.50 10.00 4.76 Chile 1.76 10.00 14.97 Colombia 1.63 10.00 14.02 El Salvador 1.58 11.02 12.54 Peru 2.27 8.00 22.10 Uruguay 1.92 12.27 13.53 Average 1.60 9.15 16.87 Notes: Administration fee includes only account and asset management charges set as a percentage of contribution/salary. Insurance premiums are excluded. Information for Colombia refers only to the mandatory pension fund system for December 2000. Information for Bolivia includes only the contribution charge (the asset management charge varies from zero to 0.23 percent, depending on the amount of assets in the portfolio). Information for Uruguay excludes an additional commission for custody, which averaged 0.00293 of total assets under management in December 2002. Source: AIOS 2002, Colombian Banking Superintendency. However, contribution rates and the number of contributors cannot ex- plain all the differences in charge ratios between countries. Chile, for ex- ample, has the same contribution rate as Bolivia and a much higher num- ber of contributors (3.4 million vs. 0.4 million in December 2002). But its charge ratio of approximately 18 percent (including fixed commissions) in 2002 is much higher than Bolivia's, which is approximately 9.1 percent (in- cluding both the contribution- and the asset-based fee). The difference in the annual commission charge per person between the two countries is even greater: US$90 in Chile in December 2002 vs. US$31 in Bolivia. The greater gap in this measure of commissions between the two countries is ex- plained by the different charge structures. The Bolivian asset-based fee col- lects increasing commissions over time as the funds accumulate. Chile's charge ratio (the second highest among the countries listed in table 7.3 when fixed commissions are included in the measure of the charge ratio) also seems inconsistent with the maturity of the system, which should have helped make the industry more efficient and competi- tive. The longer period since the inception of the new system also should have given the Chilean pension fund administrators more time to amortize the set-up (fixed) costs from their own establishment and that of the pen- sion funds that they manage. 146 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA For the same reasons, commission rates should be expected to fall over time. Figure 7.6 shows the evolution of the ratio of variable commissions to total contribution at the end of various years in three countries. Argentina and Peru have both experienced a sudden jump in variable commission rates, in 2001 and 1995, respectively. These increases in commissions resulted from reductions in the mandatory contribution rate (from 11 percent to 5 percent in Argentina and from 10 percent to 8 percent in Peru). Allowing for this policy shift, the variable commission rate fell somewhat between 1997 and 2002 in both Argentina and Chile. In Peru, on the other hand, the variable commission rate has remained at more than 20 percent since 1997. The different evolution of charges in these countries can be largely explained by regulatory policies aimed at containing operational costs (such as restrictions on switching between funds and greater control of sales forces) and direct pressure from the supervisor aimed at lowering commissions in Argentina and Chile. The imposition of restrictions on the frequency with which affiliates can switch among fund managers also applies in Peru, but this policy by itself can also foment collusion in the industry--and the evidence from Peru presents some worrisome trends. Since 1997 there has been a dramatic decline in the operational costs of the pension fund industry, accompanied by increasing return on equity to the fund managers, but little change in the fees charged to affiliates (Lasaga and Pollner 2003). Nevertheless, even the Peruvian commission rates do not appear high when compared with mutual fund and professional asset managers in both Latin American countries and abroad. A 22 percent charge ratio as in Peru is equivalent over a 40-year horizon to an asset-based fee of approximately 1 percent. While fixed income and indexed mutual funds are able to achieve somewhat lower fees in large, mature markets such as that of the United States, this is not the case for such funds in Latin America (see chapter 10 for further evidence on the cost of retail fund management). The higher commission rates that older workers faced when they entered the pension fund system also translate into higher charge ratios for these earlier cohorts. Figure 7.7 shows the cumulative charge ratio at re- tirement (including all fixed and variable commissions charged over the whole contribution period) for Chilean workers earning the average con- tributor's wage in successive cohorts, where each cohort is identified by the year in which it would normally retire, starting with the cohort that re- tired in 1982.12,13 The cumulative charge ratio at retirement was highest for the cohorts who retired soon after the inception of the new system and fell gradually for later cohorts. Because the first workers to retire from the new system started to do so only in the second half of the 1980s, few workers suffered from higher fees.14 For a worker earning the average contributor's wage who retired in December 2000 and who had con- tributed each year to the system (choosing the average pension fund), man- HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 147 Figure 7.6 Variable Commission Rates Have Risen as a Result of Declines in Contribution Rates in Argentina and Peru, but Not Chile Commission/contribution, 1990­2002 45 40 35 30 25 20 15 10 5 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Argentina Chile Peru Note: Commission/contribution, 1990­2002. Source: Superintendencies of respective countries; authors' calculations. agement fees would have consumed approximately one-quarter of his or her total contributions.15 It should also be considered that total mandatory contributions are higher than the amount that actually goes into the individual account to finance retirement benefits. In addition to administration fees, the total commission that workers pay the pension fund administrators includes a premium for disability and survivors' insurance. In Chile, for example, the total mandatory contribution rate for the average contributor between 1982 and 2000 was 13.3 percent of her salary. The contribution rate to finance retirement benefits amounted to 9.6 percent (until 1987 the pen- sion fund administrators could charge fixed and variable commissions on the stock of accumulated assets, which reduce the actual contribution rate to pay retirement benefits below the statutory 10 percent), while the total commission rate (administration fees and insurance premiums) over this period was 3.7 percent. The total contribution rate for a worker earning the minimum wage was somewhat higher (13.6 percent) and was allocated in a less advantageous manner, as the contribution rate for retirement 148 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.7 One-Quarter of the Net (of Insurance Fees) Man- datory Contributions of the Average Chilean Con- tributor Who Retired in 2000 Went to Administra- tion Fees 29 27 (%) 25 23 ratio 21 19 Charge 17 15 1982198319841985198619871988198919901991199219931994199519961997199819992000 Notes: Cumulative charge ratio by year of retirement in Chile. Salary 200,000 pesos in December 1981. Source: Superintendent of AFPs; authors' calculations. benefits was 9.2 percent while the total commission rate was 4.4 percent. It is the total mandatory contribution rate and how it is divided between total commissions and the (net of total commissions) accumulated balance in the individual account that should be considered when discussing in- centives to participate in the second pillar. Since the establishment of the system, the total mandatory contribution rate in Chile has fallen by approximately one percentage point while the ratio of total commissions to total contributions has declined substantially. As shown in Figure 7.8, the ratio of total cumulative commissions (admini- stration fees and insurance premiums) to total cumulative contributions between 1982 and 2000 was nearly 33 percent for a worker earning the min- imum wage, 28 percent for a worker earning the average contributor's wage, and 26 percent for a worker who earned three times the wage of the average contributor. The higher mandatory contribution rates and higher ratios of commissions to contributions for lower income workers can also create an important disincentive to participation in the second pillar. It should also be noted that in a country with a high level of wage inequality such as Chile, the median wage is significantly lower than the mean wage. Workers in the informal sector therefore faced potential mandatory contribution and HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 149 commission rates that were more similar to those of workers earning the minimum wage than those of the average contributor.16 Of course a regressive charge structure would be less worrisome if those AFPs chosen by poorer households offered a better service or performed better in terms of gross rates of return. There is no evidence that this is the case. Indeed, there is no correlation between the level of commissions and the performance of a pension fund. As for the service offered, more fre- quent or detailed communications on a worker's accumulated balance are unlikely to compensate him or her for a lower replacement rate or net salary. Moreover, it would appear that many low-income workers may not be choosing the lowest-priced option, given their earnings level, as demonstrated by the low price elasticity of demand calculated by Mastrángelo (1999) for Chilean AFPs. For Argentina, where AFJPs could also set fixed charges before November 2001, Rofman (2000) has calcu- lated that the average commission (including the insurance premium) would be less than 3.0 percent instead of 3.4 percent if each contributor chose the cheapest AFJP for his or her income level. Figure 7.8 Participation in the Second Pillar Is Costlier for Poorer Workers in Chile (%) 40 38 36 contributions 34 32 total to 30 28 26 24 commissions 22 of 20 Ratio 1982198319841985198619871988198919901991199219931994199519961997199819992000 Minimum wage Average wage contributors 3*average wage contributors Note: Cumulative charge ratios for different salaries in Chile, 1990­2000. Source: Superintendent of AFPs. 150 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA These distributional consequences of the pension reform deserve more attention by policymakers. The costs of the system, and in partic- ular fixed commissions, can also be a strong disincentive to participation for low-income households who may mainly desire a minimum pension guarantee that insures them against poverty in old age. In Mexico, where pension fund administrators have complete freedom to set charges, the cuota social account subsidy more than offsets these adverse distribu- tional effects and may also increase the incentive to participate in the formal pension system. The earlier cohorts who chose the funded system have also borne a dis- proportionate share of the set-up costs of the new pension fund industry; the evolution of the commission structure has led to a redistribution of income from early (older) to later (younger) participants. Unwittingly, pension reform of the Chilean kind has at least on this aspect reversed the bias inherent in generous PAYG systems in which future generations paid for current retirees. In the new funded system current workers subsidize the cost of the system for future generations.17 Such redistributions are more generalized when older workers are obliged to switch to the new sys- tem as in Bolivia and Mexico. Although expected, such large intergenerational transfers are avoid- able.18 It is also possible to lay the cost of these redistributions on the richer households of a given cohort. Uruguay's reform, which required contribu- tions to the new savings pillar only for higher-income individuals, has en- sured that the new industry is subsidized in its early stages by those most able to create "thickness" in the market and endure high commissions. Performance of the New Funded Pillar in the Distribution Stage Workers affiliated with the new funded pillars are generally allowed a greater degree of choice at retirement in the distribution phase than dur- ing the accumulation phase. As shown in table 7.4, in all Latin American countries, with the exception of Bolivia and Uruguay, participants may choose between (at least) a private annuity and scheduled withdrawals. Some countries also permit a sequential, but not simultaneous combina- tion of the two (deferred annuities). Lump-sum distributions, mean- while, are highly restricted, which is consistent with the mandate to save.19 Depending on which option is chosen, the effect on risk-bearing is rad- ically different. The scheduled withdrawal option lays all investment and longevity risk on the individual. An annuity, on the other hand, insures the policyholder against the risk of outliving his or her resources and therefore lays longevity risk on the insurance company that sells the pol- icy. Insurers, however, price annuities according to the life expectancy of HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 151 Table 7.4 The Form of Benefits Is Usually Inconsistent with an Assumption of Retiree Irrationality with Respect to Saving Country Form of benefit Argentina Annuity, scheduled withdrawal (up to five years after retirement) Bolivia Only annuity Chile Annuity, scheduled withdrawal, deferred annuity Colombia Annuity, scheduled withdrawal, deferred annuity Costa Rica Annuity, scheduled withdrawal, deferred annuity. Not implemented because products are not available. During the first 10 years workers can withdraw full accumulated balance as a lump sum at retirement. Dominican Republic Annuity, scheduled withdrawal El Salvador Annuity, scheduled withdrawal, deferred annuity Mexico Annuity, scheduled withdrawal Peru Annuity, scheduled withdrawal, deferred annuity Uruguay Only annuity Notes: Lump-sum distributions are permitted in all countries except Bolivia and Uruguay, but there are substantial constraints. Relevant data can be consulted in Devesa, Martínez, and Vidal (2000). Sources: Devesa, Martinez, and Vidal (2000); and relevant national legislation. the age cohort to which an individual belongs. Hence, annuities normally offer protection only against the risk of outliving the average individual of the cohort, while passing on the cost of anticipated increases in the average life expectancy of the cohort.20 The extent of protection offered against investment risk, meanwhile, varies depending on the type of annuity sold. In general, the scheduled withdrawal option is preferable when interest rates are very low and unlikely to increase within the time frame permit- ted for purchasing a deferred annuity. Pensioners, however, are not free to choose the amount they wish to withdraw as a benefit. Scheduled with- drawals are recalculated every year by the pension fund administrator as a function of the pension fund's return and the life expectancy of the worker and his or her family members. The scheduled withdrawal option is mandatory for workers in Chile and El Salvador who have accumulated funds that are insufficient to generate annuities above the minimum pen- sion. At retirement they must draw down a pension equal to the minimum pension. When the funds run out, the government pays the remaining amount necessary to finance the minimum pension. Annuities are attractive instruments for workers with above-average life expectancy because the price they pay reflects the average life 152 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA expectancy of the population. Mandatory annuitization of the balance of accumulated savings in individual retirement accounts may, therefore, be beneficial for these individuals--and it may correct failures in annuities markets as a result of adverse selection.21 However, it can be costly for some disadvantaged groups in society with a low life expectancy because the premiums charged by annuity providers (based on the average life ex- pectancy of the whole population) are above what would be actuarially fair for these groups. Annuities are mandatory in Bolivia and Uruguay for all workers at the time of retirement, and in Argentina, Chile, El Salvador, and Peru for workers who wish to obtain their pensions before the official retirement age.22 Given the attraction of early retirement for many workers, the early re- tirement rule acts as an indirect form of compulsion of annuities. In Chile, for example, more than one-half of annuities are of the early retirement type. Other than adverse selection, the only main justifications of manda- tory annuitization are the moral hazard of minimum pension guarantees and workers' improvidence. The moral hazard problem justifies annuiti- zation of the portion of the accumulated balance sufficient to meet the minimum pension guarantee. Such argument is certainly not applicable to countries such as Argentina or Uruguay that have contributory, PAYG- financed first pillars that fulfill the poverty-prevention function. In the other countries it would justify lower levels of compulsory annuitization for early retirement. The improvidence argument, although valid, needs to be weighed against the costs of forcing low-income individuals to buy an- nuities. Low-income workers often have lower life expectancies and greater consumption needs in retirement than richer individuals and may therefore suffer most from mandatory annuities. There are also certain restrictions on annuity sales that can be costly for individuals. Workers cannot combine them simultaneously with the scheduled withdrawals option. Since, in addition, all annuities are of the single-premium, fixed type, workers face considerable risk in the timing of the annuity purchase. Permitting partial and gradual annuitization would go a long way toward smoothing out this investment risk. Variable annu- ities--whereby investment risk is borne by the pensioner and longevity risk is borne by the insurance companies--may be particularly attractive for higher-income workers or for those who work where the public pen- sion system still offers generous pensions. Variable annuities may also be provided more cheaply by insurance companies than fixed annuities be- cause, with the exception of Chile, the financial instruments needed to underwrite fixed annuities and hedge investment risk over long periods (inflation-indexed bonds of long maturities) are in short supply. In Chile, Colombia, and Peru annuity benefits must be indexed to a measure of prices, also transferring inflation risk to the insurance com- pany. In Uruguay annuities must be indexed to wages. In Argentina, HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 153 benefits may be denominated in U.S. dollars, which offers protection against devaluation (and hence at least partial protection against infla- tion). An inflation-indexation requirement, however, is feasible only to the extent that there is a liquid market of inflation-indexed securities that insurance companies can rely on to build portfolios that match their inflation-indexed liabilities. Only Chile has such a market. In countries like Colombia and Peru insurance companies are likely to charge a hefty premium for underwriting inflation-indexed annuities. In Uruguay pre- miums are likely to be even higher because growth in wages often out- strips prices. In Chile, Colombia, El Salvador, and Peru workers can also buy deferred annuities and in the meantime draw down part of the accumu- lated balance as part of a scheduled withdrawal. Deferred annuities may be attractive when interest rates are expected to increase. Nonetheless, annuity purchases may still be badly timed because workers are not able to buy deferred annuities before retirement and only one- to three-year deferrals are permitted at retirement. In Argentina, Bolivia, and Mexico, where deferred annuity purchases are not permitted, the investment risk that workers face is even greater. However, as long as workers are able to move into conservative, fixed income portfolios as they approach the time of annuity purchase they can minimize this risk. If interest rates fall the value of the annuity that can be purchased with the accumulated balance declines, but the market value of the fixed income investments in the individual account rises. Restrictions on worker choice of investment portfolios are therefore a source of un- necessary volatility in retirement benefits. An indicator of the impact of interest rate volatility on annuity val- ues in Chile is shown in figure 7.9.23 This graph shows the value of the annuity that a premium, fixed in real terms, would buy at the end of each year since 1988. The value of the annuity is expressed in terms of the replacement rate, with the replacement rate of 1988 set arbitrarily at 50 percent. The interest rate used to calculate the benefit paid by the annuity is the annuity yield for workers retiring at the official retirement age, as reported by the Superintendencia de Valores y Seguros. As shown in figure 7.9 there is some variation in the annuity value over time. The difference between the highest and lowest replacement rates is 22 percentage points, the average replacement rate over the period is 60 percent, and the standard deviation is 6 percent. These differences in replacement rates across cohorts are caused by the volatility of interest rates over the period. In particular, the decline in interest rates in the late 1980s lowered the value of annuities for cohorts retiring in these years. Workers retiring in 1998 and 1999 especially would have been better off deferring the purchase of the annuity for two and one years, respectively. 154 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.9 Annuities Have Yielded Varying Levels of Retirement Benefits in Chile 70 65 (%) 60 rate 55 50 45 40 Replacement 35 30 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Note: Replacement rate from annuities; 1988 50 percent. Source: Authors' calculations, based on data from Chile's Insurance and Securities Superintendency. Investment and Longevity Risk Management Properties of the Funded System The new funded pension systems are a potentially attractive new instru- ment for managing intergenerational longevity risk during the accumula- tion stage. By moving to a defined contribution formula the new pension pillars ensure a more fair and efficient allocation of the cost of increasing life expectancy between generations. However, annuities, the instruments that insure individuals against intragenerational longevity risk are only available at retirement. It may be desirable, therefore, to permit the pur- chase of annuities before retirement, although this may expose insurance companies to greater risk of misjudging increases in life expectancy. It is considerably less clear whether the design of the new funded pillars is consistent with the efficient management of investment risk. Restricting individual choice and limiting portfolios to mainly interest-bearing assets make sense for mandatory contributions intended to meet minimum income requirements in old age (most if not all of the mandatory contri- butions of poorer households and part of the contributions of richer households in countries where the PAYG plan is being phased out). HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 155 Indeed, it could be argued that mandatory contributions should be in- vested exclusively in inflation-indexed government bonds of appropriate maturity and of the highest possible creditworthiness. These bonds are es- sential to building portfolios that will ensure the attainment of basic re- tirement income needs. Such a limited investment regimen, however, would call into question the rationale for decentralized, competitive pri- vate management of the new funded pillars. To the extent that society agrees on the need to set a minimum pension guarantee (defined in refer- ence to the minimum or average wage), a flat-rate benefit offered as a part of a PAYG system would achieve the objective of providing a basic retire- ment income much more effectively and at much lower cost. This proposal is taken up in more detail in chapter 9. Individual investment choice is desirable for the portion of contributions whose main objective is to ensure income smoothing above the basic income level. Richer households in countries where the funded pillar is the only source of mandatory pensions may therefore wish to invest their portfolios in a riskier manner than that permitted by current regulations. Similarly, both low- and high-income households in countries such as Argentina, Costa Rica, or Uruguay may also wish to invest their additional mandatory contributions to the funded pillar in a riskier manner. Workers in these coun- tries are still covered by a contributory PAYG system that provides for ba- sic retirement income objectives. In other words, individual portfolio choice becomes most desirable for retirement savings that are complementary to those intended to keep households out of poverty in old age. For contributions above the level needed to meet basic income needs in old age, the current restrictive design on the new funded pillars can impose costly restrictions on risk management in two ways: 1. The current mandate to save forces individuals to manage invest- ment risk through a single instrument until their retirement. Individuals are not allowed to invest their mandatory, tax-advantaged pension con- tributions in assets that may offer a more suitable risk­return trade-off or may be cheaper to manage, such as bank deposits, property, children's ed- ucation, or foreign assets. Moreover, prior to retirement participants can- not pool smooth investment risk through, for example, deferred annuities, guaranteed investment contracts, or term deposits. 2. During the accumulation stage the mandated savings instrument of- fers a risk­return trade-off that is determined by investment regulations and the fund managers' choices. Hence, it is not possible to adjust portfo- lios over the life cycle in accordance with affiliates' risk, time, liquidity, and bequest preferences. The only country that has been moving away from this design is Chile, where a second fund was introduced in 2000. Since May 2000 and until late 2002 Chilean workers close to retirement were offered the opportunity to 156 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA trade out of the so-called Fondo 1 (the original pension fund) into a fund invested exclusively in fixed income securities (Fondo 2). At the end of 2001 only men older than 55 and women older than 50 were permitted to switch their accumulated balance to this second fund. The Fondo 2 had limited popularity among those workers eligible to switch, however. The option to switch was only taken up by a handful of (lucky) workers who have benefited since then from bonds' superior per- formance relative to equities. Figure 7.10 shows the performance of Fondo 2 over Fondo 1 in terms of a higher annual real rate of return (gross of contribution fees) during every month between June and December 2001. This bumpy start of the Fondo 2 appears to be caused by the limited Figure 7.10 Chile's Pension Funds Did Not Do a Good Job of Educating Participants: Fondo 2 Had Few Takers Despite Earning Higher Returns (Percent) 10 9 8 7 6 5 4 3 2 1 0 2001 2001 2001 2001 2001 2001 2001 June July August October September November December Fondo 1 Fondo 2 Note: Annual real gross rate of return, 2001. Source: Superintendent of AFPs. HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 157 Table 7.5 The Five New Funds in Chile Vary by Proportion of Equity Investment (Percent of total assets in each type of fund) Minimum Maximum Fund A 40 80 Fund B 25 60 Fund C 15 40 Fund D 5 20 Fund E 0 0 Source: Ministry of Economy and Finance, Chile. publicity extended to the switching option. The pension fund administra- tors had little incentive to engage in advertising and information cam- paigns because the switch to Fondo 2 would not translate into any addi- tional income from commissions. In addition, the fund managers were under heavy pressure from the government to cut commissions and had re- duced the number of sales agents they employed. Nor did the government sufficiently promote the second fund. The government drew important lessons from the experience with Fondo 2. Despite the low acceptance of the second fund, a law passed early in 2002 extended individual choice over the investment of manda- tory savings even further. A multifund system has been approved that per- mits workers to choose among five funds, all with varying exposure to equities. The five funds are characterized by the level of investment in eq- uities. The equity limits for each fund are shown in table 7.5. Men 55 and under and women 50 and under are able to choose among all five funds. Men older than 55 and women over 50 are able to choose among funds B to E. Pensioners who have maintained their accumulated assets with the AFPs (instead of opting for an annuity) are able to choose among funds C to E. Those workers who do not select a specific fund when they enter the workforce will be assigned one according to their age: workers up to 35 years of age are assigned to fund B. Men between 36 and 55 and women 36 to 50 are assigned to fund C. Men 56 and older, women 51 and older, and pensioners are assigned to fund D.24 The same default options were applied 90 days after the introduction of the multifund system for those workers who had not selected a fund within that period. Men and women who have their accounts invested in fund A but do not move to one of the other funds within 90 days after they turn 56 or 51, respectively, are shifted to fund B in a gradual manner over a four-year period. Although workers' reaction to Fondo 2 does not augur well for the new multifund system, the government has made a greater effort to publi- cize the new funds. Since September 2002, when funds A, B, and D were 158 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 7.11 Chilean Workers' Choices in the New Multifund System Largely Correspond with the Default Options 90.0 fund 80.0 70.0 each in 60.0 50.0 40.0 invested 30.0 20.0 10.0 0.0 Percentage Up to age 35 (%) Ages 36­55 (%) 56 years and older (%) Fund A 3.1 1.1 0 Fund B 82.4 5.2 0.8 Fund C 11.5 83.6 11.4 Fund D 1.4 6.6 75.3 Fund E 1.5 3.5 12.6 Age Group Note: Contributions to multifunds by age group, December 2002. Source: Superintendent of AFPs. introduced, investment in the funds has been more promising.25 By June 2003 nearly 1.4 million workers had chosen a fund, representing 40.5 per- cent of contributors. It is interesting to note that the choices made largely correspond with the default options. As shown in figure 7.11, between 75 percent and 85 percent of the workers in the three age groups chose the portfolio assigned as the default option. The government has a fiduciary responsibility to ensure that workers are properly advised on the choices they face and their potential impact on retirement income. This responsibility may be laid on the pension fund ad- ministrators, but because private fund managers enjoy a trapped market of mandated demand, they have little incentive to incur the costs of edu- cating individuals on optimizing their returns from the system by shifting their portfolio allocations. Indeed, the mandate may be partly responsible for some of the internal inconsistencies of the system. Because the number of competitively priced voluntary savings and insurance instruments in Chile is increasing, the possibility that distortions in the financial sector HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 159 and capital markets could arise from the government mandating a particular form of private retirement saving should be investigated with increased scrutiny (Shah 1997; Lasaga and Pollner 2003). The new Chilean design also intermingles the poverty prevention and income smoothing objectives of a formal pension system. In its drive for individual responsibility it has put poor households in a situation where investing in equities seems to be a one-way bet. Workers who do not ex- pect to accumulate much more than the minimum pension guarantee have an incentive to invest in riskier portfolios (with a higher allocation to eq- uities). If returns are high, they receive higher income, but if they are low, the government picks up the bill through the minimum pension guarantee. The data available in fact show that poorer workers choose portfolios sim- ilar to those of richer households (see figure 7.12). Hence, the government, as guarantor of the minimum pension, will face an additional source of risk in the form of the volatility of the equity portfolios of poorer work- ers. The increasing international diversification of equity portfolios will Figure 7.12 Choices Are Similar for Different Income Groups in Chile If Controlling for Age 60.0 fund 50.0 each to 40.0 30.0 contributors of 20.0 10.0 Percentage 0.0 0­150,000 pesos 150,000­350,000 pesos More than 340,000 pesos taxable income (%) taxable income (%) taxable income (%) Fund A 1.3 2.1 2.7 Fund B 44.1 44.8 32.0 Fund C 45.4 43.6 50.3 Fund D 7.8 6.5 9.6 Fund E 1.5 2.9 5.4 Taxable income (thousands of pesos) Note: Percentage of contributors to each fund by income group, December 2002. Source: Superintendent of AFPs. 160 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA reduce this risk, but the fact remains that the government is offering poor workers an incentive for risk taking, when such liabilities may best be matched through investment in inflation-indexed government bonds. If the purpose of the guarantee was a form of redistribution through the back door, this could have been done in a way that did not create perverse incentives for risk taking and distortions to capital markets. A similar risk of moral hazard may arise in Peru, which passed legisla- tion along Chilean lines to allow fund managers to offer affiliates multiple funds. On the other hand, this problem does not arise in those savings components that are not covered by state guarantees, such as the volun- tary savings pillar of all Latin American countries and the mandatory pillar in Argentina, Costa Rica, and Uruguay (the countries that have retained a basic PAYG pillar). A move away from the one-size-fits-all model and toward more worker choice of investment may therefore be considered in these cases. Nonetheless, it is critical that the authorities en- sure the comparability and transparency of fee structures and fund per- formance and that they engage in a serious effort to raise the financial literacy of the population. Conclusion Although the savings systems have delivered relatively high gross returns and the administrative costs do not appear high when compared with retail fund managers in the region, there are still some worrying issues. Young and poor workers may be discouraged from participating by the relatively high mandatory contribution rates. Poorer workers also suffer from fixed commissions in Chile, which may further reduce their incentive to participate and which have a regressive income effect. The earnings ceiling used to calculate mandatory contributions is also very high in all Latin American countries (they are actually among the highest in the world for mandatory retirement security systems), which leaves little space for the voluntary system to develop. There are other explanations for why rational workers may not find the new savings component satisfactory in Latin American countries: the cost of pension fund administration, the lack of investment choice, the absence of international diversification, the regulatory bias toward government se- curities, and the obligation to buy annuities. In every Latin American coun- try that has introduced savings components, at least one of these features is present. Chile is probably the country with the least of these negative fea- tures (high fees and obligation to buy annuities for early retirement). There are signs that the savings systems are becoming more efficient over time, at least in some countries. Variable commission ratios have fallen in recent years in Chile, but nowhere near as much as the adminis- trators' operating expenses. Investment regulations are gradually being HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 161 relaxed (including on foreign investment) in Bolivia and Mexico. Chile in- troduced portfolio choice in 2002 and Peru will do so in 2005. Other countries, such as Mexico, may soon follow. Policymakers are also recon- sidering the high contribution rates of these systems, although, at least in Argentina, they do not seem to have taken into account the impact that lower mandatory contributions have on variable commission rates in an industry with high fixed costs. Argentina halved the contribution rate to the mandatory system in 2001, which led to a sharp increase in the ratio of fees to contributions. As policymakers reconsider these and other features of the savings com- ponents, it is essential that they take into account the interaction of the savings component with the rest of the retirement income system. High contribution rates and maximum taxable earnings are particularly detri- mental for the growth of mandatory and voluntary pension savings when a large public pillar is still in place (as in Costa Rica and Uruguay). Simi- larly, diversification into foreign securities and riskier domestic instru- ments may be more valuable for workers if they know that they will be able to rely on defined benefit promises to cover their basic retirement needs. Finally, mandating savings into a particular instrument may be costly for workers if they can find better investment alternatives elsewhere. Notes 1. In Mexico there were US$0.16 billion in the voluntary accounts in December 2001. In Chile at the end of 2001 there were 1 million voluntary savings accounts that held US$0.2 billion. Only one of six affiliates had a CAV, despite their rela- tively high liquidity (withdrawals are permitted four times a year) and absence of commissions. 2. "Funding" does not necessarily imply private management, but the reverse is nearly always the case, especially for mandatory plans. Some publicly managed plans, like the provident funds found in several Asian countries, are funded. In some OECD countries public plans that were traditionally run on a PAYG basis are starting to build up reserve funds to meet future liabilities. For ease in terminology we use the term "funded" or "savings" component to refer to the privately man- aged individual accounts that are such a prominent feature of pension reforms in Latin America. 3. We abstract here from the poverty prevention objective because this is the subject of chapter 9. Readers should note, however, that the systems in some Latin American countries (e.g., Bolivia, Chile, El Salvador, and Mexico) play a poverty prevention role with the support of minimum income guarantees. 4. This will only happen to the extent that the annuity providers do not accu- rately estimate increases in life expectancy. The later in life that people are allowed to purchase annuities, the less likely this will be the case. It should be noted that annuities are similar to defined benefit plans in their investment and intragenera- tional (generation-specific) longevity risk management properties. 5. The rest of this chapter focuses on the first type of risk. 6. Investment floors are only in place in Bolivia, Mexico, and Uruguay. The Argentine government also used nonlegislated forms of persuasion before the 2001 162 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA crisis to increase the allocation of pension funds to government bonds above the legislated limit. 7. Mexico lifted the ban on investment in foreign securities in June 2002. 8. Chile was an exception in Latin America in the 1990s, being the only country whose government debt was rated investment grade by the main rating companies. Since then Mexico has obtained investment grade as well. 9. As explained previously, in addition to these commissions workers must pay monthly premiums for disability and life insurance. 10. In Peru fixed commissions were permitted until 1996. Argentina banned fixed commissions at the end of 2001. 11. In Mexico the two measures are not equivalent because some providers charge commissions on assets and returns. In Bolivia the commission is set on assets under management. In Chile the AFPs also charge fixed commissions. 12. The cumulative charge ratio at retirement measures the total impact of charges on retirement income over a person's career. To calculate this ratio for Chile, both fixed and variable charges need to be taken into account. It is assumed that the representative worker is charged the industry average commission. It is assumed also that the participant contributes to the system on a regular basis. In the Chilean case, the cumulative charge ratio at retirement is an accurate measure of administrative costs for older participants who made contributions from the start of the system and retired before the year 2000. 13. The salary for the cohort that retired in December 2000 was set at 285,000 Chilean pesos, the average wage of the contributors to the pension. Wages are as- sumed to grow at 2 percent a year in real terms and the contribution for disability and life insurance is assumed to be a constant 0.7 percent of the worker's salary, its average level during the past 10 years. Precise information on insurance premi- ums prior to 1990 is not available. 14. Only 393 people retired in 1983 from the funded system. Even by 1987 the number was less than 8,000. 15. The contrast of these results with previous evidence (e.g., James, Smalhout, and Vittas 2001) stems from focusing on commissions at a point in time instead of focusing on their cumulative effect over a worker's career (as shown by the cumu- lative charge ratio). 16. Previous reports have used statistics reported by the superintendent in the 1990s that have since been revised. If these ratios are calculated using the commis- sion rates originally reported by the superintendent, the ratio for the average con- tributor would be about one-third. 17. Despite this fact, the net intergenerational impact in Chile is probably still positive for earlier generations because of the recognition of their accrued rights under the previous, generous system and the high gross returns during the first years of the funded system. In other countries, the intergenerational impact may be different. 18. If an asset-based structure had been chosen instead of the salary- and con- tribution-based one, any up-front costs could have been more evenly spread over time. On the other hand, workers with low contribution densities--often the poorer ones--may have been worse off than those workers with higher densities because under an asset-based structure commissions are normally deducted every month even when there are no contributions. Moving to an asset-based charge now, moreover, would be a very complex process. 19. In chapter 6 we identified "improvidence," or regretting not having saved enough when old age is reached, as the main justification for mandating savings for retirement. It would be inconsistent to mandate savings during the accumulation stage but permit lump-sum payments at retirement, unless improvidence disappears HOW WELL HAS THE "SAVINGS" COMPONENT PERFORMED? 163 gradually as the individual gets older and realizes the need for providing an adequate level of retirement income. 20. Of course, insurance companies may miscalculate increases in the average life expectancy of a specific age cohort and inadvertently provide protection against this longevity risk to those cohorts. 21. Valdés-Prieto (2002b) argued that there is limited empirical evidence of ad- verse selection in annuities markets in Chile. 22. In Chile workers who want to draw their pension--and retire from the mandatory pension system, although not necessarily from the workforce--before the official retirement age (60 years for women, 65 years for men) are required to have a benefit that exceeds 110 percent of the state-guaranteed minimum pension and must be greater than 50 percent of the worker's average real wage over the last 10 years prior to the retirement request. If the benefit exceeds 70 percent of the worker's average real wage over the last 10 years prior to the retirement request and 120 percent of the statutory minimum pension, the rest of the accumulated balance can be taken as a lump sum. Similar rules are in place in Argentina, El Salvador, and Peru. 23. We abstract here from the impact of interest rates on the value of the accu- mulated balance. 24. The assets are transferred between funds on a gradual basis over a four-year period. 25. The previous Fondo 1 and Fondo 2 have been renamed C and E, respectively. 8 The Preferences That Individuals Reveal AS POPULATIONS GROW OLDER THE economics of insurance prescribe indi- vidual saving as the most efficient and sustainable means of ensuring ade- quate income in old age. At the aggregate level it becomes increasingly costly for a shrinking labor force to finance the pensions of a growing number of retired elderly purely through public PAYG pooling arrange- ments. At the household level workers will be increasingly reluctant to pay the higher payroll taxes necessary to sustain pure pooling PAYG systems. However, pooling remains an important insurance instrument to lower losses from the relatively rare risks of poverty in old age, unexpected longevity, disability, and untimely death.1 Because the market for insur- ance often fails, and even where it flourishes cannot provide private pool- ing instruments to insure against certain loses relevant to retirement in- come--such as inflation--there is a role for government in augmenting the set of instruments at households' disposal through social insurance.2 Governments in Latin America have taken up this role with enthusiasm, as shown by the proliferation of publicly administered and/or mandated health, disability, retirement, and unemployment insurance systems in the region since the 1920s. However, these social insurance institutions can be detrimental if designed with little regard for the changing nature of the risks they seek to cover, if they ignore privately available alternatives and if they pay no attention to what individuals and households reveal as their prefer- ences for what governments should provide. When social security is de- signed with little regard for individual incentives and household preferences workers are likely to turn away from the systems, thereby jeopardizing the plans' financial sustainability and effectiveness at pooling risks. This chapter focuses on the low rates of participation in Latin America's national retirement security systems. We particularly emphasize the impact of pension reforms on the incentive for workers to seek formal coverage when the mandate to participate in a national pension system can generally 165 166 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA be evaded and where alternative saving and investment options exist--from financial instruments like bank deposits and private annuities taken up voluntarily to nonfinancial assets such as housing, other property, small en- terprise, and even the accumulation of human capital (education and train- ing that make workers more productive). As in other chapters of this book, rather than discuss the government's objective of covering the population, we focus on the household's objective of adequately insuring itself against the loss of earning ability that comes with aging: whether households have access to adequate coverage and whether they choose to participate or opt out of government-administered or -mandated pension systems. There are three main findings. First, after analyzing individual and house- hold behavior with respect to national pension systems (using data from reg- ularly deployed surveys as well as two surveys focused on risk, savings, and insurance in Chile and Peru), we provide evidence that people generally are more likely to make rational decisions with respect to securing adequate re- tirement income than is typically assumed. This manifest rational behavior regarding savings and investment choices makes it all the more important for governments to provide the instruments that people want to secure adequate income in old age. After all, if on the whole people behaved irrationally with respect to retirement income it would matter little what they wanted, and governments would be justified in stepping in to tell them. Second, this chapter presents evidence that people expect from govern- ments what governments can credibly deliver. In Chile, where there is con- siderable trust in government, the contribution behavior of workers suggests that what they want from government is some insurance against poverty in old age. Workers tend to contribute to the public system just enough to qualify for government topping-up--namely, the assurance of a minimum pension to insure against old-age poverty. This behavior may reveal a pref- erence for government-provided instruments for pooling over saving. Third, in Peru, where there is considerably less trust in government, survey data suggest that workers may value regulatory oversight of pri- vately managed pension plans more than instruments to insure against the losses associated with old-age poverty. The latter result is "contaminated" by the fact that the Peruvian government has not implemented a poverty prevention component for the majority of affiliates to the reformed pen- sion system, but the results in Peru indicate that even with weakly enforced property rights investment in real estate acts as a substitute for the formal retirement savings system. Is Low Participation Evidence of Social Exclusion or Individual Choice? Much of the literature on coverage of pension systems argues that work- ers' access to protection is determined by the degree of unionization in a THE PREFERENCES THAT INDIVIDUALS REVEAL 167 particular sector and industry of employment (Mesa-Lago 1991; Tokman 1992; World Bank 1994; Uthoff 1997; Marquez and Pages 1998; Mesa- Lago 2000). Several studies have pointed out that with the changes in oc- cupational structure in Latin America over the past two decades--that is, an increasing proportion of the workforce that is self-employed or work- ing in small firms (ILO 1999; de Ferranti et al. 2000)--a growing number of workers are excluded from pension programs because coverage in these sectors is far lower than is typically found in public and private large-scale manufacturing and in the civil service (IADB 2000). This strand of the lit- erature characterizes low rates of coverage as evidence of broader social exclusion linked to segmented, discriminatory labor markets and the fail- ure of governments to provide greater access through better education and opportunities for social advancement. Access to social security is determined by occupational category and the size of the firm where workers are employed (Mesa-Lago 1991; Uthoff 1997). Levels of participation differ according to the amount of political pressure for inclusion in the national system that certain groups of work- ers can bring to bear (Mesa-Lago 1991, 2000). Countries with greater rates of urbanization, industrialization, and unionization and with a greater share of salaried employment (relative to self-employment) will have higher rates of coverage. Countries that still have predominantly rural, agricul- tural economies, where labor is less unionized and where there is a large share of self-employed people in the workforce, have lower levels of cov- erage. Workers in part-time jobs and people temporarily employed with- out a contract are also less likely to be covered. Thus, the workers that national pension systems are especially intended to protect are those least likely to enjoy the benefits of protection (Marquez and Pages 1998). The social-exclusion literature does provide plausible arguments and convincing empirical evidence to explain why workers go without cover- age; however, it does not tell the whole story. Barrientos (1998a), James (1999), and Holzmann, Packard, and Cuesta (2000) looked beyond the social-exclusion arguments and presented a number of hypotheses to explain why rational, nonmyopic individuals and households may choose to avoid formal retirement income security systems, even if these are actu- arially fair and/or include privately owned and administered savings accounts. Where formal retirement security is bundled with unrelated government programs and regulations, the costs of compliance to the individual (or small firm) may be prohibitive. Coverage under a formal social security sys- tem is often nested deeply within the broader regulatory and taxation framework of the economy. Even where single-pillar pure PAYG systems have been replaced with less centralized systems based on individual retire- ment accounts, participation may require payment of taxes, compliance with regulations, and adherence to labor standards totally unrelated to in- come security in old age (Holzmann, Packard, and Cuesta 2000). 168 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA The constraints imposed by formal pension systems on many work- ers--especially poorer entrepreneurs--may be more binding on their in- vestments in productive enterprise than they are beneficial to smoothing consumption. Avoiding the pension system may be optimal given capital and credit constraints on investment choices. The opportunity costs of vesting scarce capital in a formal retirement security plan, no matter how actuarially fair, may be too high. The inability to draw on saved funds in times of hardship can place unacceptable liquidity constraints on workers. This is especially true of entrepreneurs, farmers, and rural nonfarm self- employed populations whose wealth is held in illiquid forms or whose in- come is largely seasonal (Holzmann, Packard, and Cuesta 2000). Furthermore, mandatory contributions to social insurance can lead to welfare losses for poorer people with high rates of discount. Where household income may be just sufficient to meet immediate, basic needs for survival, saving for old age may not be rational (James 1999; Will- more 2001b). Poorer households will place greater value on consumption today than on consumption tomorrow or far into the future. If the time preference rate is higher than the market rate of interest and credit is ex- pensive or rationed, the shadow discount rate is even greater. Thus, for lower-income households mandatory contributions can lead to major welfare losses, and participation in a formal social security system may place an intolerable constraint on household efforts to smooth consump- tion (James 1999). Additionally, income security in old age may not be the primary risk concern in poorer households. The profile of risks faced by the poor may feature the less predictable shocks to income, such as disability and sud- den illness, more prominently. This line of argument is strengthened by the link between income (nutrition/health) and mortality; that is, poorer peo- ple would rather consume income today than save and consume in the fu- ture when, because of their relatively higher mortality rate, they may not be around to collect a pension. These factors augment the implicit tax component of mandated retirement contributions for poorer households, whether they live and work under a single-pillar pure PAYG regimen or one that includes mandatory retirement savings accounts. For many households in developing countries, traditional, family-based systems of old-age security may provide superior income smoothing and cover against the risk of poverty in old age. There is ample evidence that agents (as individuals, as households, and within households) engage in consumption smoothing and risk management to mitigate the impact of fluctuating incomes. Where formal insurance markets have failed to coa- lesce or have broken down because of moral hazard and adverse selection, the extended family and community still fill the gap. The majority of the world's elderly people rely solely on informal and traditional arrange- ments for retirement income security (World Bank 1994; James 1999).3 Traditional structures involving resident elderly parents or expected THE PREFERENCES THAT INDIVIDUALS REVEAL 169 reliance on children (Becker and Tomes 1976; Appelbaum and Katz 1991; Hoddinott 1992) are still prevalent in Latin America (IADB 2000) and may act as a superior, more flexible substitute for the formal social insur- ance system. Furthermore, poor and nonpoor households may put little stock in the promises of government. Even reformed pension systems may be suffering from an inherited lack of credibility in formal social security institutions. Even if workers were fully aware of the benefits defined by social security legislation they might perceive a high political risk to their promised pen- sion stemming directly from a government's lack of credibility. If govern- ments have a track record of frequently changing the rules of the game (vesting requirements, the benefit formula, indexation, or minimum pen- sion guarantees), if inflation taxes are high, and if funds earmarked to pay retirement benefits are mismanaged or depleted, workers may consider the risk of not receiving a pension to be too high and may heavily discount the returns of investing in being covered. Where private individual accounts have been introduced, households may find the burden of financial risk in reformed systems to be too high. A criticism of the new multipillar model is that it requires workers to as- sume a greater share of risk to their income security in retirement than do pure PAYG systems (Diamond 1993, 1998; Orzag and Stiglitz 2001; Barr 2000). Under purely public PAYG regimens government assumes a hefty share of risk (demographic, macroeconomic, and financial), and workers' only risk is largely political (i.e., that their benefits will be cut or their ac- cumulated rights ignored in the political process of setting government budgets). Under the reformed systems based on privately invested savings in individual accounts, workers are burdened with the weight of a wider range of risks to adequate retirement income. In a system of individual ac- counts, although there may be relatively greater certainty of receiving some pension there is substantially greater uncertainty as to the amount of the pension--that is, of accumulating a balance sufficient to guarantee an adequate stream of income in old age.4 A significant number of workers may consider mandated arrangements where the bulk of the pension is de- termined by variable returns from private investment to be too risky. The perception of risk in the reformed systems my be even greater in countries with poorly regulated capital markets, in economies vulnerable to fre- quent external shocks, or where the full faith and credit of government are called into question and even publicly issued. Fixed-income securities, which typically dominate the portfolios of the new pension funds in Latin America, pay substantial risk premiums. Recent events in Argentina are a poignant illustration of this argument. The mix of retirement investments adopted by households will neces- sarily depend on the relative costs and benefits of each investment and its efficiency in balancing returns with risk. A portfolio of formal and infor- mal assets (financial assets, own home, other property, own business) and 170 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA household-based strategies (education of a child, reciprocal arrangements among members of the extended family) may have higher returns and lower risks (beholden children and relatives) than do the portfolios offered by the national pension system. Ultimately, pension systems based on compulsory contributions to either a purely public PAYG system or one including individual savings accounts may simply crowd out voluntary household arrangements and what are often considered supplementary third-pillar retirement investments offered privately by the formal finan- cial sector. Who Contributes to Social Security? Survey Evidence from 13 Latin American Countries In a background paper for this book Packard, Shinkai, and Fuentes (2002) analyzed the contribution behavior of workers using household- level data from 13 Latin American countries. As in Barrientos (1996, 1998a) for Chile and Holzmann, Packard, and Cuesta (2000) for Chile and Argentina, the authors performed separate probit maximum likeli- hood estimations of the probability that working individuals are con- tributing to the national retirement security system in their country. Table 8.1 summarizes the results by presenting the positive or negative impact of selected individual, employment, industry, and household characteris- tics on the likelihood that workers are contributing to the national pen- sion system in each country.5 Although the authors are restricted in the set of specific hypotheses that can be tested by using survey data from a large number of very diverse countries, interesting regional trends emerge from their analysis. It is not surprising that better-educated workers earning higher incomes are more likely to be contributing to the national pension system. It is in- teresting to note that working women, often considered an "excluded" group, are more likely to contribute in 8 of the 13 countries examined. This finding may reveal greater prudence with respect to old age among working women relative to men, but it may also be explained by their relatively greater numbers in professions traditionally well covered by public social insurance systems such as public administration, nursing, and teaching. As shown in table 8.1, however, structural barriers clearly remain be- tween formal coverage and workers in certain sectors and industries. Cer- tain segments of the working population face a lower likelihood of access to formal income protection in old age. These people include married women (although many of these may be covered through the contributions made by their husbands), workers in rural households and people employed in agriculture, transportation, retail and services, and the con- struction industries (relative to workers in manufacturing). The probability Table 8.1 Probability That Workers Contribute to Social Security Is Determined by Individual, Household, and L Market Factors Costa Dominican El Nica- Para- Bolivia Brazil Chile Colombia Rica Republic Ecuador Salvador Mexico ragua guay Peru 1993 1997 1996 1998 1997 1998 1995 1998 1996 1998 1995 1997 Individual characteristics Age Elderly Income Education Woman Married Wife Rural Employment characteristics Single job Subordinate Professional Self-employed Only self- employed Self-employed professional 171 Work hours (Table continues on the following 172 Table 8.1 (continued) Costa Dominican El Nica- Para- Bolivia Brazil Chile Colombia Rica Republic Ecuador Salvador Mexico ragua guay Peru 1993 1997 1996 1998 1997 1998 1995 1998 1996 1998 1995 1997 Part-time 5 years with firm Contract Industry of employment (standard International Standard Industrial Classification categories, where manufacturing is omitted) Manufactur- ing Agriculture Mining Utilities Construc- tion Retail Transporta- tion Finance Community Household characteristics Size Number of old people Number of male children Number of female children Extended family Observations 4,374 136,420 45,718 36,528 11,828 6,881 11,180 18,021 23,455 6,554 5,486 8,112 Pseudo R2 0.23 0.58 0.48 0.31 0.18 0.31 0.28 0.52 0.48 0.46 0.38 0.59 Notes: The table presents probit regressions; dependent variable "contributes to social security." " " and " " indicate variable has at least a 10 percent statistically significant positive or negative effect, respectively, on the probability of contrib to the national retirement security system. Blank cells indicate the variable is not statistically significant to the probability of contribution. Source: Packard, Shinkai, and Fuentes (2002). 173 174 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA of access to the social security systems is also lower for workers in small firms and those without a legal employment contract. Although this hy- pothesis was not tested, this may reflect small firms' costs of affiliating workers and making regular contributions, including the costs of compli- ance in heavily regulated and taxed product and factor markets. The results summarized in table 8.1 also show that workers who hold what have been traditionally considered "better" jobs in Latin America-- in larger firms, in manufacturing, and in the civil service--are more likely to be contributing. Indeed, many workers and their households may be ex- cluded from retirement security systems and other forms of social insur- ance in countries with deeply segmented labor markets. The evidence in table 8.1 suggests that the growing concern among policymakers for workers in these sectors may be justified, especially in countries where workers queue for covered, formal employment. However, such conclu- sions can only be drawn from a thorough country-specific analysis of labor market dynamics: how individuals insert themselves in different sec- tors and industries, whether workers choose the sector in which they work, and whether "covered" formal employment is indeed rationed.6 Finally, in table 8.1 there is little evidence of traditional forms of re- tirement income security substituting for formal institutions. In fact, a larger share of elderly people and dependent children in the household seems to complement participation in the formal system in most countries. The positive influence of children and elderly people on the likelihood of contribution to formal pension systems may not be as surprising as it seems. Workers who have many dependent children and who are more likely to face the risk of disability and sudden death may have a higher demand for coverage under the social security system and thus may be more likely to contribute (Barrientos 1998a). Furthermore, older children and resident elderly men and women may take charge of household chores such as cooking and caring for younger children, thereby freeing parents-- especially women--to take up remunerated employment and increasing the probability that they will accrue rights in the social security system. Only in Mexico is the anecdotal preference for (male) children as insur- ance against destitution in old age--the most frequently cited traditional old-age security arrangement-- borne out by the data. In Mexico a larger share of male children in the household significantly lowers the likelihood that workers contribute to the pension system. The failure to find evidence of traditional arrangements substituting for formal institutions, however, may be more a consequence of the poor proxy variables available in the existing data to capture these arrange- ments and other factors effecting demand for formal coverage and the choice to participate. Without quantitative data on the role played by chil- dren and elderly people in the household economy, and qualitative data on the expected/desired number of children of each gender, or the motivation for having larger families, it is difficult to detect and test the significance THE PREFERENCES THAT INDIVIDUALS REVEAL 175 of traditional retirement security arrangements. Analysis conducted in background papers for this book using recently collected data from sur- veys focused on risk, savings, and social insurance in Chile and in Peru shows that these and other factors significantly determine individual and household demand for formal coverage, and that when taken into account they render most of the access variables discussed above statistically in- significant. Do Low Participation Rates Reflect Lack of Demand? The analysis presented in the last section, taken from Packard, Shinkai, and Fuentes (2002), placed greater emphasis on factors affecting access to formal coverage in Latin America. This is primarily the result of the limi- tations of regularly deployed household surveys in most of the countries in the region. For example, among the variables available it is difficult to find data on asset holdings that may be preferred as alternative forms of retirement savings. Few surveys ask about access to credit, thereby hin- dering examination of the effects of capital constraints on consumption- smoothing behavior. Furthermore, there are few qualitative data on the role played by resident elderly people in the household or the motivation for having children. Surveys deployed recently in Chile (where individual retirement accounts have been in place for the longest period of time) and Peru (the second country in Latin America to introduce individual retire- ment accounts) correct these limitations and allow a fuller analysis of factors affecting individuals' decisions whether to participate in govern- ment-mandated retirement security systems (see box 8.1). Before presenting the results of the PRIESO surveys (social risk man- agement surveys), it is important to review the set of instruments for re- tirement income security that the governments in Chile and Peru either provide directly, mandate, or regulate. Table 8.2 presents the features of the reformed pension systems in both countries discussed in chapter 2, and provides additional relevant information that will help readers put into context the econometric results reported in this section. The reformed pension systems in the two countries are similar. In both countries dedicated private pension fund managers invest workers' accu- mulated savings in individually owned retirement savings accounts. Em- ployee participation is mandatory, but the self-employed freely choose whether to participate in the formal system. Part of workers' contributions pay for the services of the fund managers as well as the premiums for pri- vately provided life and disability insurance. Although the reformed systems are similar, crucial differences are also apparent. It is most important to an analysis of individual and household savings and insurance behavior that workers who choose to participate in 176 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 8.1 PRIESO: Social Risk Management Surveys in Chile and Peru Until recently, analysis of participation in the reformed retirement secu- rity systems in Latin America has been constrained by the limitations of regularly deployed household surveys. Several previously unavailable variables used in background analysis for this book were constructed from data collected in specially designed surveys on risk, savings, and so- cial insurance (in Spanish, Encuestas de Previsión de Riesgos Sociales-- PRIESO) conducted first in Santiago, Chile, in January 2000 and re- peated in Lima, Peru, in May 2002. The PRIESO surveys are specifically designed to identify the strategies taken by households to mitigate risks to income. In addition to traditional questions dealing with household com- position, income, and labor market activity, the surveys elicit respon- dents' opinions of the reformed pension systems, their preferences for alternative retirement security strategies, their access to credit, percep- tions of their own mortality, income shocks, and strategies for coping with risk. Although both surveys were limited to a sample of workers from the largest metropolitan region of each country, the results in several in- stances reflect national statistics. For example, among the sample of workers in Santiago who are affiliated with the pension system, only 62 percent were making contributions at the time of the survey, approxi- mately the same contribution level found by Edwards and Cox-Edwards (2000a) using nationally representative data. Among working men, 64 percent were contributing. Only 58 percent of working women made contributions; and 42 percent of the women of working age who were neither working nor searching for a job received some coverage from the system through the contributions of a spouse, leaving 58 percent of this cohort without formal coverage. The PRIESO protocol includes questions never previously asked combined with more traditional questions on household composition and labor market activity. For instance, the surveys include a wide range of questions about informal instruments to mitigate poverty in old age, and about how these instruments might substitute or comple- ment the formal pension system. The data collected in the PRIESO surveys thus provide an important empirical resource to buttress an area of research that has largely relied on qualitative and anecdotal evidence. To illustrate that point, two questions were posed to capture whether and how parents expected their children to care for them in their old age. Even a casual analysis of responses from Chile shows the rural/urban dis- parities frequently referred to in the literature on informal intrahousehold risk management (Alderman and Paxson 1992; Hoddinott 1992; Deaton 1991, 1997; Cox, Eser, and Jimenez 1998). Although 47 percent of THE PREFERENCES THAT INDIVIDUALS REVEAL 177 Box 8.1 (continued) respondents from rural areas expected to live with a son or a daughter in their old age, only 19 percent of urban respondents held the same expec- tation. Similarly, rural respondents seemed more confident that they would receive some sort of care from their children, with 67 percent giv- ing an affirmative response and only 14 percent unsure. Only 17 percent of rural respondents did not expect to be cared for by their children. Urban respondents, on the other hand, were more evenly distributed be- tween those who anticipated care from their children (34 percent), those who did not (30 percent), and those who did not know (19 percent). When asked why they did not expect either a son or daughter to care for them (28 percent of all respondents), the answer most frequently given was that they did not want to become a burden. In formal econometric analysis workers who expected to either reside with or otherwise receive care from their children were significantly less likely to contribute to the formal pension system than were those who did not expect to be cared for (Packard 2002). Readers can find more details on the PRIESO surveys (sample ques- tionnaires, survey field reports, and sampling techniques) in Chile and Peru in Packard (2002) and Barr and Packard (2002a, 2003), online at Web page of the Office of the Chief Economist, Latin America and the Caribbean Region: www.worldbank.org/keepingthepromise, or bancomundial.org/cumpliendolapromesa. the formal retirement security system in Peru are allowed to choose be- tween a downsized public PAYG plan and private individual accounts for their earnings-related pensions. Furthermore, although workers' retire- ment security in Chile is underpinned by poverty prevention benefits-- both a pension guarantee for workers who have made a minimum number of contributions and a targeted (but rationed) social assistance benefit to elderly, indigent people--neither type of public pooling instrument to pre- vent poverty is available to the majority of workers affiliated with the re- formed pension system in Peru (see box 8.2). In analyzing the coverage of an old-age income security system--espe- cially the demand for coverage--a revealing choice variable is an individ- ual's period of contributions to the pension system as a share of his or her life as part of the labor force--or the density of his or her contributions. This measure has long been unavailable to researchers in developing coun- tries, even those countries in Latin America that have introduced individ- ual accounts.7,8 Respondents to the PRIESO surveys were asked the month and year that they first contributed to the social security system. 178 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Table 8.2 Reformed Pension Systems in Chile and Peru Are Similar, but There Are Important Differences Chile Peru Year of reform 1981 1992­1993 Earnings-related public Closed Remains for workers who PAYG system? choose publicly managed second pillar Total payroll tax rate, 33 18 pre-reform (%) Total payroll tax rate, 20.0 20.5­22.0 post-reform (%) Participation of new Mandatory Voluntary (but workers workers? must choose a second- pillar system: either AFPs or the downsized PAYG) Participation of self-employed Voluntary Voluntary workers? Remaining separate system No No (with exceptions for for civil servants? some subnational systems) Dedicated fund managers AFP AFP Contribution to AFP/IRA 10 8 (% of wage) Fees and insurance premiums 2.31 3.73 (% of wage) Switching between fund Two times Once annually managers? annually Payout options Annuity or Annuity or scheduled scheduled withdrawal withdrawal Minimum return on Relative to Relative to average investment? average Minimum contributory Yes Only for affiliates of pension? PAYG and of the AFP system who were born before 1945 Social assistance pension ? Yes No Note: IRA individual retirement account. Source: Authors' compilation. They were then asked to estimate the total period (in years and months) during which they had failed to contribute for whatever reason--inactiv- ity, unemployment, employment without a contract, or self-employ- ment--since they started. A "contribution density" variable was con- structed from these responses.9 THE PREFERENCES THAT INDIVIDUALS REVEAL 179 Box 8.2 Peru's Reformed Pension System: Multipillar in Name Only Proponents of mandated individual retirement accounts are often unfairly accused of paying little attention to the remaining pillars of the multipil- lar model. However, the history of Peru's pension reform provides evi- dence to support this accusation. Since the December 1992 passage of the law that introduced privately managed retirement savings accounts, Peru's retirement security system has been multipillar in name only. Peru is exceptional among recent reformers in the region for not pro- viding a poverty prevention pension to the majority of workers who affiliated with the new system of private accounts. A first-pillar mini- mum pension guarantee still exists for workers affiliated with the re- formed PAYG plan, and a guarantee similar to the minimum pension guarantee in Chile was put in place in July 2002 for older affiliates of the private plan. But the majority of workers covered by a formal re- tirement security system are not covered against poverty in old age. The failure to set up a contributory pillar is particularly worrisome because Peru does not have a social assistance benefit targeted at its elderly indi- gent population. For most of Peru's workers the set of institutions put in place by struc- tural reforms in 1992 does not yet represent a diversification of the risks to income in old age as envisioned by the reformers. Rather, it is more a transfer of the bulk of these risks from the state to households. The gov- ernment is currently exploring how this imbalance can be redressed in an equitable and fiscally sustainable way. Source: Barr and Packard (2003), available online at: www.worldbank.org/keepingthepromise. In figure 8.1 the sample of affiliated men and women who responded to the PRIESO survey in Santiago, Chile, is divided into deciles by their contribution density. Taking eligibility for the minimum pension guaran- teed by the government as the minimum level of coverage offered under the retirement security system, the required months of contribution for the guaranteed benefit (240 months, or 20 years) are divided by the average number of working months for men and women. The resulting ratio is the "contribution density threshold" that affiliates must cross to qualify for the minimum pension guarantee (the bold, horizontal axis of each graph).10 Assuming that workers will maintain their reported rate of con- tribution to the system, affiliates whose contribution density places them above the threshold will qualify for (at least) the minimum level of cover- age, but those below will not.11 180 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Econometric analysis shows that the contribution density of workers who entered the labor market and began participating in the system after the introduction of individual accounts in 1981 is significantly greater than that of workers who began contributing prior to reforms (Packard 2002). This microeconomic evidence of an improvement in workers' in- centives to participate in formal retirement security systems lends support to the country-level evidence of an improvement in incentives discussed in chapter 5. However, it is immediately apparent in figure 8.1 that a large gap in coverage remains. A greater share of affiliated women--about half--lies below the threshold of contributions necessary to qualify for the minimum pension guarantee. Although this raises concern, many of these women may be entitled to some benefit through the current and past con- tributions of a husband. What is particularly worrying is that if we assume no change in workers' current contribution behavior, 30 percent of affili- ated men are unlikely to qualify for the minimum benefit. Readers should also note that the PRIESO in Chile is only representative of greater met- ropolitan Santiago, and thus is likely to understate the shortfall in regular contributions reported in national surveys that include less-industrial rural Figure 8.1 Chile: Reported Contribution Density (Contribution months/months in EAP) 1.00 1.00 0.93 0.90 0.85 0.77 0.80 0.69 0.70 months 0.61 0.60 0.51 working 0.50 of 0.40 0.41 0.30 Percentage0.20 0.26 0.10 0.07 0 1 2 3 4 5 6 7 8 9 10 Deciles of affiliated men, by contribution density THE PREFERENCES THAT INDIVIDUALS REVEAL 181 Figure 8.1 (continued) 1.00 1.00 0.89 0.90 0.77 0.80 0.70 0.65 months 0.60 0.54 working 0.50 of 0.40 0.44 0.30 0.33 Percentage0.20 0.21 0.10 0.10 0 0.01 1 2 3 4 5 6 7 8 9 10 Deciles of affiliated women, by contribution density Note: Affiliated men and women who responded to the PRIESO survey in Santiago, Chile. Source: Packard (2002) using PRIESO (2000). areas where access to the pension system is relatively limited. Indeed, a national survey of affiliates based on the PRIESO and conducted by the Chilean government reveals that average contribution density among af- filiates of the pension system is significantly lower. Additionally, because preferences for time and risk are fundamental to the decision to insure and to save, particular attention was paid to collecting and analyzing empirical measures of risk aversion and time pref- erence. In a background paper for this book, Barr and Packard (2002a) analyzed how time and risk preferences determine contribution behaviors among a subsample of respondents to the PRIESO in Chile. The results of experimental techniques used to measure individuals' aversion to risk are presented in figures 8.2 and 8.3. The first point to note is that although people who start their own busi- ness are often thought to have a greater tolerance for risk, there are no 182 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 8.2 There Is No Difference in Risk Preferences be- tween Employees and Self-Employed Workers Who Responded to the PRIESO Survey in Chile Employee Self-employed worker 0.2 0.1 Fraction 0 2,000 3,000 4,000 5,000 2,000 3,000 4,000 5,000 Chilean pesos, Ch$ Note: Individual certainty equivalents for employees and self-employed workers: risk tolerance increases right of the origin. Source: Barr and Packard (2002a) using PRIESO (2000). significant differences in risk preferences between self-employed workers and wage employees in the PRIESO sample in Chile. This finding indicates that self-employed people are of particular interest when studying contri- bution behavior, given that in Chile as in other countries in the region self- employed workers are allowed to freely choose whether to contribute. Be- cause they are free to reveal their preferences with respect to the pension system although wage employees are explicitly constrained by the man- date to contribute, and because no significant difference in the relevant preference indicator (in this case, risk aversion) can be found, inferences can be drawn about the behaviors of all workers from the contribution de- cisions of those who are self-employed.12 Furthermore, as shown in figure 8.3, despite the assumption that the formal pension system is the only source of insurance against poverty in old age, the authors found that self- employed contributors to the pension system--those workers in Chile who are completely free to manifest their preference with respect to retirement saving--are significantly more tolerant of risk than the self- employed workers who choose not to contribute to the AFP system. This suggests that the Chilean pension system may be viewed with some trepi- dation by workers considering whether to contribute. This may be because those who are more risk averse prefer to rely on alternative forms of retirement income security, or may be deterred by the financial risks asso- ciated with the capital markets in which retirement savings are invested THE PREFERENCES THAT INDIVIDUALS REVEAL 183 Figure 8.3 Self-Employed Workers Who Contribute to the AFP System in Chile Have a Greater Tolerance for Risk Noncontributing employees Contributing employees 0.2 0.1 0 Noncontributing self-employed Contributing self-employed Fraction workers workers 0.2 0.1 0 2,000 3,000 4,000 5,000 2,000 3,000 4,000 5,000 Chilean pesos, Ch$ Note: Comparison of certainty equivalents between employed and self- employed, and contributors and noncontributors to pension system: risk tolerance increases right of the origin. Source: Barr and Packard (2002a) using PRIESO (2000). under the reformed system. Alternatively, these potential clients may be poorly informed about the system and the performance of the private fund managers. Packard (2002) took the analysis of savings and insurance preferences in Chile a step farther, using a wider range of variables from the PRIESO survey, including the risk and time preference measures, in contribution re- gressions similar to those presented in the last section to capture the impact of household preferences and alternative investments. The expectation of care from children and the amount spent on their education significantly lower the likelihood of contribution to the system. Investments in tools and machinery also lower the likelihood of contribution to the system. By including the contribution history variable shown in figure 8.1 in the analysis of the contribution behaviors of all workers (not just those who are self-employed), we are able to separate the public risk pooling element 184 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA of the Chilean pension system (the minimum pension guarantee) from the dominant private savings element and find evidence of portfolio behavior among the sample of affiliates. Workers who have met the contributory re- quirements to qualify for the minimum government-guaranteed retirement pension by contributing for at least 20 years are significantly less likely to continue contributing to the AFP system. The likelihood that these work- ers make additional contributions beyond the eligibility threshold de- creases as the rental value of their homes increases. Finally, after taking account of alternative strategies and investments for retirement security as well as workers' contribution histories, the ear- lier finding that individuals with a greater tolerance for risk contribute to the pension system is confirmed. This suggests that there are retirement in- vestments in Chile that are perceived as relatively less risky than saving in the reformed pension system. The results provide evidence that housing, household enterprise, and even the education of children are among the alternative investments being pursued by individuals. It is important to note that including this wider range of variables in the analysis of contribution behavior in Chile renders most of the "ac- cess" variables--found to be significant in table 8.1 in the last section-- Figure 8.4 Peru: Reported Contribution Density (Contribution months/months in EAP) 1.00 1.00 1.00 0.98 0.90 0.80 0.80 0.70 0.69 months 0.60 0.58 0.50 working 0.45 of 0.40 0.30 0.29 0.20 Percentage 0.10 0.14 0 0.01 1 2 3 4 5 6 7 8 9 10 Deciles of affiliated men, by contribution density THE PREFERENCES THAT INDIVIDUALS REVEAL 185 Figure 8.4 (continued) 1.00 0.98 1.00 1.00 0.90 0.82 0.80 0.70 months 0.66 0.60 0.52 working 0.50 of 0.40 0.38 0.30 0.26 Percentage 0.20 0.10 0.13 0 0.02 1 2 3 4 5 6 7 8 9 10 Deciles of affiliated women, by contribution density Note: Affiliated men and women who responded to the PRIESO survey in Lima, Peru. Source: Barr and Packard (2003) using PRIESO (2002). statistically insignificant to the likelihood that men participate in the pen- sion system. Thus levels of participation in the retirement security system in Chile are likely to reflect relatively low household demand for the sys- tem as a savings and investment vehicle. That said, there are still signifi- cant barriers to women's participation in the pension system, related primarily to the size and place of employment and their (lack of) con- tractual relationship with employers. The second PRIESO survey was conducted in May 2002 with a repre- sentative sample of working individuals in Lima, Peru, where since the in- troduction of the multipillar model workers choose between alternative forms of government-mandated coverage for the earnings-related (second pillar) portion of their pensions. As described earlier, since the reform was implemented every new cohort of workers in Peru chooses between indi- vidual accounts and the reformed public PAYG plan. (The same system ex- ists in Argentina and Colombia, although in Colombia workers can choose 186 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA every three years.) Figure 8.4 plots contribution densities calculated from the history of contributions of men and women in the PRIESO sample who are affiliated with either branch of Peru's retirement security system. Al- though very similar to the contribution densities of affiliated workers in Chile (shown in figure 8.1), noteworthy differences are apparent. In figure 8.4 the bold, horizontal axis represents the threshold density of contributions that would be necessary to qualify for a minimum pen- sion guarantee similar to Chile's guarantee. However, as discussed in box 8.2, Peru does not offer a first-pillar benefit to younger affiliates of its AFP system. There are three main differences: 1. Twenty percent of men and women affiliated with Peru's pension system have a perfect (1.00) contribution density. Although there is only a small sample of countries in Latin America that have introduced indi- vidual accounts, a drop in the share of regular contributors among affili- ates of the new systems is found to be common over time. Thus the large share of affiliates with perfect contribution density relative to Chile may simply reflect a younger system. 2. A greater share of affiliated women in Peru (60 percent) than in Chile would be likely to fulfill the contributory requirements to receive a mini- mum pension guarantee. Furthermore, those women whose contributions fall short of the threshold would be closer to qualifying for the minimum benefit than would affiliated women in the same contribution decile in Chile. 3. The coverage gap--measured by contribution density below the min- imum eligibility threshold--for affiliated men in Peru is worse. As in Chile, 30 percent of affiliated men from the PRIESO sample would not be likely to meet the contribution requirements to qualify for the minimum level of coverage. However, the 30 percent of affiliated men whose contributions fall short in Peru have greater shortfall to make up. As is true in Chile the data on contribution history affirm the policy shift away from pure public PAYG pooling and toward some private sav- ing, and they indicate that the introduction of individual accounts led to an improvement in the incentives to participate in Peru's pension system. Among workers who were affiliated with the national retirement security system before the introduction of individual retirement accounts, those who switched to the AFP system have a significantly (at 1 percent confi- dence level) greater contribution density than those who remained affili- ated with the reformed PAYG system. Furthermore, even among workers who affiliated after 1993 when the pension reform came into full effect, those who chose the AFP system over the reformed PAYG have a greater density of contributions. Turning to the measures of risk preference gathered in Peru (presented in figures 8.5 and 8.6), as was found in Chile there is no significant differ- THE PREFERENCES THAT INDIVIDUALS REVEAL 187 Figure 8.5 There Is No Difference in Risk Preferences be- tween Employees and Self-Employed Workers in Peru Employees Self-employed workers 0.2 Fraction 0.1 0 5 10 15 20 5 10 15 20 New Peruvian Soles Note: Individual certainty equivalents for employees and self-employed workers; risk tolerance increases right of the origin. Source: Barr and Packard (2003). ence in the risk preferences between employees and self-employed workers. Unlike the situation in Chile, however, self-employed workers with a greater aversion to risk contribute to Peru's retirement security system. Among contributing self-employed people, those most averse to risk choose the private branch of the pension system and make contributions into an individual retirement account. Two sets of choices can be analyzed: the choice to switch from the pub- lic to the private system made by workers affiliated prior to 1993, and the choice of one branch of the system over another made by individuals join- ing the labor market for the first time since 1993, namely, those workers for whom competing public and private options existed the first time they made a choice with respect to the formal retirement security system. Among workers affiliated with the pension system prior to reforms, those who switched to the private system after 1993 have a significantly greater aversion to risk than those who stayed (although the statistical sig- nificance of this result is very weak). Furthermore, among individuals who affiliated after the introduction of individual accounts in 1993, those with a greater aversion to risk also chose the private AFP system over the re- formed PAYG still administered by the government. The inverse also holds. Among affiliates who joined the system after reforms, those who contribute to the public PAYG branch have a significantly greater toler- ance for risk. 188 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 8.6 Self-Employed People in the AFP System in Peru Are More Risk Averse Nonaffiliated employees Employees affiliated Employees affiliated with the public system with the private system 0.2 0.1 0 Nonaffiliated self-employed Self-employed workers affiliated Self-employed workers affiliated Fractionworkers with the public system with the private system 0.2 0.1 0 5 10 15 20 5 10 15 20 5 10 15 20 New Peruvian Soles Note: Comparison of certainty equivalents between employees and self- employed workers, and between affiliates and nonaffiliates of the pension system; risk tolerance increases right of the origin. Source: Barr and Packard (2003). What of those workers who do not contribute to either branch of the formal pension system in Peru? As in Chile investment in housing and other residential property acts as a substitute for the formal retirement savings system, despite weakly enforced property rights in Lima. Workers for whom the mandate to save is not binding hold the greater share of their accumulated assets in the form of housing. Additionally, the greater the share of children in their households, the less likely workers are to con- tribute to the formal pension system. If investment in housing and children can be considered substitute strategies for securing well-being in old age, workers' participation decisions are rational and consistent.13 Although there is no minimum pension guarantee in Peru's AFP sys- tem, the cutoff in workers' contributions to the pension system is similar to that observed among workers who cross the eligibility threshold for the guarantee in Chile. The missing first pillar in Peru limits the conclu- sions that can be drawn and the comparisons that can be made with re- sults from Chile, but this behavior could reveal perceived falling marginal THE PREFERENCES THAT INDIVIDUALS REVEAL 189 returns to workers' investment in the formal pension system and a pref- erence for greater diversification among saving options. Implications of Household-Level Analysis: The Preferences Individuals Reveal Public interventions to help households mitigate loss of earnings ability and poverty in old age--and other adverse shocks to income over the life cycle--are a relatively new phenomenon even in industrial countries. In most developing countries the majority of households still rely on mecha- nisms that lie in the private domain--provided formally through the mar- ket or informally though family and social networks. However, evidence that investments for income security that require cohesive social networks are increasingly less reliable (Deaton 1991; Hayashi, Altonji, and Kot- likoff 1996) indicates that a role for government clearly exists. That said, it is important in assuming this role that careful attention is paid to com- plement and not distort private choices or to displace private options. Latin America's experience with overly generous guarantees from poorly managed social security institutions that privileged a relatively small, select group at an unsustainable cost to the majority of households provides an example of what policymakers should seek to avoid. It is of- ten argued that failing formal public institutions have caused the majority of workers to turn away from social security. Pension reforms that intro- duced the multipillar system, based primarily on individual retirement accounts, were intended to bring workers back to the formal economy by increasing incentives to contribute. Evidence presented in this book sug- gests these reforms have had a positive impact on worker participation.14 Despite the positive impact of reforms, policymakers still have cause for concern. Although reforms have increased the incentives to contribute, the low share of regular contributors in the labor force provides the best ar- gument against complacency. The empirical evidence summarized in this chapter suggests that although reforms to social security were intended to sever the link between coverage and a worker's sector or place of employ- ment, barriers to access remain. Across the region certain segments of the working population may still be excluded. Binding minimum-wage legis- lation, tenure-based employment security provisions, overly regulated product markets, or even the transaction costs of affiliation and contribu- tion to the pension system could increase the costs of participation for small businesses above what they can afford to remain in operation. If structural reforms to the social security system and to related institutions succeed in eliminating explicit barriers to participation (e.g., conditioning access on employment in a certain industry or sector), remaining implicit barriers (primarily to do with transaction costs) may be lowered as re- formed systems become more efficient. 190 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA However, the data on contributors to government-administered and/or -mandated pension systems ignore other forms of insurance and savings that individuals and households may engage in on their own, and thus can overstate the degree of vulnerability to the loss of earnings ability and the risk of poverty in old age. Because the actions households take to secure income in retirement mostly lie outside government-mandated systems (re- liance on family, private savings, house purchases, and take-up of private insurance policies) and are not reflected in participation rates, the truly "vulnerable" population--households that really are facing the likelihood of poverty in old age--is probably smaller than that reflected in official statistics. More households may be covering themselves and securing an adequate retirement income. For this reason it is important to take account of al- ternative assets, savings, and insurance in the widely applied simulation analysis of retirement savings in developing countries where data are avail- able, and to collect these data where not available. Although some barri- ers to access remain even after structural reforms, the analysis presented in this chapter, using new data on private savings and insurance and the re- tirement investments preferred by households, indicates that researchers and policymakers concerned with low rates of participation should focus relatively less on issues of social exclusion and relatively more on the fac- tors affecting household demand for formal coverage. The results of our analysis of PRIESO data from Chile indicate that with respect to individual preferences the Chilean pension system may be overly engineered. Workers seem to be using a system intended to act pri- marily as a vehicle for savings--with a small pooling component--prima- rily as a risk pooling device. Each cohort of workers that completes the minimum required months of contributions to the system may be content to receive the government's pension guarantee. Given the modest amount of the guarantee (which has averaged between 80 percent and 90 percent of the minimum wage in the last 10 years), one would hope that these workers would continue to save or invest for retirement outside the system. Contrary to the default assumption of worker irrationality, the evidence drawn from the PRIESO survey in Chile suggests that many do, because the likelihood of further contribution to the pension system is lower as the market value of affiliates' property increases. Several researchers have attributed the fall in regular contribution af- ter 20 years to rational moral hazard because simulations will show that for lower-income workers every additional contribution above the threshold necessary to qualify for the minimum pension benefit is a pure tax (Vittas 1996; Cox-Edwards 2000; Edwards and Cox-Edwards 2000a; James, Cox-Edwards, and Wong 2003).15 It is important to point out that the portfolio behavior apparent among affiliates of the pension system in Chile is seen not among the working poor (for whom the rational moral hazard explanation applies) but among respondents in the fifth income THE PREFERENCES THAT INDIVIDUALS REVEAL 191 decile and higher. When they have contributed for 20 years, affiliates have the right to the minimum benefit--the pension system's remaining public pooling instrument--but only if they suffer a dramatic fall in their accumulated assets and actually need it, because eligibility is also means- tested. Given their likely lifetime earnings, many workers will not qual- ify. Each contribution to the AFP system above the eligibility threshold for the minimum pension guarantee is purely a form of savings for this group. If affiliates perceive the AFP system to be a relatively risky, costly, and illiquid savings instrument, it comes as little surprise that they mani- fest a preference for alternative, voluntary forms of savings (such as hous- ing and life insurance programs) when they have secured the guaranteed minimum annuity. Furthermore, the analysis from Chile indicates not only that house- holds save outside the mandated system--but also that with regard to the portion of their retirement portfolios the government has mandated, they may place a relatively greater value on security than on real rates of return. In Chile those workers who freely choose to contribute to the system show a significantly higher tolerance for risk. Households are content to gain el- igibility for the low, government-guaranteed annuity and continue to save outside the system, despite the variable but high real returns they could earn in the system.16 In Peru, on the other hand, workers averse to risk choose the private AFP system over the government's reformed PAYG option. Rather than contradicting the findings from Chile, however, the result may reveal a similar preference for security when set against data indicating that private financial institutions are trusted more than all three branches of the Peru- vian government (Barr and Packard 2003). Given recent revelations of sys- temic corruption in Peru's public institutions and the events leading to the ouster of the Fujimori regime, these findings may not be surprising. Fur- thermore, the relative youth of the new pension system, the failure of Peru's government to effectively implement the promised minimum pen- sion guarantee for workers who choose private individual accounts, and the lack of secure third-pillar instruments are likely to lead to very differ- ent patterns of saving and investment behavior than that seen in Chile. De- spite precarious property rights in Peru relative to Chile, however, there is evidence of workers replacing investment in the pension system with in- vestment in housing and other residential property. Conclusion Contrary to the assumption often made by policymakers with respect to retirement security, the results presented from Chile and Peru indicate that workers make rational choices. The results from Chile--where individual retirement accounts have been in place for more than 20 years--suggest 192 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA that what households may be seeking from a government-mandated instrument is a greater degree of security relative to the retirement invest- ments that they undertake freely. This may also indicate that when a social safety net to prevent households from falling into poverty in old age (ei- ther because of an exogenous shock or their own improvidence) is in place and reflects the particular degree of redistribution valued in that society, mandating private savings should be considered a largely transitory policy device. Publicly mandated but privately owned and managed individual retire- ment savings may be most effective as a transition institution to substitute for missing or inadequate instruments until private capital markets can provide efficiently priced savings and insurance vehicles. In the longer term the role of the state in mitigating loss of earnings ability and poverty in old age may be reduced to protecting households against systemic shocks to which private insurance cannot respond (such as inflation, through pri- vately traded indexed instruments), and ensuring that a retirement income safety net, financed with pooled tax payments, is securely in place. Further investigation of the level at which such a safety net should be placed so as to effectively protect households from improvidence and governments from moral hazard and a Samaritan's dilemma is sorely needed. In the medium term the optimal role of governments concerned with mitigating the loss of earnings ability and the risk of poverty in old age may be to spend less on trying to increase compliance with the mandate to save and more on ensuring that the economic and regulatory conditions enable markets to offer a greater array of secure, voluntary private savings and insurance instruments. In the relatively developed, middle-income countries of Latin America where households are already engaging in strategic portfolio behavior and where sophisticated savings and insurance products are increasingly available at competitive prices, mandating a par- ticular form of private savings instrument can present a new set of distor- tions. A government-mandated savings instrument offered by a small number of dedicated private providers may produce adverse effects on the supply of private and potentially more competitive alternatives. Notes 1. The distinction among the appropriate instruments to mitigate investment versus longevity risk is discussed in chapter 10. 2. The importance of indirect government intervention (such as the regulation of private providers of contractual savings and the provision of inflation-indexed securities) is also addressed in chapter 10. 3. These traditional strategies can take the form of larger families or preferences for male offspring, especially in agricultural economies and in labor markets whose wage structure discriminates against women (Hoddinott 1992). Furthermore, there is ample evidence that households still rely heavily on reciprocal relationships within THE PREFERENCES THAT INDIVIDUALS REVEAL 193 the extended family; remittances arising from rural-to-urban and international mi- gration of household members (Hoddinott 1992); strategic marriage arrangements; intrahousehold arrangements; establishment of a portfolio of assets with uncorre- lated risks; purchase of livestock or jewelry; forward sale of agricultural crops (Al- derman and Paxson 1992); and community-based credit systems. 4. Of course this argument rests on the strong assumption that governments are less likely to default on public pension promises. A fair counterargument is that most purely public PAYG systems in Latin America were bankrupt prior to reforms, and the likelihood of government default on its public pension promises made work- ers' investment in the pre-reform systems just as risky if not more so. 5. As described in Packard, Shinkai, and Fuentes (2002), specification tests strongly (at the 1 percent level) rejected pooling the data from the different coun- tries into a single regression. 6. Several recent publications have focused on whether individuals are queu- ing for formal employment in Latin America (see Maloney 1998a, 1998b, 1999, 2000, 2003). Contrary to a large body of literature on labor markets in develop- ing countries, these studies found little evidence that self-employment is the resid- ual sector. Data on transitions in and out of the labor market and across sectors in Argentina, Brazil, Chile, and Mexico show that movement into self-employ- ment is more consistent with an entrepreneurial "pull" than the popular notion that workers are "pushed" out of formal jobs into small enterprise. However, informal wage employment often does exhibit many of the characteristics of a residual employment safety net. Individuals in this branch of the informal sector are often indistinguishable in their age and education from the unemployed pop- ulation. In Chile informal employees are more likely to have a greater number of dependents as parents and heads of household than those still searching for a job, and thus are more likely to take up informal employment out of greater income necessity. This raises the concern that informal employers may be unwilling to in- cur the costs of "formalizing" their workers by providing access to the national retirement security system. 7. Ironically, although a worker's density of contributions is relatively more important in assessing whether he or she is covered in a defined contribution sys- tem than under a purely PAYG regimen, the private and decentralized structure of the reformed system in Chile has made data on contribution history unavailable to government researchers interested in exploring contribution patterns. The private fund management industry resisted earlier efforts for the government to gain access to these data, even successfully arguing their case in Chile's courts. The government has found a second-best avenue around this obstacle by surveying a random sam- ple of AFP affiliates. The nationally representative data on affiliates will soon be available to researchers. 8. Cox-Edwards (2000) and James, Cox-Edwards, and Wong (2003) used cross-section survey data to estimate longitudinal patterns of contributory behav- ior and wages. Because information on years of contributions was previously unavailable, the researchers were forced to create synthetic cohorts to estimate years of contributions. They found that men typically accumulate 40 years' worth of contributions from the age of 16 to 65. Women tend to have more interruptions, especially women with lower levels of education. 9. Contribution density is constructed by first calculating respondents' history of contributions in months and dividing that number by their number of months in the labor force, using the Mincer (1974) formula for labor market experience: (age minus years of education minus 5). For further details see Packard (2002). 10. Eligibility for the minimum pension guarantee in Chile is not only deter- mined by a minimum contribution requirement, but also by an income test. How- ever, poor elderly people who have not contributed for at least 20 years can only 194 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA receive the noncontributory social assistance pension, roughly equal to 50 percent of the contributory minimum pension guarantee. 11. The contribution densities shown are a cross-section. Affiliates of different ages are grouped together by contribution density. However, it is likely that work- ers may increase their contribution density as they age, and even as they approach retirement age. However, plotting contribution densities by age does not reveal a clearly increasing pattern. Regression analysis shows that affiliates may even per- ceive diminishing marginal returns from contributions, leading many to slow their contributions well before retirement age. 12. There is also no difference in time preferences between employees and self- employed workers. Furthermore, self-employed contributors to the pension system show lower rates of time preference, as traditional life-cycle consumption and sav- ings theory would suggest, thereby validating the techniques used to gather the data. 13. However, the security of investments in housing depends greatly on secur- ing property rights; and the ease with which wealth held in physical assets can be used to finance consumption will depend on the functioning of the market for land and real estate. Neither can be taken for granted in a developing country like Peru. Furthermore, the prevalent mode of investing in children may become an unreliable source of income security in old age because of greater migration, urbanization, and the resulting dispersion of family groups. As an additional note of caution, although respondents are behaving in a manner consistent with their current pref- erences, this is no guarantee that these preferences will remain stable with age or over time. 14. However, the results presented cannot be used as evidence that a transition to private individual retirement accounts is the only way to improve incentives and achieve greater rates of participation. A similar improvement in incentives may arise from aligning contributions and benefits within a PAYG system--as with the establishment of notional defined contribution (NDC) retirement accounts in sev- eral Eastern European countries. In Latin America, although not an NDC reform strictly speaking, only Brazil has chosen to align benefits and contributions while maintaining the PAYG financing structure of its retirement security system. Not enough time has passed since the Brazilian reform in 1999 for a conclusive analy- sis of its impact on incentives to be made. The analyses in background papers writ- ten for this book indicate that workers respond to improvements in incentives to contribute to formal retirement security systems. To the extent that public social security institutions in developing countries lack credibility, and that privately managed, individually owned retirement savings are perceived as wholly owned, the impact on participation brought about by a transition to private individual re- tirement accounts may be greater than adjustments in the parameters of PAYG systems. Time will tell as more evidence becomes available. 15. Simulation analysis conducted in the past (Edwards and Cox-Edwards 2000a; Cox-Edwards 2000) indicates that the minimum pension guarantee in Chile may be set too high relative to average income and may foster moral hazard among workers with lower lifetime earnings. The analysis in Packard (2002), however, shows that meeting the minimum contribution requirement for the guaranteed benefit has no significant effect on the contributory behavior of workers in the fourth income decile and below. On the other hand, workers in the fifth income decile and above who become eligible for the minimum guaranteed annuity are less likely to continue contributing. The likelihood of additional contributions is low- ered even more as the market value of their homes rises. 16. There is an interesting alternative interpretation of workers' observed be- havior. A saver's risk does not stem solely from the risk in the investment oppor- tunities. The risk of unforeseen cash needs is also important. Thus a risk-tolerant THE PREFERENCES THAT INDIVIDUALS REVEAL 195 worker accepts a larger amount of liquidity risk. Middle-income workers may cease contributing to the second pillar after completing 20 years because the size of the second-pillar replacement rate they have accumulated by then (say, 30 per- cent) is adequate for their needs (and is optimal given the illiquidity of second- pillar savings). Middle-income workers may value the minimum pension because it is a subsidy they can access by choosing programmed withdrawal and living longer (the programmed withdrawal pension falls with survival, but the state sub- sidizes a floor). High-income workers have a tax motive to stop contributing for 20 years: second-pillar pensions are subject to the income tax, whereas profits from the holding and sale of houses are exempt--and profits in other investments can be hidden by creative accounting. We are grateful to Salvador Valdés-Prieto for pointing this out. As in the first interpretation, however, the weight of this al- ternative interpretation rests on a greater assumed rationality and informed choice on the part of affiliates rather than on the default assumption of irrationality so prevalent in the pension policy literature, and would justify only a minimal man- date to save. III Prospective: The Future of Social Security in Latin America 9 Preventing Poverty in Old Age: Improving the Pooling Component HISTORICALLY, GOVERNMENT-ORGANIZED (mandated and/or administered) retirement security systems have been designed with two complementary functions in mind: maintaining levels of consumption in retirement by replacing a portion of individuals' incomes when they are no longer able to work, and preventing widespread poverty among elderly people by fixing an income floor below which covered workers will not fall. In the past two decades social security reforms in Latin America and elsewhere have focused mainly on restructuring the income replacement function of pension systems, whereas the poverty prevention function (once ex- plicitly identified and separated into a first or even a zero pillar) has not received the attention it deserves. In fact, this component should be the main attraction of a social security system, not a sideshow. Building on the evidence and reasoning in earlier chapters, this chapter shows why the lack of attention to this core component of government policy is a serious mistake. Government's Essential Role: Preventing Poverty Among Elderly People In earlier chapters we saw how changing economic fundamentals imply the growing importance of saving over pooling for income replacement in retirement where longevity is increasing (although some degree of pooling through the purchase of annuities is appropriate to cover longevity risk among members of roughly the same age cohort).1 Figure 9.1 shows how longevity is likely to evolve in selected Latin American countries, providing an empirical basis for this shift in policy 199 200 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 9.1 Rising Life Expectancy Increases the Share of Elderly People in the Population and Upsets the Balance of Pure Pooling Pension Systems 17.0 16.0 65 at 15.0 men of 14.0 13.0 expectancy Life12.0 11.0 10.0 2000 2010 2020 2030 2040 2050 2060 2070 Argentina Bolivia Chile Colombia Ecuador Mexico Peru Uruguay Venezuela Source: PROST projections based on simulation data in Bos, Massiah, and Bulatao (1994). with the passage of time: the average 65-year-old man is likely to live al- most 20 percent longer in 2050 than he did in 2000. With the loss of earn- ings ability while living becoming more frequent, countries that still rely on traditional "defined benefit" systems will experience strong fiscal pres- sures to reform. In contrast, as countries develop, poverty among the elderly portion of the population should become increasingly rare relative to poverty among other age groups. Table 9.1 shows head-count poverty rates for selected Latin American countries calculated using income reported in household surveys. The incidence of poverty among elderly people by this measure can actually be significantly greater than among the population as a whole. Indeed, in 11 out of the 19 countries shown in table 9.1 this is the case, seemingly contradicting our assertion that poverty among old men and women should be relatively rare. PREVENTING POVERTY IN OLD AGE 201 The numbers, however, often fail to capture wealth.2 When the value of accumulated assets is taken into account (see figure 9.2 using data from the PRIESO survey in Lima, Peru), elderly people are somewhat better off. In the context of economic development, all else being equal, average life- time incomes should rise, the opportunities to save and accumulate assets should increase, and poverty among the elderly population (measured in wealth--accumulated assets) should become even more rare. In a review of poverty among elderly men and women in 44 countries--largely wealthier, industrial countries--Whitehouse (2000) found that although the incomes of elderly people are 80 percent of the incomes of the popu- lation as a whole, these people are not typically poor. Old people are ei- ther represented proportionately or underrepresented among the poor population. For this reason minimum PAYG benefits need not be justified on the ba- sis of solidarity. As a relatively rare loss, risk pooling is the most appropri- ate and efficient insurance mechanism to mitigate poverty in old age. To the extent that poverty among old people is increasingly rare, and effective tar- geting instruments are in place to accurately measure household wealth, a social insurance mechanism to keep elderly men and women out of poverty should become more affordable. Obviously, in making this argument the legal and financial institutions that protect property rights and allow households to convert illiquid assets into income for consumption in old age are critically important. The existence and efficiency of such institu- tions cannot be taken for granted in Latin America or elsewhere. Over time, therefore, the rationale to pool the risk of old-age poverty has grown stronger, not weaker. Comprehensive insurance strategies im- ply a growing importance of the pooling pillar because it is increasingly well matched in terms of design to the risk of old-age poverty. In most cases, however, its size should shrink relative to the savings pillar. It is curious that the poverty prevention function of pension systems should have been relatively ignored until quite recently and that the in- come replacement function should have received so much attention from policymakers and the pension specialists who advise them. After all, the public policy objectives of mandating participation in retirement security systems are to counteract individuals' improvidence and myopia during their working years and to prevent elderly poor people from becoming an economic burden on society. Although it is determined by specific social values that can vary widely from country to country, the minimum level of income required in retirement to fulfill these two policy objectives is rel- atively modest--certainly less than the benefits promised by pre-reform, single-pillar PAYG systems in Latin America, and often less than the level of income replacement targeted by the architects of structural reforms in the region.3 Indeed, the poverty prevention features of a retirement income security system should be at the center of policy discussions rather than at the Table 9.1 Poverty among Elderly People Is as Frequent as among Other Age Groups When Measured by 202 Current Income (Percent) Per capita income Per capita income 50% Equivalent income 50% under $2a of medianb of medianc KEEPING All 60 65 All 60 65 All 60 65 Argentina 16.64 5.29 4.49 20.71 6.40 5.22 18.60 9.48 8.48 THE Bolivia 46.98 47.96 50.17 28.36 30.16 31.35 27.86 36.59 38.38 Brazil 22.21 6.32 5.38 25.58 7.45 6.28 23.46 10.33 9.44 PROMISE Chile 8.88 4.41 3.78 13.11 6.58 5.74 11.81 7.89 7.34 Colombia 30.37 29.91 31.19 26.57 26.43 27.61 25.75 29.20 30.79 Costa Rica 11.75 16.83 19.06 18.12 22.22 24.19 17.30 30.38 33.63 OF Dominican 13.84 19.81 21.75 25.18 34.07 36.39 23.76 38.09 40.89 Republic SOCIAL Ecuador 40.92 47.27 49.38 26.52 33.53 34.69 26.02 38.39 40.03 El Salvador 36.86 33.71 33.63 30.46 28.48 28.37 30.06 30.35 37.74 SECURITY Guatemala 36.60 -- 44.70 36.60 -- 44.70 34.10 -- 45.90 Honduras 36.58 28.89 30.28 27.45 19.91 20.87 26.81 24.27 25.67 Jamaica 17.50 17.50 18.19 26.21 24.57 26.44 25.02 29.71 32.29 Mexico 27.22 30.23 30.95 22.86 27.44 28.04 21.89 30.50 31.27 IN Nicaragua 38.26 31.83 32.55 20.34 15.81 16.20 19.70 18.25 18.95 LATIN Panama 21.35 18.14 17.97 22.16 18.80 18.66 20.64 21.71 21.69 Paraguay 30.21 23.32 23.22 25.91 19.40 18.60 24.46 24.35 23.92 Peru 31.80 21.90 21.56 24.70 15.78 15.24 24.17 18.32 18.69 AMERICA Uruguay 1.41 0.14 0.15 21.95 6.52 5.43 17.84 8.67 7.45 Venezuela 29.52 24.98 25.28 19.95 17.36 17.49 19.69 20.85 21.18 Notes: Poverty rates for older age groups (over age 60 and 65) in selected Latin American countries; -- not available. PREVENTING a Per capita income equal to or less than US$2 per day. b Per capita income equal to or less than 50 percent of median income. c Income equal to or less than 50 percent of median income using OECD equivalence scale. OECD scale is equal to 0.5 0.5 number of adults 0.3 number of children (age 16 or less). Source: Bourguignon, Cicowiez, Dethier, Gasparini, and Pestieau (2004). POVERTY IN OLD AGE 203 204 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 9.2 Accumulated Wealth Increases with Age and Is Greatest among Old People 30,000 14,000 Asset value Earned or pension income 25,000 12,000 Earned 10,000 20,000 income dollars 8,000 U.S. in 15,000 in U.S. 6,000 value dollars 10,000 4,000 Asset 5,000 2,000 0 0 14­25 26­39 40­64 65+ Age in years Source: Barr and Packard (2003) using PRIESO (2002). margin. When a pension system that will prevent poverty among elderly men and women at a fiscally sustainable cost is in place, government has largely fulfilled its role with respect to retirement income security and should then concentrate on ensuring the efficient and secure supply of pri- vate savings and insurance options. In this chapter we briefly set the earnings replacement function of pen- sion systems aside and place the poverty prevention function at center stage. We examine the emerging distinction between contributory "pillar one" and noncontributory "pillar zero" public pension benefits, using the theoretical framework employed in chapter 6. Next we describe three dif- ferent options for structuring the poverty prevention pillar, drawing on in- ternational and regional experience. Finally, we present the results of sim- ulations to show the cost of each public pooling alternative in the Latin American countries that have undertaken structural reforms to motivate a discussion of how best to structure public poverty prevention systems for elderly people. Readers should note that we do not advocate any particular option over the rest because the optimal poverty prevention institution or set of PREVENTING POVERTY IN OLD AGE 205 institutions will vary widely by country, level of development, and adminis- trative capacity. We do, however, place poverty prevention (i.e., covering the risk of poverty in old age) at the top of the list of objectives of pension pol- icy, and we present the set of issues that policymakers should keep in mind when determining what sort of poverty prevention pillar to put in place. Pillar Zero vs. Pillar One: Does the Distinction Matter? Since the publication of Averting the Old Age Crisis (World Bank 1994), in which the basic architecture of the multipillar system of public and pri- vate pension provision was described, a curious distinction has emerged between two types of pensions still directly administered by government: first-pillar pensions paid to individuals with a history of contributions to an earnings-related retirement security regimen, and zero-pillar pensions offered universally or targeted at the elderly poor population regardless of whether they have contributed to an earnings-related pension regimen. This distinction is probably more helpful to pension specialists (who are confronted with public pension systems that, even after structural reforms, can exist alongside other transfers to elderly people) than it is to policy- makers or to the individuals and households they represent. Conceptually we find the distinction unnecessary. According to the analytical framework borrowed from the economics of insurance (presented in chapter 6), when the policy objective is to pre- vent poverty among elderly men and women, the distinction between a first pillar that conditions benefits on contributions and a zero pillar that pays pensions regardless of contribution history is largely political. Both pillars pool the risk of poverty in old age; both pillars are directly ad- ministered by government; both pillars are typically financed on a PAYG basis; and both pillars receive "contributions" either from payroll taxes or from other levies with a broader base such as taxes on income or on consumption. By definition, a pooling or insurance instrument will always have fewer "losers"--those who receive a payoff from the pool if they suffer the bad state that the instrument seeks to cover--than the number who contribute to the pool. After all, there has to be a sufficient number of "winners" to compensate the "losers" or the instrument will not be able to pool risks effectively (Barr 2001). The payment to contributors who suffer the bad state may be actuarially fair, but not all contributors are guaranteed payment. When the logic of prescribing public, defined benefit, PAYG pooling to cover the relatively rare risk of poverty in old age is accepted, and the level of pooled benefits is set with care not to upset individual and household incentives to save privately, to insist on separating a zero pillar from the 206 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA first appears politically expedient at best. At worst, perpetuating a dis- tinction between the two pillars can lead to perverse outcomes and even increase the vulnerability of the poorest segment of society. On the one hand, public pooling PAYG systems that deny minimum pensions to individuals who lack a history of explicit contributions but that pay benefits guaranteed by government transfers, can redistribute in- come from all current and future taxpayers to those who have accumu- lated minimum pension rights. Even where the contribution and benefit parameters of pillar one are set to be "self-financing," who pays for short- falls between benefits and contributions during economic downturns, or for indexation to protect benefits during bouts of inflation? All current and future taxpayers "contribute" to maintain the number and the value of benefits paid to a relatively smaller group of covered workers. On the other hand, separate, seemingly noncontributory transfer arrangements to the elderly poor population are perceived as charity rather than just another instrument with which households can manage risks to income, and are often only reluctantly considered in budget allo- cations. Small budget allocations to "social assistance" pensions sepa- rated from the first pillar typically count on the support of small, rela- tively weak political constituencies, and historically have been vulnerable to budget cuts. Options for Preventing Poverty among Elderly People There are at least three basic alternatives for structuring the public pooled component of a government-organized retirement security system with a poverty prevention objective (Willmore 2001a). These are (1) a minimum pension guarantee or benefit top-up to workers who have contributed for a specified number of years to a retirement security regimen; (2) a benefit targeted at the elderly poor population; and (3) a universal flat pension, sometimes called a "demogrant," paid to all men and women over some threshold age, regardless of their means. In several countries these alter- native poverty prevention pension structures overlap. Each option has its particular merit, but all impose a cost. A minimum pension top-up will cover workers with low lifetime earnings who con- tribute to the retirement security system, and may provide some incentives for participation in the earnings-related pillar, but it leaves workers with- out a history of payroll contributions uncovered. A targeted pension will provide benefits to only those elderly people whose income or accumulated wealth lies below some specified level, but accurate targeting may lie be- yond the capacity of many governments and can introduce poverty traps and opportunities for corruption. A universal pension covers all individuals of a certain age regardless of their income, accumulated wealth, or contri- bution history, but can entail considerable and even prohibitory fiscal costs. PREVENTING POVERTY IN OLD AGE 207 Many countries have at least one of the three arrangements, and most countries in Latin America offer both the contributory minimum pension guarantee and an often poorly targeted benefit for the elderly poor popu- lace. Chile's minimum pension guarantee ensures that affiliates with at least 20 years of contribution history to the pension system retire with a minimum annuity amount, which is initially financed out of the accumu- lated balance in the affiliate's individual account and then by the govern- ment directly when these savings are exhausted. This model has been adopted in Colombia, Mexico, and El Salvador. Mexico offers to all affil- iates an additional account subsidy that has an important poverty preven- tion role for poorer affiliates (see box 9.1). Few countries offer a univer- sal flat pension. We are not aware of any Latin American country other than Bolivia that offers such a plan (see box 9.2). Judging from the conceptual discussion in the previous section, in a country where all individuals contribute to the earnings-related pension system, a contributory minimum guarantee structured as a top-up is a satisfactory public pooling arrangement: it encourages workers to save privately and guarantees a minimum level of retirement income at a mini- mum cost to taxpayers. As we already pointed out, however, in countries where most workers will not have a sufficiently long history of contribu- tions to the earnings-related pension system, a top-up conditioned on a long history of participation can not only exclude large segments of the population, but also can lead to perverse transfers. Readers are reminded of the large share of affiliates in the reformed pension systems in Chile and in Peru that may not have a sufficient contribution density to qualify for the minimum pension guarantee (i.e., if this guarantee is eventually imple- mented for all affiliates of the private second pillar in Peru). Incentives can also be distorted if low-income workers are offered their choice of invest- ment (as in Chile) because the pension guarantee eliminates downside risk, thereby turning high-risk investments into a one-sided bet. Targeting public pensions at the elderly poor population is the public pooling arrangement closest to the risk pooling ideal for middle-income countries in which many workers fail to contribute regularly to an earn- ings-related pension pillar. This is especially true if the targeted benefit is financed with a broad-based tax, such as the value-added tax. Means testing to target the benefit efficiently, however, comes with complications and costs. Means tests increase administrative costs and provide opportunities for corrupt behavior on the part of public officials (World Bank 1994; Willmore 2001a). Furthermore, just as overly gener- ous social assistance benefits can lead to moral hazard, so means tests can discourage private saving and wealth accumulation for retirement (Hub- bard, Skinner, and Zeldes 1993, 1994) as well as continued work in old age. Finally, as mentioned in the previous section, means-tested benefits are often regarded as charity, which reduces their political appeal, makes the benefits vulnerable to budget cuts--especially in economic downturns 208 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 9.1 The Cuota Social: Preventing Poverty among Elderly Men and Women in Mexico To strengthen the efficacy of Mexico's new multipillar pension system in preventing poverty among elderly men and women, the government of Mexico introduced a flat contribution subsidy along with private indi- vidual accounts in 1997. The government makes a daily payment to the individual retirement accounts of all workers affiliated with the new pri- vate defined contribution pillar whose contributions are up to date. This contribution, called the Cuota Social, is commission-free and does not vary with workers' income. When the new AFORE pillar came into effect in July 1997, the Cuota Social was set at 5.5 percent of the minimum wage and now equals 6.05 percent, reflecting adjustments for inflation. Because all contributing affiliates of the reformed pension sys- tem in Mexico receive the same amount of Cuota Social contributions in their individual retirement accounts, the share of total pension contribu- tions represented by the Cuota Social is inversely proportional to work- ers' incomes. That is, the Cuota Social bolsters the value of low-income workers' pension accounts more than the value of high-income workers' accounts. In fact, Cuota Social contributions represent more than one- quarter of the value of retirement contributions for the 68.5 percent of the Mexican workforce earning three times the minimum wage or less; it ac- counts for almost 55 percent of the retirement contribution for a worker earning the minimum wage. The Cuota Social introduces a poverty prevention mechanism directly into the defined contribution pillar created by the Mexican pension re- form of 1997. For workers earning up to three times the minimum wage the Cuota Social is greater than commissions charged to their individual contributions taken from wages, thereby allowing their future pension benefits to be larger than their individual contributions. This could serve as an incentive for low-income workers to join or remain in the formal sector by contributing to the defined contribution system. It is still too early to adequately estimate the effects of the Cuota Social on workers' incentives to join the AFORE system and to keep their contributions up to date. The contribution subsidy, however, is not the only poverty prevention instrument in Mexico's multipillar system. AFORE affiliates with at least 25 years of contributions who have not accumulated savings sufficient to finance a determined minimum annuity qualify for a minimum pension guarantee top-up similar to that in Chile's AFP system. This poverty pre- vention instrument is more targeted at the likely elderly poor population than is the Cuota Social. The Cuota Social's fiscal burden on the Mexican government will be substantial in the first decade of reform. At 0.33 percent of GDP at the outset of reform, it is the largest cost item among government PREVENTING POVERTY IN OLD AGE 209 Box 9.1 (continued) contributions to the reformed pension system. However, as GDP and real wages grow the relative burden of the Cuota Social will decline, even al- lowing for increases in pension coverage. Grandolini and Cerda (1998) projected that Cuota Social liabilities will decrease to 0.2 percent of GDP by 2025 and increase again, to 0.5 percent by 2067. Furthermore, Azuara (2003) projected potential fiscal savings by increasing the Cuota Social as a percentage of the minimum wage and investing the amounts over time, thereby reducing the government's liability at the time of retirement to finance the minimum pension guaranteed by the reform. Source: Contributed by Todd Pugatch, based on Grandolini and Cerda (1998) and Azuara (2003). (Snyder and Yackovlev 2000)--and may discourage eligible applicants dissuaded by social stigma (Barr 1992; Willmore 2001b). Despite this, some countries have improved their targeting efficiency at modest cost, notably Mexico and Chile. A universal flat pension does not strictly comply with the conceptual ideal of a public risk pooling mechanism to ensure against poverty in old age because all individuals above a specified age would receive some ben- efit, not only those suffering poverty. However, Willmore (2001a, 2001b) claimed that of the three public options for pooling against the risk of poverty in old age, universal pension benefits have numerous advantages over both systems that condition coverage on a minimum period of con- tributions and those that target benefits to poor people. He argued that be- cause universal pension benefits provide coverage to all individuals of pen- sionable age they are the simplest public poverty prevention mechanism to administer, with the lowest transaction costs. Willmore believed that uni- versal flat pensions prevent governments from creating the disincentives to work and save that are often inherent in means testing. In a review of public poverty prevention pensions around the world, however, Willmore noted that most governments regard this type of pro- gram as a luxury that will be difficult to sustain. He demonstrated alge- braically that where per capita income is growing, the cost of providing a universal benefit to a growing share of elderly people need not imply an onerous levy on taxpayers and that this can even be true when income growth is kept constant. Figure 9.3 compares the generosity and cost to taxpayers of public poverty prevention pensions in a first set of countries that provide universal benefits, with a second set of countries that means- test to target benefits at the poorest individuals and households. Willmore claimed that government can easily regain a large portion of the transfers to elderly men and women by taxing universal pensions as 210 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 9.2 BONOSOL: Bolivia's Universal Pension Program A unique feature of Bolivia's pension reform was the creation of the BONOSOL program, which uses income from privatization of state en- terprises to fund an old-age social assistance pension. The Sanchez de Lozada government (1993­97) partially privatized Bolivia's six largest public enterprises, selling 50 percent of the firms to foreign companies. Noting the poor results of recent public sector projects (a possible alter- native use of the funds generated), the absence of a safety net to prevent old-age poverty, and the Bolivian people's status as the ostensible owners of the state enterprises that had been capitalized, the government decided to distribute the proceeds of privatization to all Bolivians 65 years old and older. Political considerations were also important in establishing the program: distribution of dividends from capitalization would galvanize support for privatization and for pension reform. BONOSOL pays an annuity to all Bolivians 65 or older. The initial an- nuity was scheduled at US$248 per year for the first five years of the pro- gram, with adjustments to follow every three years thereafter to reflect changes in portfolio income and life expectancy. Citizens who were 21 or older in 1995 are eligible to receive the BONOSOL annuity when they reach 65 on the grounds that this group paid for the state enterprises that were sold to finance the program. No similar benefit has been planned for subsequent generations. The portfolio through which the program is fi- nanced was US$1.65 billion at its inception, representing 22 percent of Bolivia's GDP. Bolivia's two AFPs manage this portfolio. Although the BONOSOL annuity is not means-tested, the program serves as the poverty prevention pillar in Bolivia's multipillar pension re- form. Because all Bolivians 65 or older are eligible for the benefit, its cov- erage is much greater than for those receiving pension income under the old PAYG system or the new defined contribution system. At just 27 per- cent of average per capita income and 11 percent of average earnings in 1997, BONOSOL payments are intended to provide a broadly shared subsistence income in old age, not a replacement income to a small seg- ment of the workforce. BONOSOL replaces 85 percent of the income of extremely poor people and 50 percent of the income of the other poor people (1997 figures). Early results indicated that the program would be a major success: of the more than 300,000 Bolivians eligible to receive the benefit at its inception, 63 percent received the annuity within the first two months of the program, with only 14 cases of attempted fraud reported. The BONOSOL program is also less onerous to the government than most other poverty prevention pillars. The program enabled the Bolivian government to establish a poverty prevention pillar without minimum pensions or minimum return guarantees, thereby reducing its contingent liabilities. And because the program is financed by capitalization income rather than payroll taxes, it does not distort labor markets. PREVENTING POVERTY IN OLD AGE 211 Box 9.2 (continued) The political risk inherent in the program was apparent when, after the new government came to power in 1997, the BONOSOL program was renamed BOLIVIDA and payments were reduced to US$60 per year. The reduction was made based on a Ministry of Finance study that con- cluded that the higher BONOSOL benefits would be exhausted in 30 years, rather than the 70 years for which the program was envisioned. Nevertheless, the new Sanchez de Lozada government that came to power in 2002 restored the BONOSOL program at its original benefit levels, ful- filling a campaign promise but disregarding financial simulations that indicated the program's lack of sustainability. Current and future benefi- ciaries of BONOSOL remain subject to portfolio and political risk, but the latter can be minimized by establishing a sense of ownership over BONOSOL among its beneficiaries to increase the political cost of divert- ing its assets elsewhere. Source: Contributed by Todd Pugatch, based on Von Gersdorff (1997) and Escobar (2003). any other form of income, and he cited New Zealand's universal flat pen- sion as an ideal poverty prevention system. Assuming a large part of uni- versal benefits paid to the elderly population could be clawed back through taxation channels, the administrative ease of a universal flat pen- sion might provide a public pooling model that effectively and efficiently covers the risk of poverty in old age. Contributory Minimum Pension Guarantees Minimum pension guarantees are used in most Latin American countries to set a minimum level of benefits in mandatory pension systems. The min- imum benefit is usually set at a level close to the minimum wage, but there are significant differences between countries. For example, in Colombia the minimum pension is fixed at the same level as the minimum wage. In Chile, on the other hand, the minimum pension is fixed discretionally by the government and is currently equivalent to approximately two-thirds of the minimum wage. All countries apply minimum contribution require- ments that range from 20 years in Chile to 30 years in Argentina, the Do- minican Republic, and El Salvador. These guarantees can distort incentives and lead low-income workers to contribute only for the period needed to qualify for the guarantee. In chapter 8, however, we showed that the contribution behavior of poorer workers in Chile and Peru did not change significantly after they qualified 212 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 9.3 Relative Generosity and Cost of Alternative Public Poverty Pension Arrangements in Selected Countries 45 4.5 40 4.0 capita 35 3.5 Percentage per 30 3.0 GDP 25 2.5 of 20 2.0 of GDP 15 1.5 10 1.0 Percentage 5 0.5 0 0.0 Africa States India ZealandMauritius NamibiaBotswana Australia New South United Poverty prevention pension as percentage of GDP per capita (left axis) Total annual transfer to the elderly population as percentage of GDP (right axis) Notes: Countries in the first group (Botswana, Mauritius, Namibia, and New Zealand) provide a minimum universal benefit. Those in the second group (Australia, India, South Africa, and the United States) target minimum benefits at elderly poor people. Source: Willmore (2001a). for the minimum pension guarantee (whereas the contribution behavior of workers in the fifth income decile and higher did change). A more worrisome issue is that minimum pension guarantees can dis- tort incentives for risk taking in savings-based pension systems. A mini- mum pension guarantee is a pooling instrument to prevent poverty, but access to this instrument in most Latin American countries4 is condi- tioned on a minimum contribution to the savings system with a con- sumption-smoothing objective. There is an inherent tension between the need for risk pooling to counter poverty and the need for individual choice to achieve efficient consumption smoothing that can only be resolved if different instruments (pillars) are assigned these separate func- tions. Minimum pension guarantees work well in defined benefit and no- tional defined contribution systems, where there is no individual choice, PREVENTING POVERTY IN OLD AGE 213 but are not adequate for the savings-based systems of Latin America, where it is only a matter of time before individual choice of investment portfolio is more widely introduced. To date, only Chile permits investment choice in its mandatory pillar (the one that guarantees protection to workers). In June 2002 a new mul- tifund structure was introduced that allows all young workers to choose among funds that allocate different percentages of their assets to equities. Poorer workers, knowing that they are covered by the minimum pension guarantee, have a strong incentive to choose the portfolio with the high- risk, high-return profile. Other countries are likely to follow the Chilean example. These countries face a significant challenge if they continue to exercise their poverty prevention goals through minimum pension guar- antees conditioned on participation in the mandatory funded system while trying to improve the design of the funded component's consumption- smoothing properties by permitting workers to invest a higher proportion of their mandatory savings in equities and foreign securities than is cur- rently permitted. Calls for more flexibility in investment and greater individual choice are already being heard in other Latin American countries. In Mexico the liberalization of individual choice and investment portfolios is particularly worrisome for the government, because workers who contributed to the old system are covered by a guarantee equal to the benefit under the old PAYG plan. If they were offered the choice these workers would quite ra- tionally opt for the riskiest portfolios, knowing that they will be bailed out by the government if the funds perform badly. Only those countries in which there is no minimum pension guarantee in the funded component, such as Argentina, Costa Rica, and Uruguay, can safely relax restrictions on individual choice and investment because workers there still benefit from a defined benefit pension that provides a basic income. No country in the rest of the world has completely done away with a ba- sic defined benefit pillar. This feature of several of the Latin American pen- sion systems is practically unique by international standards. The only country that comes somewhat close to facing a similar moral hazard prob- lem is the outsourced pension scheme in the United Kingdom. In the United Kingdom there is a minimum pension guarantee, but this is set at 20 per- cent of national average earnings, a level only slightly higher than that of the mandatory basic pillar that offers a flat-rate benefit equivalent to 16 percent of average earnings. But given current requirements for minimum contributions into outsourced plans, the risk of moral hazard is minimal. The moral hazard problem of individual choice has not emerged in any Latin American country other than Chile. This problem has been ad- dressed by restricting individual choice and limiting investment to interest- bearing assets. By the same token, however, these restrictions damage the consumption-smoothing properties of the funded component, as argued in chapter 7. To the extent that the savings system is to become the mainstay 214 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA of retirement income in countries such as Chile, Colombia, El Salvador, Mexico, and Peru, it should be complemented with a poverty prevention, risk pooling pillar that has separate financing. Policymakers in these coun- tries should consider transferring the part of contributions sufficient to combat old-age indigence into a separate pooling mechanism offering flat- rate,5 basic benefits, as is done in Argentina. The savings component could then be freed of any state guarantees and evolve into a flexible consump- tion-smoothing system offering workers meaningful investment choices. Noncontributory Poverty Prevention Pensions in Latin America Prior to structural reforms retirement security systems in Latin America often combined the poverty prevention and earnings replacement func- tions into a single, defined benefit PAYG system. When a single policy in- strument is designed to meet both objectives, it is difficult to take stock of its success in preventing poverty--the subject of this chapter. Some stock- taking of Latin America's experience with PAYG systems was briefly pre- sented in chapter 3 where the substantial fiscal impact of these systems and their negative effects on equity were also discussed. In this section we briefly review experience in the region with pension systems explicitly designed (in most cases) to prevent poverty--the so- called noncontributory pension programs.6 A telling statistic is that three of the countries reviewed in this book (El Salvador, Mexico, and Peru) do not have noncontributory pension systems. In the countries where they exist, these programs developed alongside the contributory, payroll tax­financed, single-pillar PAYG systems to become a dominant instru- ment among the battery of targeted social assistance benefits. We draw the comparative statistics on noncontributory pensions from an extensive (and highly recommended) review by the International Labour Organiza- tion (Bertranou, Solorio, and Van Ginnekin 2002). Noncontributory systems have developed in countries in the region with a longer history of formal social security. Whereas the noncontribu- tory component of the larger social insurance infrastructure in these coun- tries usually developed parallel with the contributory programs, in many cases (Brazil being a prime example) noncontributory programs may still be bound within the contributory programs. Tables 9.2 and 9.3 compare noncontributory pension programs in Argentina, Brazil, Chile, Costa Rica, and Uruguay according to the criteria commonly used to assess so- cial protection programs: fiscal cost, coverage, generosity, and effective- ness at reducing poverty. Table 9.2 presents expenditure on noncontributory pensions in the five countries studied by the ILO, as a percentage of GDP; as a share of total public spending on social security; and as a share of spending on the wider PREVENTING POVERTY IN OLD AGE 215 Table 9.2 Expenditure on Noncontributory Pension Programs Expenditure as Expenditure Percent percentage as percentage financed Expenditure of total of total from as percentage expenditure on social sector general Country of GDP social security spending revenues Argentina 0.2 3.6 1.1 100.0 Brazil 0.3 5.3 2.0 100.0 (Assistance) Brazil 1.0 17.2 6.7 91.6 (Rural) Chile 0.4 5.5 2.3 91.6 Costa Rica 0.3 7.0 1.8 48.3 Uruguay 0.6 5.5 2.6 100.0 Note: Brazil's assistance pensions include the purely noncontributory BPC (benefi- cio de prestação continuada, continuous pension benefits), which replaced the RMV (renda mensal vitalícia, lifetime monthly income) in 1996. Brazil's rural pensions pro- gram is a regimen of preferential contribution parameters for rural sector workers that acts implicitly as a noncontributory pension program because it is poorly enforced. Source: Bertranou, Solorio, and Van Ginnekin (2002). social sectors (including education and health). Public spending on non- contributory pensions is highest in Brazil (by all three measures). Although contributing to Brazil's precarious fiscal situation (along with the egre- giously generous pension regimen for civil servants in that country), this spending has had positive impact on reducing poverty among elderly men and women. As shown in table 9.1, poverty among old people in Brazil is among the lowest in the region. Uruguay is the second-highest spender on noncontributory benefits (as share of GDP), followed in third place by Chile. The final column shows the source of financing for the noncontrib- utory programs. In each of the countries (with the notable exception of Costa Rica) noncontributory pensions are financed almost entirely from general revenues. The first important point about coverage to note in table 9.3 is that benefits paid to the elderly poor populace are just a portion (although frequently the dominant portion) of noncontributory pension programs. These programs typically mirror the benefits paid by payroll tax­ financed systems, namely old-age, survivor, and disability benefits. The share of noncontributory pensions paid for old age is shown in the third column of table 9.3. Brazil's rural pensions program--a special regimen of preferential contribution and eligibility criteria nested within the country's social security regimen for workers in the private sector--pays the greatest share of benefits for old age, followed closely by Costa Rica. Argentina's noncontributory pension system pays the smallest share of Table 9.3 Noncontributory, Poverty Prevention Pensions Cover a Significant Portion of Pension Recipients 216 Beneficiaries as Old-age Number of Percentage percentage of beneficiaries as Number of beneficiaries of benefits all pension percentage of elderly KEEPING Country beneficiaries for old age for old age recipients poor population Argentina 350,660 40,152 11.5 10.1 47.0a Brazil (assistance) 2,022,708 706,345 34.9 11.1 86.3b THE Brazil (rural) 6,024,328 4,012,127 66.6 33.0 -- PROMISE Chile 358,813 165,373 46.1 22.6 36.5 and 78.7c Costa Rica 76,009 46,597 61.3 31.2 44.5d Uruguay 64,053 18,515 28.9 9.0 17.3 and 11.9e OF -- Not available. a Share of indigent population aged 65 and older who receive a public pension based on Encuesta de Desarrollo Social (Social Development Sur- SOCIAL vey 1998), in Arriagada and Hall (2000). b Share of recipients among population 67 and older with per capita household income less than the minimum wage, receiving a public pension (Assistance and Rural), September 1999. SECURITY cShare of population 65 and over in deciles 1 and 2, urban and rural areas, respectively, in 2000. d Share of poor people aged 65 and older in 2000. eShare of poor people aged 50 and older in Montevideo and the interior of Uruguay, respectively, in 1997. Source: Bertranou, Solorio, and Van Ginnekin (2002) and others where noted. IN LATIN AMERICA PREVENTING POVERTY IN OLD AGE 217 benefits for old age. The next point to note in table 9.3 is the importance of noncontributory programs in the overall retirement security infra- structure of the countries examined. Beneficiaries of noncontributory pensions make up more than a third of pension recipients in Brazil and Costa Rica. In Chile, 22.6 percent of pension recipients receive the non- contributory pensiones asistenciales (PASIS, social assistance pension). The share of noncontributory benefit recipients among pensioners in Argentina and Uruguay is much smaller (11.5 percent and 9 percent, respectively). Figure 9.4 shows the relative generosity of the old-age noncontributory pensions in the countries studied by the ILO. Uruguay's noncontributory pension is the most generous as a percentage of the average contributory pension, followed by those of Chile and Argentina. However, the gen- erosity of noncontributory benefits in Chile and Costa Rica is slightly overstated because the minimum rather than the average contributory pension is used for comparison. Figure 9.4 Average Noncontributory Pensions Are between 30 Percent and 60 Percent of Average Contribu- tory Pensions 100 90 80 benefit 70 60 contributory 50 average 40 of 30 20 Percentage 10 0 Uruguay Chile Argentina Costa Rica Brazil Source: Bertranou, Solorio, and Van Ginnekin (2002). 218 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA As discussed in greater detail elsewhere (Bertranou, Solorio, and Van Ginnekin 2002), noncontributory pension programs have had an impor- tant effect on poverty. Case studies show that in 2000 and 2001 noncon- tributory pensions lowered the rate of poverty among elderly men and women by 67 percent in Argentina, 95 percent in Brazil, 69 percent in Chile, and 21 percent in Costa Rica. But many elderly poor people still go without an old-age pension of any kind, as shown in the last column of table 9.2.7 Forecasting the Cost of Public Pooling Alternatives How much would each of the alternative public, poverty prevention pool- ing arrangements described in earlier sections cost? We present at least a preliminary answer to this question in this section, along with the results of simulations using the same income and demographic data and the same macroeconomic assumptions employed in the simulation results presented earlier in this book.8 Rather than try to find a minimum level of public pension benefit that would be both a salient and a socially acceptable lower boundary on in- come across a set of very diverse countries, we decided to use each coun- try's legal minimum wage to reflect the lowest level of income that even populist politicians would find politically acceptable. Setting minimum pension benefits at the minimum wage is a very gen- erous decision with dangerous consequences for incentives to save and fis- cal sustainability. In most of the countries in our sample the minimum wage as a share of average wages is high (table 9.4). The minimum wage Table 9.4 As a Share of Average Wages, Minimum Wages Are Relatively High in Chile, Colombia, and Some Other Latin American Countries MW/10th Country Year MW/mean MW/median percentile wage Argentina (urban) 1998 0.26 0.33 0.67 Bolivia 1997 0.22 0.34 0.80 Brazil (all) 1998 0.24 0.43 1.00 Brazil (urban) 1998 0.22 0.37 1.00 Chile 1996 0.34 0.55 1.09 Colombia (urban) 1998 0.40 0.68 1.00 Honduras 1999 0.62 0.90 2.26 Mexico (urban) 1999 0.34 0.48 0.87 Uruguay (urban) 1998 0.19 0.27 0.64 Note: Ratio of the minimum wage (MW) to comparison wages. Source: Maloney and Nuñez (2001). PREVENTING POVERTY IN OLD AGE 219 is often set above the market-clearing wage by governments acting on po- litical pressure from trade unions. Furthermore, although the minimum wage is presented as the lowest level of income that society is prepared to let workers earn legally, this judgment often assumes that during their working lives individuals have dependents to support and other expenses that retired households no longer have to meet. If we assume that, other than health expenses, elderly people consume less than households headed by individuals of working age, poverty prevention benefits should typically be set as some fraction of the minimum wage. In Chile the minimum pen- sion guarantee oscillated between 80 percent and 90 percent of the mini- mum wage during the 1990s (Cox-Edwards 2000). In Colombia the con- tributory minimum pension is set equal to the minimum wage. In Peru the minimum pension guaranteed in the reformed PAYG system is actually a bit higher than the minimum wage. By selecting the minimum wage as the level of the poverty prevention benefit in these simulations we are not advocating this as the optimal size of the public pooling pillar. To the contrary, given the typical distribution of wages in Latin America, setting the public poverty pension equal to the minimum wage would provide a strong disincentive for workers to save privately. Indeed, in Colombia, where the minimum wage is binding, there is strong emerging evidence that the consequently high minimum pension lowers the share of workers who contribute to the formal pension system and threatens fiscal sustainability. Nor are we advocating one type of pooling structure over another. We are simply addressing a fiscal concern often faced by policymakers when try- ing to determine the appropriate size of their poverty prevention pillar. We provide cost projections of what we believe to be a likely upper-boundary benefit--that which even the most ardent populist politicians might be willing to support as the acceptable level of the minimum pension in their country--to arrive at conservative estimates of the cost of restructuring the pooling pillar in different ways. In this spirit we have also generously indexed the minimum benefit to current wages. Significant cost savings can be had by indexing benefits to inflation or to a combination of prices and wages. To calculate the cost of the targeted benefit we have simply assumed that the current rate of poverty among the elderly population (as shown earlier in this chapter in table 9.1) will remain constant throughout the simulation. In light of the earlier discussion of the likely level of poverty among elderly people, when poverty is measured by income or by accu- mulated wealth, our assumed poverty rate is likely to be very conservative (i.e., to overstate poverty or need among the elderly population) in some of the countries included in this exercise and less so in others. The cost of pay- ing the targeted benefit is simply derived by multiplying this poverty rate by the cost of the universal benefit. This cost will in some cases vary signifi- cantly from the costs of the noncontributory programs reviewed in the previous section largely because we assume that the targeted benefit reaches all the elderly poor people. The simulation results are presented in table 9.5. 220 Table 9.5 Cost of Providing and Guaranteeing a Public Pension Equal to the Minimum Wage (Alternative public pooling arrangements as a percentage of GDP) Uruguay Argentina Mexico Current Current Current transfer Top-up Targeteda Universal transfer Top-up Targeted Universal transfer Top-up Targeted Universal (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) 2001 4.00 0.00 0.14 1.90 2001 2.50 0.80 0.34 4.00 2001 0.50 0.50 0.41 1.30 2010 2.60 1.00 0.14 1.90 2010 2.20 0.80 0.36 4.30 2010 0.60 0.40 0.44 1.40 2020 2.10 0.70 0.14 1.90 2020 2.30 0.80 0.39 4.60 2020 0.70 0.40 0.59 1.90 2030 2.20 0.40 0.16 2.20 2030 2.80 0.90 0.44 5.20 2030 0.70 0.30 0.75 2.40 2040 2.50 0.30 0.19 2.60 2040 3.60 1.30 0.55 6.50 2040 0.70 0.30 1.06 3.40 2050 2.80 0.30 0.22 2.90 2050 4.40 1.60 0.64 7.60 2050 0.60 0.40 1.34 4.30 Bolivia Colombia Chile Current Current Current transfer Top-up Targeted Universal transfer Top-up Targeted Universal transfer Top-up Targeted Universal (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) 2001 3.50 0.00 1.57 4.10 2001 -- 0.20 1.08 3.50 2001 7.20 0.00 0.25 3.40 2010 2.20 0.20 1.42 3.70 2010 -- 0.40 1.14 3.70 2010 4.60 0.40 0.27 3.70 2020 2.10 0.30 1.54 4.00 2020 -- 0.80 1.57 5.10 2020 3.40 0.90 0.33 4.50 2030 2.10 0.20 1.80 4.70 2030 0.70 1.60 2.31 7.50 2030 1.50 1.50 0.48 6.60 2040 1.70 0.30 2.23 5.80 2040 3.40 3.30 2.96 9.60 2040 0.50 1.30 0.60 8.20 2050 0.90 0.60 2.99 7.80 2050 5.40 4.70 3.45 11.20 2050 0.80 1.30 0.64 8.70 El Salvador Peru Current Current transfer Top-up Targeteda Universal transfer Top-up Targeted Universal (%) (%) (%) (%) (%) (%) (%) (%) 2001 1.40 0.10 1.74 4.60 2001 0.70 0.00 0.36 1.90 2010 2.20 0.10 2.00 5.30 2010 0.90 0.20 0.37 2.00 2020 3.20 0.10 2.34 6.20 2020 0.90 0.30 0.45 2.40 2030 2.90 0.00 2.98 7.90 2030 0.90 0.40 0.56 3.00 2040 2.60 0.20 4.23 11.20 2040 0.80 0.50 0.75 4.00 2050 0.50 0.50 5.36 14.20 2050 1.00 0.70 0.93 5.00 a. Poverty rates used to calculate the cost of the targeted benefit are those presented in the last column of table 9.1, as 50 percent of equivalent mean income in each country (Bourguignon et al., 2004). Source: PROST simulations using same assumptions as Zviniene and Packard (2002), documented in the technical annex at the end of this volume. 221 222 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA There are few surprises in the projected costs of structuring the pooled benefit according to the three alternatives discussed earlier. Limiting pub- lic pooled benefits to a top-up paid to those who contribute to the earnings- related pillar is often the cheapest option but also the one that leaves large segments of the population uncovered against the risk of poverty in old age. Targeting benefits at the elderly poor population will be cheaper than the exclusive top-up in countries like Argentina, Chile, and Uruguay, where poverty among the elderly population is low--but more expensive than a top-up in countries like Bolivia, Colombia, and El Salvador, where old-age poverty is more widespread. The surprise arises from the cost of the universal flat benefit. Assuming countries could start from scratch (admittedly, a very strong assumption, but helpful nonetheless), the cost of paying benefits to all individuals over age 65 is in many cases not too different from what governments are cur- rently paying to cover the deficits of their reformed pension systems (much of which is made up of transition costs, and the payment of benefits cov- ering risks other than those of old age). The difference in costs of the uni- versal benefit become apparent, however, later in the simulation horizon when countries have finished paying the transition costs of reforms and longer life expectancy makes the universal benefit paid at 65 exhorbitant. The projected costs of the universal benefit in Uruguay would be less than the current pension deficits the government has to pay until 2030, and then is roughly equal to annual transfers to elderly people if there is no change in the current system. In Argentina, even when the benefit is generously set equal to the minimum wage and indexed to wage growth, the cost of the universal pension, although higher than the deficits of the current public pillar, mirror costs of a similar benefit in New Zealand. However, even in wealthier New Zealand, the transfer to the elderly pop- ulace is forecast to grow to 9 percent of GDP by 2050 (Willmore 2001a). The cost of the universal pension is highest in El Salvador, Colombia, and Chile--countries considered to be outliers in the region with respect to the level at which they set the minimum wage. Readers should keep in mind that these are simple simulations in- tended only to inform an important debate on the size and structure of the poverty prevention pillar. Clearly, the cost of setting benefits to the minimum wage and indexing to wage growth would make universal pen- sions inviable in most Latin American countries. The simulated costs of the different poverty prevention benefits are shown alongside current spending on public pensions to give readers an idea of their relative size. Much of the current cost of public pensions represents acquired rights that governments would find difficult to ignore in favor of purely poverty prevention programs. Furthermore, these simulations are only prelimi- nary and are intended to motivate a closer investigation of options for a poverty prevention pillar in each country. There is much that could be done to improve these cost estimates in a country-specific study of viable PREVENTING POVERTY IN OLD AGE 223 options. The projected costs do not take account of (a) the portion of the current transfers to the elderly population that is still financing the tran- sition cost of introducing individual retirement accounts; (b) the cost of excluding workers without a contribution history from receiving benefits; (c) the revenue that the government could claw back by setting a sur- charge on pension payments for workers above certain income levels or by simply making the universal flat benefit taxable; (d) the administrative costs of means tests to efficiently target a benefit to elderly poor people; and (e) the costs of corruption and leakage to people who are not poor that must be kept in mind, given the track record of public pension ad- ministration in the region. Adjusting the minimum retirement age to match changes in life expectancy would go far toward containing the costs of this benefit. Each of these considerations can change the pro- jected costs of poverty prevention benefits considerably and should be ex- amined carefully in country-specific analyses using PROST or other actu- arial simulation tools. Conclusion We have argued that there is little distinction between contributory and noncontributory public pension systems when the public policy objective is to prevent poverty in old age. The origin of this distinction between con- tributory and noncontributory benefits is political. Indeed, there may still be sound political economy grounds for requiring workers to pay some premiums to social insurance to maintain a strong constituency to protect budget allocations to these poverty prevention programs. That said, in countries where labor is very mobile among sectors and the informal economy is large, structuring the premiums for social insur- ance programs as payroll taxes is an increasingly ineffective and unreliable way to finance social insurance. A more reliable source of financing for public pooling mechanisms may be a value-added tax. In fact, financing public poverty prevention pensions through taxes other than payroll taxes would erase the distinction between the covered and uncovered sectors of the labor force. This requires that public poverty prevention pensions be viewed not as the social assistance charity of the state and society but as an additional instrument available to individuals and households to man- age shocks to their income should they need it. As pointed out in Averting the Old Age Crisis (World Bank 1994), greater reliance on a broad tax base, such as an income or consumption tax instead of a payroll tax, is the most efficient course for policymakers to pursue in the long run because it reduces the tax rate needed to finance benefits. It is also most consistent with the poverty prevention and redistributive functions of the remaining public pooling pillar after in- troduction of the multipillar model (Willmore 2001b). 224 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA We have also argued in this chapter that if Latin American countries de- cide to continue to meet poverty prevention objectives through contribu- tory systems (coupled with means testing for the uncovered population), the contributory part should ideally be assigned to a separate pooling pil- lar offering a minimum benefit to all qualifying contributors. Making the minimum pension guarantee conditional on participation in the savings component creates an unnecessary tension between the need for risk pool- ing to optimize poverty prevention and the need for individual choice to tailor consumption smoothing to individual preferences. Policymakers in El Salvador, Mexico, and Peru should also consider the introduction of means-tested social assistance for the elderly population. Another promis- ing avenue is outreach to elderly households in programs that target poverty broadly like Chile's Puente and Mexico's Oportunidades. With- out these additional instruments the basic promise of old-age social secu- rity, poverty prevention among elderly men and women, will not be kept. Ultimately the experience with single-pillar, defined-benefit, PAYG arrangements in Latin America teaches us that the proverbial devil in pub- lic pooled pension plans is in the details. The details can only be deci- phered with more focused, country-specific analytical work conducted with an eye to competing demands on limited fiscal resources, such as so- cial assistance transfers to other, perhaps needier groups in the popula- tion. This analysis should address at least the following questions: (a) Should a separate, poverty prevention first pillar be universal or targeted? (b) Will political considerations allow a new minimalist, universal public pooled pillar to replace partly or wholly other contributory and noncon- tributory benefits mandated and guaranteed by governments? (c) In a country where the informal sector is large and tax evasion is rampant, how could the benefit best be financed? (d) Would the benefit be a flat amount or related to the number of years of contributions? (e) What would be the maximum old-age poverty prevention benefit the government could cred- ibly guarantee and sustain? (f) How would the benefit be indexed to pro- tect its real value? (g) Would the benefit be taxable? and (h) At what age should individuals be eligible? Notes 1. This is not to say that the market for private annuities can be taken for granted. Even in middle-income countries like those in Latin America that intro- duced private individual accounts, much still has to be done to ensure that there is a well-regulated, transparent, and competitive annuities market to cater to the de- mands of affiliates retiring under the new pension systems. 2. Poverty statistics typically are calculated using current income from em- ployment, pensions, or public and private transfers reported in representative household and labor market surveys. Older respondents, many of whom are likely to be retired, will report lower levels of income, which will naturally place them in the lower tail of the income distribution. PREVENTING POVERTY IN OLD AGE 225 3. Purely public Bismarkian PAYG systems typically defined benefits of be- tween 80 percent and 100 percent of average wages earned in the last few years (and in many cases the last month) prior to retirement. Although based primarily on defining contributions in individual accounts, the replacement rates envisioned by policymakers setting the contribution parameters of the new multipillar systems ranged from 60 percent to 70 percent of some average of earnings prior to retire- ment (Pinera 1995; Valdés-Prieto 2002b; Rofman 2002). 4. Only Argentina, Costa Rica, and Uruguay have assigned the poverty pre- vention role to a separate pillar based exclusively on risk pooling. 5. The targeted or minimum benefit of this component could be set as is cur- rently the case in most Latin American countries, as a percentage of the minimum wage. Alternatively, it could be set, as in the richer OECD countries, in relation to average earnings. In Latin America, as in OECD countries, the minimum pension benefit tends to be somewhere between one-quarter and one-half of average econ- omywide earnings. 6. Notable exceptions are the pensiones graciables found among the noncon- tributory pension programs in Argentina and Uruguay. These pensions were cre- ated to recognize notable contributions made by citizens to what were deemed na- tional interests (e.g., in recognition of military service or international distinction in the arts, sports, and science). In both cases these pension programs have deteri- orated into fiscally regressive and costly tools of political patronage. 7. The case study for Chile in Bertranou et al. (2002) reported that many more PASIS benefits were being paid than there were elderly poor people in the country in 2000. Taking the share of PASIS beneficiaries as a percentage of the eld- erly poor population gives a coverage rate of 154 percent. Because PASIS benefits are paid to poor disabled people and to widows, the rate reported in Bertranou et al. (2002) is likely to overstate coverage of the elderly poor populace. The figures reported for Chile in the last column of table 9.2 are taken directly from the CASEN (2000) household survey. 8. PROST simulations reported in this section were performed by Asta Zviniene (HDNSP; World Bank Human Development Network, Social Protection Unit), using the same macroeconomic assumptions employed in the simulations re- ported earlier and presented in Zviniene and Packard (2002). 10 Facilitating Consumption Smoothing: Improving the Savings Component IN THIS CHAPTER WE EVALUATE various policies aimed at improving the functioning of the savings pillar so that it best serves its consumption- smoothing role. In some countries, however, the funded system has a dual role, serving also a poverty prevention function. In chapter 9 we examined alternate ways to deal with the poverty prevention goal and argued that it would be best served by a separate defined benefit pillar, which should be financed by general tax revenues. The discussion in this chapter focuses only on the goal of consumption smoothing for most of the population, above what is required to prevent poverty in old age. The evidence provided in chapters 7 and 8 raises some concern over the ability of Latin American governments to mandate a retirement savings instrument that is attractive for a reasonably high proportion of individu- als. The low density of contributions to the mandatory pillar and the reported strategic behavior by workers to qualify for the minimum pen- sion guarantee may be primarily a consequence of widespread informality, low disposable incomes, and pressing consumption demands that leave lit- tle space for long-term savings. However, we also saw in chapter 8 that Latin American workers are saving and investing through other means, such as buying a house. It is therefore possible that the size of the mandate may be too large, that some of the mandated products' characteristics (cost, risk, and liquidity) are not sufficiently attractive, and that the prod- ucts' weaknesses are not compensated for by preferential tax treatment of retirement savings. High contribution rates may discourage poor and young workers from participating in the mandatory pension system because these groups are most likely to have competing demands on their income. One policy option that may be considered is to reduce the size of mandatory 227 228 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA contributions for these groups. Poorer workers may even be exempt from the mandate to save. After all, their basic needs should already be taken care of by the poverty prevention pillar. We also suggest reducing the earn- ings ceilings for mandatory contributions so that the voluntary pension system can develop. The limited acceptance of opportunities to make additional, voluntary contributions to individual accounts in countries such as Chile, Colombia, Mexico, and Peru could be taken as further evidence of problems in the funded system. In these countries voluntary retirement savings in the new private second pillar are relatively liquid and therefore compete directly with alternative investment instruments. Although voluntary saving in the products offered by the AFPs is relatively stagnant, saving in alternative financial instruments has grown at fast rates in Chile, even if those instru- ments often have a less advantageous tax treatment and involve additional costs of intermediation. Aside from the size of the mandate on many formal workers, the lack of interest in voluntary savings could be explained by a desire among in- dividuals to diversify their savings into other products. It seems strange, however, that workers would prefer to save through instruments that are nearly always more expensive than the pension funds (bank deposits be- ing the main exception) and that offer significantly lower tax benefits. We suggest several changes to the new private second pillar to lower administrative costs and improve the management of investment and longevity risks, drawing on work commissioned specifically for this book as well as a large and growing body of literature. These recommended changes in the structure of the private second pillar (changes such as greater competition, enlarging fund choice, centralizing account management and record keeping, and diversifying investment in foreign securities more widely) can indeed lower administrative costs of the new systems and fur- ther improve investment performance, but it is critical that governments ensure that these benefits are passed on to affiliates of the pension system, and not simply kept by the fund managers to increase their profit margins. As in most areas, the critical ingredient to ensuring that affiliated workers benefit from the improved performance of the new private second pillar is competition. Competition might be introduced with small but im- portant changes to the structure of the dedicated financial industry. But it is even more likely to arise from creating an enabling environment for pri- vate providers of voluntary savings and insurance instruments that can substitute for mandated savings and insurance in the private second pillar, and from ensuring a level playing field with respect to the forms of long- term saving and investment the government rewards with efficient regula- tory oversight and moderately favorable tax incentives. Fortunately, this is the direction that governments already seem to be headed in countries where structural reforms have begun to mature. FACILITATING CONSUMPTION SMOOTHING 229 Improving the Mandatory Savings Pillar: Lowering Costs to Affiliates In this section we focus on ways to improve the income-smoothing role of the mandatory funded pillar. We propose various reforms aimed at elimi- nating some of the failings of the existing systems identified in chapter 7, such as the high cost and lack of efficient management of investment and intragenerational longevity risk. The objective of the proposed reforms is to offer workers a better (risk- and cost-adjusted) return than alternative investment products offer. The last subsection discusses the mandate itself and how it affects populations such as the vast majority of Latin Americans who live with monthly incomes of less than US$500. Helping People Meet Their Differing Life Cycle Needs The main difficulty in designing mandatory savings programs that complement poverty prevention pillars as described in chapter 9 is that individuals have different consumption needs and objectives and face innumerable investment choices throughout their life cycle. While their ultimate goal is recognized to be consumption smoothing, individual time and risk preferences will determine specific consumption paths and investment choices, which may include not only financial assets but also housing, education, and for some people even their own business. Provident workers who recognize their needs and the value of saving for retirement may even have negative savings rates when young because of expectations of higher labor earnings in subsequent years. Indeed, efficient consumption smoothing may dictate that such individuals borrow when young. Mandating savings can be costly for such workers in two ways. First, the mandate may force workers to borrow even more than they would do in the absence of the mandate, at rates that are typically higher than the returns earned on their savings. This spread is highest in developing coun- tries and alone may be sufficient reason to evade the mandatory system. For those people who are unable to borrow, the mandate may force them to bypass other investment opportunities (e.g., housing and education) and may leave them unprotected against income shocks. Second, the opportunity cost of mandatory savings is compounded in Latin America by the fact that the new system duplicates the fixed costs of financial intermediation. Poorer and improvident workers are likely to suffer the most from the mandate. Low-income workers normally face greater credit constraints and may therefore be less able to adjust their consumption to any mandate to save. Although they may suffer the least in terms of spread costs be- tween borrowing and savings rates, they suffer the most in terms of lowered 230 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA disposable income, missed investment opportunities, and exposure to un- expected income shocks. The duplication of fixed costs of financial inter- mediation also impinges most on the income of poorer workers. High mandatory contributions can also bring these workers closer to the poverty line during their youth. Mandating contributions of poorer workers above the level required to prevent poverty in old age is therefore contentious. By limiting mandatory contributions of poorer households to a level sufficient to ensure a mini- mum income at retirement, the negative effects of the mandate can be minimized. Such contributions should be directed toward a separate plan with purely poverty prevention objectives, as discussed in chapter 9. Some Latin American countries, however, do not have separate contributions for poverty prevention and for consumption-smoothing purposes. While it is possible in principle to organize the first pillar as a minimum pension guarantee on fully funded individual accounts (as in Chile or El Salvador), the operational cost of such an arrangement can be exacting. The advan- tages of fully funding poverty-level benefits are also debatable when one considers that pension contributions that have a poverty prevention ob- jective should be invested in long-term, inflation- or, even better, wage- indexed bonds with minimal credit risk. Such conservative investment strategies are needed in order to ensure that the minimum pension is reached with a high degree of certainty. To the extent that lower-income workers are covered by a separate poverty prevention pillar, there seems to be little need to require them to make additional contributions to the pillar that serves a consumption- smoothing role.1 The level of retirement income that brings poor people out of poverty is similar to that needed for consumption-smoothing purposes.2 For higher-income workers one can also question to what extent the consumption-smoothing objective should be achieved through a mandatory savings plan. Improvident workers, regardless of their income, suffer from the mandate. Improvident workers are likely to borrow more than provi- dent workers and as a consequence pay a higher price in terms of the spread charged by credit institutions. How much both provident and improvident workers pay in terms of higher intermediation costs will depend on their access to credit. Improvidence, however, is a function of age. In general, older workers tend to be more provident because they realize the conse- quences of having insufficient resources in old age. Hence they are more likely to accept mandatory contributions to a retirement savings plan. There is evidence that workers in Latin America may be evading the mandatory system to invest their limited spare savings in housing (Packard 2002a; Barr and Packard 2003). This fact is hardly surprising given the acuteness of housing deficiencies and the importance individuals attach to home ownership as a shock absorber. Permitting investment of mandatory pension contributions in housing could be a necessary first step to help FACILITATING CONSUMPTION SMOOTHING 231 individuals meet their consumption and savings objectives throughout their life cycle. Doing so would require reform of the pension system to permit affiliates to use accumulated funds as down payments to obtain mortgages from banks. Controlling the final use of such withdrawals, however, can be a daunting task for governments. For example, it may be difficult to prevent those people who already own a house from selling it to qualify for a fund withdrawal. Withdrawals could also be permitted for investment in human capital, such as education and health care. Such a possibility is particularly important for low- and middle-income households who have limited sources of savings and who face the greatest borrowing constraints. The fungibility of funds withdrawn, however, may make it very difficult for policymakers to control the use of withdrawals destined for these specific purposes. It would also add tremendously to costs. Similarly, permitting withdrawals in cases of emergency, such as unexpected income or wealth shocks, may make sense from the perspective of individual welfare, but the monitoring costs would be very high. Introducing such flexibility into the system, moreover, can create incentives for moral hazard. Poorer workers wishing to access their accumulated savings may hide their assets or join the informal sector. In general the more governments aim to satisfy individual preferences over investment, the more complicated the monitoring of the system becomes and the greater the risk that governments will be called to bail out unfortunate workers who made bad choices. Hence, if it is deemed that mandatory contributions may be unnecessarily high for many workers, a better policy response is to lower the mandate rather than permit with- drawals. High contributions coupled with a liberal policy on withdrawals may lead to mistaken perceptions about the adequacy of accumulated re- tirement funds among both individuals and the government. The ease with which individuals avoid the mandate in Latin American countries casts doubt on its usefulness, at least at current contribution levels. It is likely that the extent of evasion is partly linked to the level of mandated contributions. That is, if mandatory contribution rates are decreased and the maximum taxable wage reduced, the contribution frequency is likely to increase. Participation is especially likely to increase among poorer and younger workers who have many competing consumption demands (purchasing a home, professional education, raising children). Saving for retirement may therefore come at a high cost for them. For richer and older workers closer to retirement, mandatory contributions may be less costly because there are fewer urgent demands on their income. At the same time, however, these workers are the least improvident and therefore the least likely to change their behavior as a result of a mandate to save. Overall, the goals of maximizing social welfare translate into establish- ing contribution rates and maximum taxable wages that overcome the 232 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA problem of improvidence in youth but that minimize the costs of financial intermediation and leave workers sufficient disposable income for other purposes (maintaining a family, buying a home, or meeting unexpected income shocks). Contribution rates should also be set to minimize moral hazard from the first pillar (reduction in voluntary savings). To achieve these goals, mandatory contribution rates above the level re- quired to finance the poverty prevention pillar could be linked to the age and income of the individual.3 A useful reference is the Swiss system in which mandatory contribution rates in the mandatory occupational pension system (Switzerland's second pillar) increase gradually with a worker's age (from 5 percent in the 30s to 15 percent in the 50s). Earnings-linked con- tribution rates are also applied in Colombia and in some OECD countries. The maximum contribution rate and taxable wage consistent with social welfare objectives are likely to be much lower in Latin America than in countries where disposable household incomes are much higher. Although mandatory contribution rates (to the PAYG or funded compo- nent) are generally lower in Latin America than in OECD countries (Argentina, Brazil, and Uruguay are the main exceptions), the maximum taxable wage is much higher. Except in Chile, the maximum taxable wage in Latin American countries is more than five times average earnings, whereas in OECD countries it is often less than twice the average wage. From a social welfare perspective, therefore, it seems appropriate for mandatory contributions above those that finance the poverty prevention pillar4 to be related to both the income and age of the individual.5 The ceiling on taxable earnings could also be gradually lowered from its current level in countries with a high ceiling, such as Bolivia or Mexico, and a ceiling could be introduced in countries that do not currently have one, such as Costa Rica and Peru. This would free up resources that could be invested in a wider array of financial instruments and other attractive long-term assets such as housing or education. Such a solution may not be perfect because it does not take into account all possible variations in the degree of improvidence or access to credit among workers of a similar age cohort or income group. It would be un- reasonable to expect governments to manage a mandated system in the best interest of all individuals because such an ideal system would require governments to cater to each individual's specific needs and constraints. By reducing the size of the mandatory funded pillar for the young and the poor, however, policymakers can reduce distortions to individual choice while still helping to mitigate the dangers of improvidence. Enacting Reforms to Decrease Administrative Costs and Commissions One of the central policy concerns in any mandatory pension system that is privately managed is the cost of administration. The Latin American FACILITATING CONSUMPTION SMOOTHING 233 systems have been generally successful at reducing costs, but it appears that this has come largely by restricting individual choice and competition among the pension fund administrators. In particular, marketing and distribution expenses have been reduced by restricting the number of times per year that an affiliated worker can switch between pension fund admin- istrators (once in Mexico, Peru, and Bolivia; twice in other countries). These restrictions essentially have created a captive clientele for each pension fund administrator and institutionalized what was already an oligopoly. Even un- der such restrictive conditions, however, commissions may still be unaccept- ably high for a large percentage of the population. More can still be done to lower costs. Indeed, the high returns on capital for the pension fund ad- ministrators in some countries prove that only a small portion of the decline in operating expenses is being passed on to affiliates as lower commissions. In Peru, for instance, profit margins for the fund managers are growing with efficiency, yet the price of the private pillar paid by affiliates has re- mained stubbornly high. Fees charged by AFPs in Peru were the highest-- 30 percent of affiliates' net contributions (that is, contributions net of fees and premiums for insurance)--considerably above the regional average of 15 percent. Carranza and Morón (2003) argued that Peruvian fees are much closer to the regional average when one allows for the lower contri- bution rate in Peru. Despite this, the average cost to affiliates has persisted for almost five years--even though the number of contributors has in- creased and the industry has had more time to become more efficient and amortize the initial set-up costs (see figure 10.1). The minimal variation in that percentage across all four AFPs during the past five years is another indication of the weakness of competitive market forces in the industry. Lasaga and Pollner (2003) attributed persistently high AFP fees in Peru to a number of structural factors that hamper competition among the fund managers and may have contributed to forming an oligopoly. These fac- tors include (a) the high start-up costs of entering the dedicated private pension fund administration industry; (b) the difficulty of comparing AFP fees (assessed on a prepaid, percentage-of-income basis) with the charges of similar service providers in the private financial sector (often assessed as a percentage of assets under management); (c) opaque and outdated provision of insurance services tied to mandatory retirement savings; (d) the high transaction costs to affiliates opting to switch from one AFP to another; and (e) the model of "dedicated provision" itself and its implied exclusion of other regulated financial sector actors (some, like commercial banks, with established and farther-reaching supply networks) from the managing of mandated retirement savings. As the Latin American funded systems become more flexible and permit participants to exert some degree of portfolio choice, policymakers will have to search for more creative solutions to the problem of adminis- trative costs. Possible solutions fall into two main categories: those that aim to reduce operational costs through supply-side measures or caps on 234 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 10.1 In Peru, Fees Remain Persistently High, Despite Increasing Returns and Declining Administrative Costs 100.0 30 Fees/net 90.0 Fees/net contributions Operational expenses/net fees 25 80.0 contributions (%) Return on equity 70.0 20net 60.0 fees equity 50.0 15 on (%) 40.0 and 10 Return 30.0 expenses/ 20.0 5 10.0 0 0 1998 1999 2000 2001 2002 Source: Lasaga and Pollner (2003). commissions (see table 10.1), and those that employ demand-side measures aimed primarily at increasing the price elasticity of demand (see table 10.2). Supply-side measures are certainly the most effective. One promising measure is to permit group contracts. This would involve raising the pro- file of employers in pension plan administration. Firms use retirement plans as part of their human resources policies to attract and retain talented employees. A greater involvement of employers in pension provision could therefore result in higher voluntary pension contributions and a better diversification of retirement income sources. The possible role of the employer could range from record keeping and assistance in the selection of investment products and providers to direct provision of pension plans. Nonetheless, it is unlikely and possibly unde- sirable for any but the largest Latin American employers to offer defined benefit plans. Such plans require expensive administrative structures and regular supervision. Moreover, the trend toward greater transparency in company accounts (as international accounting standards are adopted) is increasing the visibility of the pension liabilities of employers, who in turn are shrinking or altogether closing down these plans and substituting plans with more limited guarantees or purely defined contribution plans.6 Table 10.1 Reform Options to Reduce the Operational Expenses of the New Private Second Pillar FACILITATING Reform proposal Rationale Country case Potential problems Limit the frequency with Marketing costs are lowered All Latin American countries Competition among pension which workers can switch because the ability to attract limit switching to one or two fund administrators is fur- between pension fund new members is curtailed. times per year. In Bolivia ther limited. The govern- CONSUMPTION managers. switching was prohibited ment's fiduciary responsibil- after the two pension fund ity is even greater because administrators merged. the only element of choice (and therefore of market dis- cipline) is removed. SMOOTHING Allow group contracts, per- Employer is a countervailing No Latin American country Self-employed workers are not mitting the establishment of force to the pension fund has yet permitted group con- able to benefit from commis- occupational pension plans administrator, with much tracts. sions as low as those avail- where the employer negoti- more bargaining power than able to wage and salaried ates the commission with the individuals have in the retail workers. pension fund administrator. market. Separate pension services and Record keeping and individ- Mexico and the Dominican The auctioning process is centralize some services, such ual account management are Republic have centralized subject to political risk. as record keeping related to services with significant record-keeping systems. The the management of individual economies of scale. In addi- new funded system in Swe- accounts. These services may tion, if pension fund admin- den also has centralized con- be auctioned off by the gov- istrators cannot know the tribution collection and ernment in an international identity of the account hold- record keeping. open bid. Alternatively, ers, marketing costs are employers may be responsible significantly curtailed. 235 for record keeping. (Table continues on the following page.) 236 Table 10.1 (continued) Reform proposal Rationale Country case Potential problems Allow other financial sector Economies of scale in asset No country permits other Big financial institutions will KEEPING institutions to manage management and additional than dedicated providers to dominate the pensions mar- pension fund assets. investment expertise will manage pension funds. ket even more. Monitoring lead to lower costs and conflicts of interest could THE better investment become more difficult. performance. PROMISE Source: Authors' compilation. OF SOCIAL SECURITY IN LATIN AMERICA FACILITATING CONSUMPTION SMOOTHING 237 That said, there is much to be gained from the employer acting as an intermediary between pension service providers and employees in defined contribution plans. Latin American governments have a lot to learn from the long, international experience in occupational pension provision. There are many generally successful examples of mandatory occupational plans based on defined contribution formulas, such as those in Australia, Denmark, Hong Kong (China), and Switzerland.7 Of all Latin American countries that have introduced mandatory indi- vidual retirement accounts, only Costa Rica permits employers to contract the funds management directly with pension fund administrators. This authorization, however, only extends to voluntary pension contributions. Costa Rica has also permitted the continuation of occupational defined benefit plans for some public sector workers, some of which are comple- mentary to the individual account system. Arrau and Valdés-Prieto (2001) proposed reforms to the mandatory pension system in Chile along the lines of the United States' 401(k) system, but with limits on investment in company stock. The Chilean parliament has considered a similar proposal. Further discussion of this policy option is needed in other Latin American countries. Occupational pension plans that involve the employers in the negotiation of fees with pension service providers (asset managers) can help keep costs down and simplify investment choices for workers. If other countries follow Chile and liberalize product and provider choice in their funded pension systems, the employers' role will become all the more valuable for securing retirement income for affiliates. One important caveat that should be considered here is that employer plans are notoriously subject to agency risk. However, regulations can be designed to ameliorate this problem, as has been done in the current system to minimize conflicts of interest between plan members and fund managers. A potential conflict of interest could arise in an employer- intermediated plan if workers pay commissions that are negotiated between employers and fund managers. It may, therefore, be better for employers to pay commissions, as proposed by Arrau and Valdés-Prieto (2001). Nonetheless, it should be noted that supervision of conflicts in occupational schemes may be much more expensive than in the current personal plans, given the large number of employers present. That said, one could argue that employees already have means of monitoring their employers' actions via labor market institutions in many countries. Another problem of the proposal is that self-employed workers and those working for small employers (the majority of workers in Latin America) may be left at a disadvantage. In fact, salaried workers already benefit from voluntary employer contributions in the existing system. Small employers could also sponsor plans jointly (multi-employer pension plans are common in Brazil, for example). If the government is concerned about the interests of the self-employed population, the adequate response should be to subsidize contributions by these workers rather than impede 238 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA access for salaried workers to more efficient retirement arrangements.8 This solution was recently put into practice in the Dominican Republic and should be monitored closely and evaluated carefully, particularly with regard to its fiscal costs. It may also be argued that employer intermediation, although superior to personal arrangements in terms of cost efficiency, may still be more expensive than a more centralized solution. Bolivia is the Latin American country that comes closest to a centralized model, especially because the two Spanish banks that owned the two pension fund administrators merged in 2000. The Bolivian reform also included caps on commissions and the auctioning of licenses for pension fund administration. Another example of a bidding contest for assigning fund management was the Chilean auction for managing the individual accounts of the new unem- ployment insurance program. This contest led to commission levels close to US$4 per worker per year. Chilean pension fund administrators charge 20 times this amount for managing individual retirement accounts. One could argue that full centralization of pension administration would obliterate the benefits of competition in private pension fund man- agement. Evidence from Latin America, however, clearly shows that the current pension fund industries are anything but good examples of competition. The Bolivian case also demonstrates that governments can organize transparent bidding contests for pension fund management that are less vulnerable to political risk. An alternative to full centralization (as in the case of the de facto Bolivian monopoly) is to centralize only certain services, such as contribution collection and record-keeping functions that have significant economies of scale and lower policy risk. Centralized contribution collection is already in place in some Latin American countries, such as Argentina and Mexico, but it does not seem to have led to lower costs than in countries with decentralized mechanisms (Demarco and Rofman 1998). An assessment of the causes is now overdue. Outside Latin America, Sweden and Estonia have embraced centraliza- tion in pension administration services other than asset management. Sweden has centralized account management, record keeping, and benefit payment (in the form of annuities). Only asset management is in the hands of the private sector. Workers choose among the full range of mutual funds licensed in the country.9 The advantage of this system is that mutual fund managers ignore the identity of the owners of the assets that they manage. As a result, sales agents that have proved so costly, because there is no way to check whether a person targeted by an agent, has actually become a new client of the fund manager. Such proposals are certainly worthy of consideration in Latin America, although the experience with public pension systems raises doubts about the ability of governments in the region to manage centralized systems as is done in Sweden.10 One way to avoid this problem is to transfer the FACILITATING CONSUMPTION SMOOTHING 239 record-keeping function to employers. Employers are already required to ensure that contributions are paid on time and must maintain a record of the contributions made to the pension system on behalf of their employees. Parallel efforts should be made also to promote competition and choice of provider in asset management and insurance services. Competition could be enhanced by opening up the pension fund management industry to other financial institutions that would act as professional asset managers. Greater competition is also needed in the market for insurance products (e.g., annu- ities) in countries such as Argentina, where the affiliates are offered only the products of the company that is tied to the pension fund administrator. Caps on commissions would also seem to be highly relevant for Latin America, where financial literacy is lower than in OECD countries and financial products are still relatively standardized given the limited invest- ment opportunities (table 10.2). Caps are also becoming popular in some OECD countries such as Ireland, Sweden, and the United Kingdom. Under the United Kingdom's stakeholder proposal private providers may only charge fees on assets under management, subject to a 1 percent limit. In Sweden fees are also capped as a percentage of assets, but they vary depending on the size of the fund. Governments can also do much to increase the affiliates' responsiveness to commissions so that they choose those providers who offer the lowest fees. The transparency of fees can be enhanced by requiring the disclosure of fees for the whole industry (clearly separating administration fees from insurance premiums), deducting the fees from the accumulated balance (though still calculated on the basis of contributions or salaries), and sim- ulating their impact on the expected pension benefit. Such information could be provided in the regular account statements that affiliates receive. A similar degree of disclosure could be used for annuity rates. Another, more interventionist mechanism is to allocate indecisive workers to the fund manager offering the lowest commissions. Finally, pension fund ad- ministrators could be permitted to offer lower commissions for "loyal" af- filiates who stay for a certain period with the same administrator. This is currently only permitted in some Latin American countries. There is much ongoing experimentation in this area, and no one-size-fits-all solution is likely to emerge. Improving Risk Management of Mandated Retirement Savings The second key feature of the consumption-smoothing component that should be reformed is the design of investment regulations and of individ- ual choice in the system to ensure (a) a better fit with the affiliates' appetite for risk and liquidity and (b) better management of investment and longevity risk. Reforms are needed in both the accumulation (saving) and the distribution (retirement) phases, although probably more radical Table 10.2 Reform Options to Ensure That Savings Are Passed on to Affiliates as Lower Commissions 240 Reform proposal Rationale Country case Potential problems Place caps on commissions. This is an effective way to limit All Latin American countries The cap may be set too low fees. When accompanied by except Mexico already regu- and will need to be adjusted KEEPING regulations on fee structures late commission structures, as technology and market it ensures a high degree of but only Bolivia has actually conditions change. transparency for investors. imposed limits on fees. THE Improve the visibility and Commissions paid out from Since November 1998 Chile The participant may pay PROMISE comparability of commis- the accumulated balance are requires that all information higher fees for the stricter sions by requiring that all more visible to workers than regarding commissions be disclosure requirements. commissions be deducted are commissions deducted included in the quarterly from the accumulated from payroll. statements sent to partici- OF balance rather then deducted pants, including a compari- SOCIAL directly from payroll. son of commissions. Since Visibility can also be May 2000 the types of com- enhanced by disclosing com- missions charged (fixed, SECURITY parative fees across pension variable, and transfer fees) fund administrators. must also be clearly disclosed. IN Establish an information Lack of transparency in In Chile pension fund admin- There is a danger of cartels LATIN process for transmitting annuity sales has driven istrators must disclose vari- being formed between information about benefit costs up. Workers take cash ous annuity options avail- pension fund administrators AMERICA options to plan members, so rebates instead of focusing able to the plan members and insurance companies if that they can compare on annuity rates. prior to annuity sales. they are allowed to exchange annuity rates and other information about plan relevant variables. members. FACILITATING Allocate workers who enter By choosing the lowest possi- This mechanism for allocating The only element of individual formal sector jobs for the ble commission the average indecisive workers was choice in the system is elimi- first time and do not choose cost of the system is reduced. introduced in Argentina at nated. All key decisions are a pension fund administra- the end of 2001. made by the pension fund tor, to the cheapest adminis- administrators under the trator (according to the surveillance of the state. CONSUMPTION expected income level). Permit pension fund "Loyalty" discounts create an Argentina and Mexico allow Discounts may not discourage administrators to offer lower incentive for workers to "loyalty" discounts. other administrators from commissions for "loyal" choose their fund adminis- offering rewards to workers affiliates who stay with the trator more carefully who change affiliation. SMOOTHING same administrator for a because when they change certain period. they forgo the reduction in commissions. Source: Authors' compilation. 241 242 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA changes are needed in the former given the practical lack of individual choice. Some basic reforms are needed to improve the management of invest- ment risk (table 10.3). First, investment in securities issued by foreign entities should be permitted. Chile is currently the country that has gone farthest in liberalizing overseas investment by pension funds, raising the ceiling to 20 percent of the pension funds' portfolios. Diversification abroad is especially important in small countries with limited opportuni- ties for domestic investment, and in countries with unstable macroeco- nomic conditions. But it is precisely these countries that typically impose the strictest limits on foreign investment. By steering investments toward domestic markets governments hope to develop their capital markets and reduce dependency on volatile foreign capital. They instead increase workers' investment risk. Second, the single-fund model should be relaxed and at least two other funds should be introduced. One of the funds should be a low-risk fund invested mainly in debt of the highest credit rating (both domestic and international), inflation-indexed as far as possible, and covering long maturities. The second fund should be a diversified portfolio of riskier assets, including corporate debt and equities. The third fund would con- sist of a domestic money market portfolio. Workers would be allowed to choose among these three funds, but limits might be imposed on the last two funds for prudential reasons. The logic of this arrangement is as follows. The first fund would be, to the extent possible, the long-term risk-free asset that an investor needs to build an optimal portfolio for retirement. In countries such as Chile, where there is a liquid market for inflation-indexed, long-term government debt of high credit standing, such a portfolio could consist mainly of debt in- struments. In other countries where domestic debt has a higher default risk there would be a need to permit greater investment in debt issued by for- eign governments and corporations. The second fund would permit work- ers with greater appetites for risk to invest in a well-diversified portfolio of bonds and equities. Such a fund would contain both domestic and for- eign securities, depending on the quality of the former. The third fund would permit workers near retirement to increase the liquidity in their portfolio to cash out some of their savings. This basic three-fund system is similar to the five-fund system introduced in Chile in 2002 and legislated in Peru in 2003 (to be introduced in 2005), as well as the three-fund system of the thrift savings plans (TSPs) for federal government employees in the United States.11 The main difference is that the three-fund model would allow for greater flexibility in the choice be- tween domestic and foreign assets, and would clearly identify a portfolio that is as far as possible a long-term risk-free asset. In the TSP, the G fund approximates this because Treasuries are arguably the safest securities in the world. However, because most of the instruments in this portfolio are Table 10.3 Reform Options to Improve Investment and Longevity Risk Management FACILITATING Reform proposal Rationale Country case Potential problems Liberalize investment Given high volatility of domes- Chile is the country that has Foreign investment may bring undesir- abroad by pension tic securities, foreign invest- gone farthest in liberalizing able macroeconomic consequences, funds. ment (especially equities) can foreign investment. Mexico institutionalizing capital flight. It is CONSUMPTION significantly improve is the last country to permit also difficult for governments to risk­return trade-offs. such investment. allow such investment when the domestic economy has insufficient access to financing. Introduce a multifund Individuals would be able to Chile introduced a five-fund Some individuals may find it difficult SMOOTHING structure with adjust their portfolios to suit structure in 2002. Peru legis- to make choices. In the Chilean case different portfolio their risk preferences. lated a similar multifund undecided people are allocated to a allocations. structure in 2003. fund according to their age. Authorize variable Some individuals may prefer Bolivia is the only country Variable annuities expose individuals to annuities. to bear investment risk dur- that has authorized variable the volatility of financial markets. If ing the retirement age and annuities. the funds are largely invested in equi- thus aim at higher benefits. ties there can be massive changes in the pension benefit during retirement. Authorize purchase By spreading the purchase of No Latin American country The administrative cost of multiple- of annuities with an annuity over many years, allows individuals to spread premium annuities tends to be multiple premiums. one reduces timing risk (the out the annuitization of the greater than that of single-premium risk that interest rates are accumulated balance. annuities because money has to be low at the time of purchase). collected and annuitized on many occasions. 243 (Table continues on the following page.) 244 Table 10.3 (continued) Reform proposal Rationale Country case Potential problems Authorize deferral of The value of annuitization Chile and Peru allow workers Workers may be forced to enter annu- KEEPING annuity purchase increases as a worker to switch from scheduled ity contracts under adverse condi- after retirement, becomes older: there is a withdrawals to annuities tions (at times of low interest rates) opting for scheduled higher risk of outliving one's after retirement. if they leave the annuitization option THE withdrawals in the resources, the bequest until late in their working lives. meantime. motive is less important, and PROMISE the cost of annuities is lower because of the higher proba- bility of death. OF Modernize mortality Mortality tables should be Chile created mortality tables Local annuity tables require suffi- SOCIAL tables. based on recent historical based on annuitants in the ciently large samples and history to experience of annuitants in 1970s; the tables are used in reduce statistical errors. the country. Peru, too. SECURITY Permit saving through Mutual funds are very similar No country has opened up the The lack of transparency in perform- other financial to pension funds but offer a mandatory funded compo- ance and fees may impede appropri- instruments such as much broader array of nent to competition. Chile ate individual choices. Allowing a IN mutual funds and investment choices. Banks opened up the voluntary wider set of financial sector actors to LATIN savings products and life insurance companies component to other financial manage mandated savings makes offered by banks can offer interest rate guar- instruments and providers in supervision more difficult and could and the life insur- antees that may be attractive 2002. increase the risk of losses from AMERICA ance industry. for workers. mismanagement. Source: Authors' compilation. FACILITATING CONSUMPTION SMOOTHING 245 not inflation-indexed (the U.S. government only started issuing such bonds in the 1990s), there is no protection against inflation over a long investment horizon. The introduction of individual choice in investment allocation would not only permit the construction of optimal retirement savings portfolios, but would also help further insulate the pension system from political risk. In its current form, where a handful of investment directors control all pension assets, it is too tempting for governments to change the invest- ment regimen on an ad hoc basis to meet their needs or objectives, such as easing a fiscal deficit or veering investment toward unprofitable or risky industries that support the elected government. It is crucial also that the introduction of portfolio choice is accompanied by intensive financial education campaigns. Reforms are also needed during the retirement stage to permit better management of investment and longevity risk. Some aspects of the design of the distribution stage appear problematic. First, while a case can be made for mandating annuities for the portion of the accumulated balance needed to keep the worker out of poverty in old age, this is better accom- plished by a separate poverty prevention pillar. For contributions above this level, mandating annuities is less compelling. Poorer workers tend to have shorter lives and greater consumption needs, so obliging them to buy annuities can have regressive effects on the income distribution. Yet in Bolivia and Uruguay annuities are mandatory at the retirement age. Second, variable annuities, in which investment risk is borne by the pensioner and longevity risk is borne by the insurance companies, do not yet exist in any country and are only permitted in Bolivia. These annuities may be particularly attractive for higher-income workers or for workers in areas where the public pension system still offers generous retirement income. Third, the requirement to buy annuities with a single premium impinges on the ability of individuals to smooth the risk of mistiming the annuity purchase. Some people will find that at the time of retirement long-term interest rates, and thus the payout that insurance companies offer them, are lower than they had expected. Ideally, an individual would buy an annuity in installments or would purchase successive annuities to smooth this volatility. Fourth, more flexibility is also needed to enable workers to switch from scheduled withdrawals to annuities at some point during their retirement, as is the case in Chile and Peru. The value of annuitizing the remaining balance increases as a worker grows older: bequest motives become less important as the fund is drawn down, the risk of outliving one's savings becomes higher, and the cost of annuitizing decreases. Fifth, the requirement that private annuities be indexed to inflation in countries that lack liquid inflation-indexed government bonds (such as Colombia and Peru) seems irresponsible. Although the intention behind 246 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA this requirement is good (protecting elderly people against price increases), the results are likely to be detrimental to workers. Private insurance companies will charge hefty fees for protecting benefits against a risk that ultimately only the government can affect. Reforms are also needed to improve information on life expectancy of annuitants. Some Latin American countries (e.g., Argentina) rely on mor- tality tables from OECD countries that may be out of line with develop- ments in the "younger" Latin American populations. Countries with young funded pension systems may be better off using data from countries with older systems, such as Chile, although there is a need to modernize the mortality tables even in Chile because they date from the 1970s. Latin American governments should also consider more generally whether there are other financial instruments available that would be suit- able for retirement savings purposes. In principle, well-regulated financial systems can offer suitable alternative retirement products, such as mutual funds and insurance policies. Mutual funds are in fact likely to come into close competition with pension funds in Chile because the nature of the investment products they offer is similar. Indeed, in many ways Latin America's new mandatory pension funds are simply a special type of mutual fund.12 Life insurance companies are also in a position to offer products over the accumulation stage that are attractive for more risk-averse individuals. In Chile insurance companies already market a life insurance policy in the new voluntary pension system (the so-called seguro de vida con ahorro, or life insurance with savings) with a savings component that can be covered by a real return guarantee. Such a guarantee can be offered because there is a liquid, long-term, inflation-indexed bond market that insurance com- panies can tap to hedge their liabilities. Chilean insurance companies al- ready rely on these bonds to hedge liabilities arising from the underwrit- ing of annuities contracts. The life insurance industry in Argentina and Peru offer similar (if admittedly less sophisticated) products that combine insurance and savings features. The long-term guaranteed rate of return offered by insurance companies in Chile may be superior to the bond portfolio offered in an individual pen- sion fund account for many workers because the companies can tie down a real rate of return on contributions over a long investment horizon. A portfolio of bonds of different maturities is subject to reinvestment risk and may therefore be less attractive for risk-averse investors. Nonetheless, insurance companies are not able to offer interest rate guarantees over a period longer than a few years in other Latin American countries because of their higher interest rate volatility and lack of long-dated instruments. The only countries that resemble Chile in the extent of capital markets development are Brazil and Mexico. Both of these countries have a more developed equity market that accounts for more than 90 percent of all stocks traded in the region, although their debt markets are still concentrated FACILITATING CONSUMPTION SMOOTHING 247 in shorter maturities and have not developed as liquid, inflation-indexed securities markets. Hence products offering investment guarantees are still some time away for even these relatively developed Latin American countries. The problem with permitting choice over savings products in the manda- tory system arises from the difficulty of comparing different fee structures. If Chile and eventually other Latin American countries were to open up their mandatory pension system to other financial products, they should first en- sure that commissions structures are comparable, perhaps by permitting only commissions based on accumulated assets. In addition, many Latin American countries have not reinforced the regulatory and supervisory framework for mutual funds and insurance companies. It would be unwise to open up the mandatory system to these financial institutions in their current state. Of course, if all the resources that have been dedicated to es- tablish and supervise the new private second pillar effectively had instead simply been used instead to improve the regulation and supervision of other financial institutions, these conclusions may have been different. Improving the Voluntary Savings Pillar: Increasing Options and Incentives Ideally, in an enlarged voluntary retirement savings pillar the decisions of how much to invest and in what instrument to invest should be left to the individual. To avoid distortions to inter-temporal consumption choices, all suitable retirement savings products should be subject to the same expenditure tax.13 To maximize the extent of flexibility of the system workers should be allowed to carry forward unused tax deductions to later years. There are two main drawbacks with this liberal model of voluntary retirement savings. The first is that individuals may not have a sufficient level of financial literacy to make adequate investment choices. Even if they do the extent of differentiation among financial products may make it too complicated or time consuming for the average worker to compare fees and performance across savings vehicles. The liberal model of voluntary savings, therefore, calls at the very least for a regulatory framework that ensures the transparency of the cost and benefit of different financial products. Another real-world drawback with this ideal model is that richer workers often take most advantage of the tax incentives offered by governments to encourage retirement saving to reduce their tax liability. This problem can be significantly ameliorated by offering instead a public subsidy to work- ers who contribute to a savings plan, as is done in the Mexican manda- tory funded pillar and in the mandatory pillar for self-employed workers in the Dominican Republic.14 Such a subsidy would probably be neces- sary in Latin American countries where a large section of the population 248 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA (sometimes more than half) are not subject to income taxes. The progres- sive impact of the subsidy can be ensured by fixing it in absolute amounts (as in Mexico) and by setting an upper ceiling on taxable earn- ings for subsidy eligibility. In all Latin American countries (except Chile since 2002 and Brazil) the regulation and taxation of voluntary pension savings are far from this liberal model. Chile has recently extended the preferential treatment of voluntary retirement savings to financial products offered outside the AFP industry (see box 10.1). In addition to bank deposits (which provide a less volatile return than pension funds), mutual funds and insurance policies are competing as voluntary pension savings products in Chile since the beginning of 2002. Box 10.1 The Big-Bang Approach to Voluntary Pension Savings Reform in Chile Since March 2002 Chilean workers have been able to save up to 50 UFs1 of their monthly pre-tax income in any voluntary pension plan authorized by the securities and insurance regulator, as well as in voluntary savings accounts managed by the private second-pillar AFPs. There are no restrictions on the number of plans or AFPs in which workers may deposit their voluntary savings. Workers also may cash out these plans at any time before retirement, subject to a 10 percent special excise tax (in addition to the relevant income tax). Employer contributions, however, may only be liquidated at retirement. The development and performance of the voluntary market in the coming years will be followed closely by Chilean policymakers, who may use the voluntary pensions market as a testing ground for a possible reform of the mandatory system. Two of the most complicated issues in the operation of the voluntary pensions market will be the transfers between plans and providers and the commissions charged. Although the law does not envisage any limits in switching among plans, it does regulate the commission structure firmly. AFPs may only charge a fixed commission for collecting voluntary contributions and transferring them to the plan chosen by the worker. This fixed commis- sion must be the same regardless of the plan chosen. AFPs and voluntary pension plan providers may not charge a commission, however, for partial or full transfers of the accumulated balance in the individual's voluntary account to another AFP or to a different voluntary plan. The voluntary pension plan providers and the AFPs may charge commissions for fund management based on the stock of accumulated assets. Note: Unidades de fomento are inflation-indexed monetary units used for all financial transactions in Chile. FACILITATING CONSUMPTION SMOOTHING 249 Prior to the 2002 reform in Chile the only option available to individuals wishing to benefit from tax incentives was to park their voluntary savings up to retirement in the mandatory, illiquid AFP accounts. Other savings vehicles, including the more liquid Cuenta 2 offered by the pension fund ad- ministrators, were not as tax advantaged. The Cuenta 2 itself, however, was less popular than other financial products even though there were no addi- tional commissions and it had a somewhat more attractive tax treatment than mutual funds and savings products offered by life insurance companies before March 2002. Probably the only negative feature of the Cuenta 2 is that withdrawals are only permitted twice a year. Despite its benefits there has been a net withdrawal of voluntary savings from the AFP Cuenta 2 over the last two years, while there has been sustained growth in premiums for savings products offered by life insurance companies (see figure 10.2). Prior to Chile's reform in March 2002, Brazil was the Latin American exception in permitting a high degree of choice for tax-advantaged, Figure 10.2 Savings Products Offered by Insurers Are More Popular Than the AFPs' Liquid Cuenta 2 (December 2000 and December 2001) (Monthly net flows) 16 14 12 10 8 millions 6 4 US$ 2 0 -2 -4 2000 2001 Premiums for savings products offered Cuenta 2 net deposits by insurers Source: Securities and Insurance Superintendency and Superintendent of AFPs, Chile. 250 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA long-term savings. Its mutual fund industry is highly developed and offers both occupational and personal pension plans (see box 10.2). The mutual funds are in fact at the center of the new voluntary personal pension plans that have been instituted since the mid-1990s. Life insurance companies also participate actively as providers of personal pension plans and can sell retirement savings products to companies. High contributions to the mandatory pension system (both the PAYG and the funded second pillar), the illiquidity of voluntary contributions to Box 10.2 The Role of the Financial System in Brazil's Voluntary Pension Savings Plans The voluntary pensions system in Brazil consist of two distinct sectors-- the occupational and the personal sectors. Although occupational plans still dominate the pensions landscape, accounting for the vast majority of members and assets, their growth has been disappointing over the last two decades. Personal pension plans, however, are expanding at a rapid pace (contributions to personal plans grew at an average annual rate of 38 percent between 1994 and 2001). Occupational plans are established by employers and have been tradi- tionally of the defined benefit variety, financed through the establishment of pension funds. The importance of mutual funds and professional asset man- agers in this system cannot be overstated. Mutual funds account for one-half of all assets held by occupational pension funds. Professional asset managers handle more than one-quarter of all occupational pension assets. Three main types of personal plans exist--the traditional pension plans, the personal pension plans without benefit guarantees (plano gerador de beneficio livres, PGBLs), and the individual programmed pension funds (fundo de aposentadoria programada individual, FAPIs). All receive a favorable tax treatment (deductibility up to 12 percent of salary) but have different liquidity requirements. Although traditional personal plans and PGBLs must be kept until retirement, investments in FAPIs must be kept for at least 10 years to receive tax benefits. If savings are withdrawn before that time a penalty is imposed. Both the traditional plans and the PGBLs are administered by insur- ance companies or specialized providers. The traditional plans are essen- tially deferred annuity contracts. They offer guaranteed real rates of return that may become unaffordable for insurance companies as soon as macro- economic conditions improve and interest rates fall. On the other hand, the more recent PGBLs are unit-linked products in which contributions are invested in mutual funds. At retirement the accumulated balance must be transformed into an annuity. FAPIs are managed by banks but must also offer a set of investment options consisting largely of mutual funds. FACILITATING CONSUMPTION SMOOTHING 251 Box 10.2 (continued) In addition to the PGBL and the FAPI, a new generation of products combining life insurance and savings has recently gained approval by the insurance supervisor. Like its Chilean equivalent (seguro de vida con ahorro, or life insurance with savings), the life insurance plan without benefit guarantees (vida gerador de beneficio livre, VGPL) offers a lump- sum payment in case of death during the accumulation stage and benefits in the form of annuities when the policyholder retires. The benefits depend on the funds accumulated at the time of the claim. Given the rich variety of products and financial service providers, to the extent that the government of Brazil considers a partial shift from PAYG financing to funding for income replacement in old age, it would rely on its existing financial institutions and products rather than establish new ones (as other countries in the region have done). Source: Gill, Packard, Schwarz, and Yermo (2001). the pension funds, and limited tax advantages for saving in other retirement products may explain in part why saving in Latin America is low. Econo- mists still dispute whether the net effect of structural pension reforms has been positive to household savings, and in particular whether mandatory savings "crowd out" or "crowd in" voluntary savings (Schmidtt-Hebbel 1998a). Even in Chile, the only country in Latin America where total con- tribution rates to the pension system fell after the reform, household sav- ings in 2000 were at only a slightly higher level than in 1987. There is little doubt that by excluding other financial products from the benefits of both the mandatory and the tax-advantaged voluntary retire- ment savings pillars, their development has been handicapped. Saving in financial instruments such as mutual funds and life insurance (other than the mandatory insurance for death and disability) is much lower than pension fund saving. Except in Brazil, employer-sponsored retirement programs are also rare, covering only some public sector plans and a few multinational corporations. Nonetheless, opening up the sector to compe- tition from alternative products and providers would require a tremen- dous regulatory effort to ensure the transparency of the cost and benefits of different products and adequate supervision of their providers. Do Mutual Funds Have a Role in the Voluntary Savings Pillar? Savings in mutual funds was less than 10 percent of GDP in all Latin American countries that had introduced a mandatory private pension pillar as of 2001. In contrast, Brazil's mutual funds have experienced 252 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 10.3 Mutual Funds Have Grown Significantly Only in Brazil, Where Pension Funds Are Voluntary 35 1998 2001 30 25 20 Percent15 10 5 0 Argentina Brazil Chile Mexico Peru Source: Investment Company Institute; Association of Mutual Fund Administrators of Peru. tremendous growth in the past decade. As shown in figure 10.3, mutual fund assets grew from 15 percent to nearly 30 percent of GDP between 1998 and 2001. This growth can be largely accounted for by pension funds and other corporate investors who invest in mutual funds because they are not subject to the tax on financial transactions. Up to one-half of pension fund assets were invested in mutual funds in December 2001, whereas pension funds and other corporate investors accounted for more than one-half of mutual fund balances. The limited development of mutual funds in Latin American countries other than Brazil offers some interesting insights on the implications of introducing a mandatory private pension system for the development of a voluntary savings pillar. The new second-pillar pension funds and mutual funds have a lot in common. They both operate on the principle of indi- vidual ownership of a quota of a fund that is invested in financial instru- ments by a company that is dedicated exclusively to that purpose. The few differences between the two industries are in fact driven by regulatory requirements, which tend to be more onerous for the new second-pillar FACILITATING CONSUMPTION SMOOTHING 253 pension funds. Although individual investors face no restrictions in the type and number of mutual funds in which they can invest, mandatory saving in second-pillar pension funds is highly regulated. Workers may save in only one fund (except in Chile and soon in Peru where five funds are available), which is subject to strict investment and performance regu- lations. Of course, the prohibition against accessing savings until retire- ment also affects the functioning of the pension fund industry. Pension fund administrators have benefited enormously from this captive market, which has helped boost their profitability. The obligation to save through pension funds has set a lower boundary on the potential saving in mutual funds because it has reduced the dispos- able income of workers.15 Had workers been free to place their mandatory savings in different financial products, there is little doubt that mutual fund growth would have been more pronounced. Even in the voluntary savings market, however, the mutual fund industry has been placed at a disadvantage relative to pension funds. First, mutual fund savings have been subject to a less attractive tax treat- ment than voluntary contributions to the pension fund system. This asym- metry in tax treatment exists because voluntary contributions to the pension funds cannot be cashed out before retirement in most Latin American coun- tries. Even in those countries where they can be cashed out (Chile, Costa Rica, Mexico, and Peru) there is a maximum frequency for distributions.16 A second bias against mutual funds is that governments have made less of an effort at regulating and supervising them than they have at oversee- ing new pension funds. The disclosure of mutual fund performance and commissions in particular is far less transparent than that of pension funds. In most Latin American countries mutual funds can charge commissions on entry and exit, as well as an annual management charge (set normally as a percentage of assets under management). Competition in the mutual fund market of countries such as Argentina, Chile, or Mexico does not take place via commissions and performance. Instead, the main determinant of the mutual fund administrators' profitability is access to a distribution channel. In Chile, for example, mutual funds sold through bank branches by administrators tend to have a higher profitability than those sold by in- dependent administrators. The Chilean mutual fund industry has also been hampered in the past by regulations requiring separate administrators for open-ended, closed-ended, and real estate mutual funds. This artificial separation was lifted during the capital markets reform in 2002. Investment regulations for mutual funds at times have verged on the side of imprudence, failing to address conflicts of interest effectively. In Mexico, for example, mutual funds can invest freely in assets of the administrator's parent company. In the past mutual funds have been used by the main Mexican financial groups to obtain financing for their opera- tions. Even today such investments are permitted. Latin American pension funds, on the other hand, are subject to an investment limit of 5 percent 254 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA (less in some countries) on assets of companies related to the pension fund administrator. These policy decisions (tax treatment and regulatory framework) have further weakened the attractiveness of mutual funds relative to pension funds and probably explain the slow development of mutual funds in Spanish-speaking Latin America relative to Brazil. The mutual fund industry has not been able to reap the benefit from economies of scale present in financial markets as have the second-pillar pension funds, and it has been subject to less scrutiny of its market practices. As a result com- petitive forces have been weak in driving commissions lower. The fees charged by open-ended mutual funds are certainly much higher than those charged by pension funds. Maturana and Walker (1999) showed that average commissions (including entry, exit, and annual man- agement fees) for the Chilean mutual fund industry represented 3.1 percent of total assets managed in 1996 (down from 3.8 percent in 1990). Pension fund commissions are significantly lower, verging currently toward a level equivalent to approximately 0.7 percent of assets under management.17 The commissions charged by equity mutual funds in Chile also seem high when compared with other countries. By the end of 2001 annual manage- ment fees for equity mutual funds in Chile were 5 percent of assets under management, whereas in Brazil they were only 3 percent. The evolution of annual fund management commissions charged for the three main types of mutual funds during the 1990s is shown in figure 10.4. It is noteworthy that commissions charged by equity funds have re- mained stubbornly high despite the sustained growth in total assets under management. This suggests that the problem of high commissions is not just the result of economies of scale but that there are barriers to entry in the industry that allow players to keep commissions high. A priori, however, it would seem that operational costs and hence economies of scale may account partly for the higher fees charged by Chilean mutual funds relative to second-pillar pension funds. The operating expenses of Chilean pension fund administrators equaled 0.5 percent of assets in 2001, whereas the figure for mutual fund administrators was much higher (0.9 percent). As shown in figure 10.5, however, economies of scale operate largely at low levels of assets under management (less than 100 billion pesos) and thereafter they stabilize rapidly. Indeed, the two mutual fund administra- tors that had cost-to-asset ratios similar to those of the pension fund administrators (0.7 percent) managed less than 150 billion pesos. Another important factor that may explain the higher fees for mutual funds is the lower transparency of fee structures and greater extent of product differ- entiation. That there are 20 mutual fund administrators but only 5 pension fund administrators in Chile despite the much larger size of the pension fund industry, demonstrates that it may be easier to create a competitive advantage in the mutual fund industry. FACILITATING CONSUMPTION SMOOTHING 255 Figure 10.4 Fees for Equity Mutual Funds in Chile Have Remained Stubbornly High 7 6 assets 5 of 4 percent 3 a as 2 Fees 1 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Money market fund Bond fund Equity fund Source: Maturana and Walker 1999; Securities and Insurance Superintendency. Looking ahead, what role should the mutual fund industry play in Latin America's reformed pension systems? The similarities in the opera- tion of second-pillar pension funds and mutual funds raise doubts about the efficiency of maintaining separate administrators for each type of fund and supervising the two systems under different regulatory authorities. In- deed, to the extent that a mandatory savings system requires additional prudential regulations (such as performance rules, and quantitative invest- ment regulations), there is no reason why regulators could not apply them also to mutual fund administrators. Maintaining separate administration and supervision for pension funds and mutual funds is particularly costly in poorer countries where skilled labor is in short supply in both public and private sectors. In many Latin American countries, however, the mutual fund industry is not ready to play a central role in retirement savings. Transparency and conflicts of interest with related financial groups are rife, and com- missions seem to be even higher than those charged by second-pillar pen- sion fund administrators. Before policymakers consider opening up the voluntary pension pillar to mutual funds they should ensure that the in- dustry is ready to compete with second-pillar pension funds as the agent 256 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 10.5 Economies of Scale in Fund Management Kick in at Low Asset Levels 8.0 7.0 6.0 under 5.0 4.0 ratio/assets management 3.0 2.0 Expense 1.0 0 0 100 200 300 400 500 600 700 800 Assets under management (billion pesos) Source: Securities and Insurance Superintendency, Chile; authors' calculations. of workers in channeling their pension contributions into productive investments. Is There a Role for Life Insurance Providers in the Voluntary Savings Pillar? Unlike mutual funds, the life insurance industry has not been excluded from the new mandatory private pension pillars in Latin America. On the contrary, death and disability insurance is also mandatory and managed by life insurance companies in all countries that have reformed their pen- sion systems (except Mexico and Costa Rica, where the social security institute administers these benefits). Life insurance companies also play a central role in the retirement stage of the new systems selling annuities. They are not allowed, however, to manage funds from forced savings dur- ing the capital accumulation phase before retirement. Some savings products sold by insurance companies could have been used, like mutual funds, as vehicles for long-term saving. Life insurance FACILITATING CONSUMPTION SMOOTHING 257 companies in Latin America often provide policies with savings accounts. In some cases, as in Chile, the rate of return on these accounts is protected by absolute return guarantees, which are attractive for risk-averse, long- term investors. In the context of a volatile interest rate environment, how- ever, such guarantees may only be affordable over relatively short periods. Insurance companies in some OECD countries have experienced severe difficulties in recent years as a result of guarantees offered during the 1970s when interest rates were at much higher levels. The availability of inflation-indexed government bonds would ease asset liability manage- ment by insurance companies because the interest rate volatility of real bonds tends to be much lower than that of nominal bonds. Exclusion from the accumulation side of the mandatory pension pil- lars and a less advantageous tax treatment than voluntary contributions to the second-pillar pension funds have handicapped the development of the market for savings policies sold by life insurance firms. Questions re- main even now about the readiness of the industry to become a viable competitor with the second-pillar pension fund administrators. Histori- cally the supervision of insurance companies has been less effective than that of second-pillar pension funds. One continuing problem is the lack of separation of life insurance and non­life insurance operations, which creates risks for potential savers in life insurance products. This is prob- ably more of a problem for investors in small countries close to the Andean Mountains and in the Caribbean region where risk of natural ca- tastrophe can be high. The licensing process is also deficient in several Latin American countries. Specifically, some countries do not require the submission of a business plan or they require a feasibility study that covers only certain aspects of the insurance operation. The work of the supervisor is made more difficult by the fact that the appointment of ac- tuaries is not obligatory. The consequences of the weakness of the regulatory and supervisory framework are evident in the number of insurance company bankruptcies over the past few years. The OECD (2001) reported a total of 57 bank- ruptcies in seven Latin American countries between 1996 and 1998, distributed as shown in figure 10.6. Only in Mexico and Venezuela were policyholders spared financial losses. In Argentina and Colombia losses were limited by the existence of policyholders' protection funds (only for retirement insurance and workers' compensation). Despite its turbulent past the insurance industry has recently begun a slow process of modernization in many Latin American countries. Part of the impetus is provided by structural pension reforms, because insurers play a central role in providing coverage against the risks of death and disability before retirement--and can sell annuities during the retirement stage. Chile, the country with the longest-lived private second-pillar pension system, was understandably also the first to tackle some of the problems in the insurance sector. 258 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Figure 10.6 Insurance Company Bankruptcy Is a Threat to Policyholders Number of compulsory terminations or liquidations, 1996­98 30 25 20 bankruptcies 15 of 10 Number 5 0 Argentina Bolivia Brazil Colombia El Salvador Mexico Paraguay Venezuela Source: OECD (2001). Are There Options to Improve the Voluntary Funded System? The evidence discussed in the previous section raises two main policy ques- tions for Latin American governments that have undertaken reforms of their pension systems: 1. Does the limited liquidity for accumulated funds in the tax-advantaged voluntary retirement savings system of Latin America help or hinder the growth of retirement savings? 2. Are other financial products (such as mutual funds and other savings products managed by financial institutions) attractive as vehicles for long- term saving in the tax-advantaged voluntary pillar? Although part of the reason for the lack of interest in voluntary saving in pension funds may be lack of public trust in institutions linked to the social security system, the illiquidity of such savings would seem to be a more dissuasive factor. The reform of voluntary pension savings in Chile FACILITATING CONSUMPTION SMOOTHING 259 demonstrates the difficult challenge faced by policymakers when designing a voluntary pillar. Until March 2002 workers wishing to make additional savings for retirement while benefiting from the same tax treatment as the mandatory contributions could deposit their contributions only in the mandatory pension fund accounts (Cuenta 1). Furthermore, workers were able to make voluntary contributions to separate accounts managed by the pen- sion fund administrators, the so-called Cuenta 2 (or voluntary savings ac- count). As many as four withdrawals from these accounts were permitted in a year. The tax treatment of Cuenta 2 contributions, however, was not as attractive as that of voluntary contributions to the mandatory ac- counts. Despite the less advantageous tax treatment of the Cuenta 2, its liquidity made it the preferred choice for most Chilean workers. As of December 2001 there were more than 1 million Cuenta 2 accounts, but only 155,000 mandatory Cuenta 1 accounts had received voluntary con- tributions. Only the richer tax-paying individuals have found it attractive to place more of their voluntary savings in the mandatory account rather than the Cuenta 2. This explains why the total funds accumulated in the Cuenta 2 were approximately half of those in the mandatory accounts (US$213 million versus US$453 million in September 2002). Since March 2002 Chilean workers have been able to place the volun- tary savings that were originally destined for the mandatory account in a variety of financial products. In addition, the requirement to maintain the balance up to retirement has been eliminated and workers can withdraw as much of their funds as they wish at any time, subject to tax penalties. As a result, voluntary contributions have increased dramatically. Accord- ing to Chile's superintendent of AFPs, voluntary contributions to the mandatory pension fund accounts over the period March to September 2002 were 14.7 percent above those of the same period in the previous year. Reform of the voluntary savings market in Chile is also expected to give a huge boost to the mutual fund and life insurance industries. Pre- miums paid for voluntary retirement savings products sold by life insurance companies between March and September 2002 equaled US$5.7 million, a sizable amount but significantly less than the volun- tary deposits made into the mandatory pension fund accounts over the same period (US$59 million). The boost to voluntary savings since the 2002 reform calls into question the logic of the prohibition against withdrawal of funds before retirement to the extent that it dissuades workers from participating in the pension system. It would seem that, at least for voluntary retirement savings, expecting workers to accept parting with their assets until retirement is not very reasonable. The Chilean evidence clearly shows that voluntary retirement savings can be increased by allowing greater access to the ac- cumulated funds before retirement. Tax benefits can be clawed back if 260 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA funds are cashed out before retirement, thereby creating an incentive for long-term saving. Such mechanisms have been successfully implemented in many OECD countries and could be explored in Latin America. As to the role of mutual funds and other savings products sold by financial institutions in the pension system, the Chilean evidence clearly shows that these products can also be attractive for workers. The fact that life insurance companies have captured 10 percent of the second-pillar pension funds' market is a clear indication of the competitive pressure to which the second-pillar funds are now subject. At the same time, however, it is not clear whether the appeal of mutual funds and other savings products is the result of an inherent superiority in performance terms or whether their popularity is simply a result of market- ing efforts. In particular, mutual funds in Chile benefit from an extensive bank-branch network through which these products are sold. Insurance companies, meanwhile, can use the same sales agents and brokers to sell both insurance policies and voluntary savings products. Indeed, the savings prod- ucts authorized for life insurance companies carry a life insurance policy. The lack of transparency in the commission structure of both mutual funds and life insurance companies is certainly a cause of concern and would caution against a rushed opening of the voluntary retirement savings system to product choice. The past experience of high commissions in the mandatory private second pillar is likely to be repeated in the context of a voluntary savings market where commissions on pension funds are not directly comparable with those on mutual funds and on life insurance companies. Pension fund commissions are applied exclusively to contri- butions, but mutual fund commissions are primarily applied to the accu- mulated assets. The complexity of some insurance products and their opaque commission structure do not bode well for a population with low levels of financial literacy. Extending choice over different financial instruments under these conditions can be a recipe for confusion and ill-informed decisions. Governments bear less fiduciary responsibility in voluntary retirement sys- tems--but to the extent that individual choice among different products is permitted, governments may be blamed for inadequate choices and may be called on to provide compensation for unfortunate workers. The first task, therefore, is to improve the regulation of these financial products and their providers. A useful reference in this respect will be the Mexican experience with fee calculators that were introduced by the supervisory authority in 2002 to make it possible to compare fees among second-pillar pension fund administrators.18 New regulations require pension fund members to sign a form that states what effect switching administrators will have on their account balance as a result of different commissions. The pension fund administrators are now also required to inform plan members about any increase in fees. Although it is yet to be seen how effective these policies FACILITATING CONSUMPTION SMOOTHING 261 are at increasing investors' awareness of the relative impact of different fee structures, a simpler and more effective solution would seem to be the one implemented in Chile's new voluntary pension pillar, where only asset- based charges are permitted. Conclusion The reform proposals that have been put forward in this chapter have con- sidered how to improve the management of investment and longevity risk and reduce administrative costs of consumption smoothing in the savings pillar, assuming that the primary goal of poverty prevention is met through a separate pillar. As readers will recall from chapter 7, the other main risks that individuals bear in funded systems are policy and agency risk. We did not dwell exten- sively on the latter because governments generally have been successful at ensuring the proper functioning of the new second-pillar pension funds. Fortunately fraud and conflicts of interest are issues that have not yet sur- faced on a large scale in countries where the new private pillars were in- troduced with high standards of disclosure and transparency. Regulation and supervision in other parts of the financial system, however, have yet to achieve the same level of rigor as that of the second- pillar pension fund industry. The mutual fund industry is a case in point. To the extent that greater space is created for voluntary savings and free choice of product and provider is introduced, there will be an even greater urgency to improve the regulation of the financial system. Countries with limited resources should also consider the benefits of consolidated super- visory structures through which the scarce professional staff can be more effectively used in a variety of financial service markets. To the extent that individual choice of savings products and providers is permitted, regulations will also be needed to ensure the comparability of fees and performance. The more choices that people face, the greater the need for access to clear, independent, and timely information about retirement products. Financial literacy programs should be given the highest priority at both government and employer levels. The Chilean experience with the five funds in the mandatory private pillar and with the liberalization of the voluntary savings pillar will be a good testing ground for these proposals. Peru's experiment with offering greater investment choice to affiliates of the mandatory private pillar also can provide valuable lessons. The approaches taken should be evaluative rather than dogmatic. It is clear that neither the poverty prevention nor the income-smoothing pillars can be completely insulated from the dangers posed by policy risk. Poverty prevention instruments require a government-mediated transfer of resources among workers. Hence, policy risk will always be present in 262 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA such plans, especially when governments are unable to maintain fiscal discipline and macroeconomic stability. Funded, privately managed programs, as the Argentine example clearly shows, can also easily fall prey to cash-strapped governments. Just as the repayment promises of a fiscally profligate household are not taken seriously, the promise of old-age income security by a fiscally imprudent government will not be credible. Fiscal prudence is therefore essential to ensure the population's confidence in both the PAYG and funded pillars of the pension system. Countries that require retirement savings to be mainly invested in government bonds with a high default risk are not in a good position to mandate savings. For such countries policy risk is likely to be overwhelming for the new funded system, as was illustrated by Argentina in 2001. Yet eliminating the mandate is hardly the best solution, even when a basic PAYG pillar still exists. In Argentina the sale of even a small part of the pen- sion funds' government bond portfolio could send waves of panic through the market and interest rates may rise, further damaging the country's frag- ile economy. This option is not even currently open to the pension funds, as the government debt they hold is in the form of nontradable loans. On the other hand, gradually reducing both the ceiling on earnings subject to the mandate and contribution rates can serve as a disciplining device for the government, which will not be able to make promises it cannot keep. In general, governments should focus their attention on poverty pre- vention and prudent macroeconomic management. Mandating investment in government bonds during the initial years of the new system can be a useful vehicle to ease the fiscal cost of structural reforms and to ensure a stable performance of the new second-pillar pension funds, but only as long as governments have previously built a track record of prudent macroeconomic policy. The Chilean government significantly strength- ened its primary fiscal surplus prior to the pension reform in 1981 and was therefore in a position to limit pension fund investment to investment- grade bonds. As the transition costs decreased over time, investment re- strictions were gradually lifted. Argentina presents the exact opposite sce- nario. The fiscal deficit of the Argentine government deteriorated significantly in the years following structural pension reform, and that forced the government to rely increasingly on the second-pillar pension funds as a source of captive financing for the ballooning debt. There is a similar risk in other countries, such as Bolivia. The main lesson from these diverse experiences is that a significant fiscal tightening is a basic precondition for the success of structural pension reforms, especially for governments that start from a fragile fiscal stance. Having already reformed its system, Argentina is not in a position to start afresh. To guarantee the sustainability of its retirement system, however, the government's first step should still be to put its internal finances in order. It could then turn the current inflation-indexed loans that it extended to the pension funds into tradable government bonds of increasing maturity. FACILITATING CONSUMPTION SMOOTHING 263 As long as the fiscal situation is under control, such bonds could help strengthen interest in long-term saving. In Latin America only Chile has managed to develop a liquid market of investment-grade, long-term, inflation-indexed government bonds. Chile is the only country in the region where long-standing fiscal discipline has successfully raised government bonds to investment grade. Thanks to the development of the government debt market, Chilean insurance compa- nies are able to offer performance guarantees and annuities at competitive prices. Second-pillar pension funds also find Chilean government bonds an attractive and safe source of investment income. At the very least, countries insisting on a retirement security model inspired by the Chilean experience should remain cognizant of these developments. The persisting differences between Chile and these countries over the past decade may even provide cause for reflecting on whether such an approach is suited for all the countries in the region. The importance of providing these types of sound financial instruments cannot be overestimated. The shift from a defined benefit to a defined con- tribution system in Latin America involved not just a fair and welcome transfer of cohort-specific longevity risk to individuals, but also laid all investment risk on workers. Prudent governments, however, are in a posi- tion to offer long-term investment guarantees that protect individuals from improvidence and governments from their own myopia. Moving forward, policymakers will need to reconsider the design of their savings pillar. Some of the reforms proposed in this chapter hang on the ability of governments to regulate markets and promote consumer protection. Some basic questions need to be answered: (a) Can the effi- ciency of centralized record keeping and account management be spared from political manipulation? (b) Can governments ensure as effective su- pervision for mutual fund administrators and other financial providers as they do for pension fund administrators? (c) Can they ensure transparency in performance and fee structures across products and providers? (d) Can conflicts of interest in employer pension plans be adequately addressed? (e) Is the level of financial literacy of the population sufficient to permit in- troducing individual choice of investment portfolio or even financial prod- uct? (f) Can limits on pension fund investment abroad be relaxed without endangering macroeconomic stability? and (g) Is it politically acceptable to introduce age- and earnings-specific contribution rates and to reduce max- imum earnings subject to the savings mandate? Notes 1. There is no reason why low-income workers should be prohibited from making contributions to the funded pillar, especially if the savings product offers better performance than other financial instruments available in the financial system. 264 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA 2. Uruguay is the only country that somewhat conforms with this arrangement because poorer workers are not obliged to contribute to the funded system but are covered by a public defined benefit pillar. However this pillar offers earnings-related pensions (the replacement rate is about 50 percent) rather than just a level of income sufficient to avoid old-age poverty. 3. Valdés-Prieto (2002c) has also suggested introducing age-related contribu- tions. 4. As argued earlier and in chapter 9, the poverty prevention objective is best served through a separate, PAYG-financed pillar. 5. The floor for mandatory contributions to the savings pillar would depend on the poverty alleviation target of the first pillar. This is usually set in relation to the minimum or average wage. 6. The experience with occupationally defined benefit plans in Brazil--the country with the largest occupational system in the region--has hardly been a suc- cess. Despite their relative maturity (some have existed for more than 40 years), these plans suffer from chronic underfunding and offer little protection to benefi- ciaries in terms of vesting and portability rules. The Brazilian government has been trying to revamp the regulatory and supervisory system over the last few years and some improvements are starting to be seen. Much work remains to be done, however, before the Brazilian system can be a reference model for other Latin American countries. 7. The voluntary defined contribution occupational plan system in the United States is also worthy of analysis, although possibly as an example not to be fol- lowed. The 401(k) plan in the United States may have been heralded as a success and an international model to be commended in Latin America, except for the lack of limits on investment in plan sponsor assets. The risks arising from investment in shares of the sponsoring employer to the extent observed in some U.S. companies may be bearable for relatively wealthier American workers, but are certainly not advisable for plans that substitute for a public social security system, or indeed for the poorer Latin American population. 8. In fact, governments sometimes do the opposite. In Chile, for example, the pension contributions of self-employed workers are not tax deductible. 9. The range of choice available in Sweden (more than 600 mutual funds) may not be suitable for Latin America, given the perceived need for ongoing, expensive supervision and the level of financial literacy of the population. A choice between three or five types of funds, as in Chile, may be more appropriate. 10. It should be noted, however, that despite the extent of centralization of administrative functions, Swedish workers have much more investment freedom than Latin American ones during the accumulation stage. Hence it may be argued that the Swedish system is more protected against political risk than the Latin American one, where investments are decided by investment regulations and the pension funds' portfolios resemble one another. 11. The three-fund classification matches that of mutual funds in Chile. The Mexican pension legislation also permits multiple funds, but this has not been put into practice yet. 12. To be precise, the Latin American pension funds resemble open-ended mutual funds, which stand ready to redeem shares or units of the fund at its net asset value at any time. 13. An expenditure tax enables individuals to receive investment income gross of tax and is therefore neutral to the choice between consumption now and in the future. It can be of two forms: EET (exempt contributions and investment income but tax benefits), or TEE (tax contributions but exempt investment income and benefits). Individuals typically prefer EET because tax rates are lower in old age and because there is a risk in a TEE environment that a new government would FACILITATING CONSUMPTION SMOOTHING 265 decide to impose taxes on benefits. In most Latin American countries, except Mexico and Peru, voluntary retirement savings in the pension funds are subject to the EET taxation system, but savings in other financial products are subject to a less attrac- tive tax treatment. Moreover, voluntary pension fund savings are illiquid and must be kept until retirement. In Peru voluntary savings in the pension funds are subject to a TEE system (as are mandatory retirement savings), whereas in Mexico there is double taxation. In both countries voluntary savings can be withdrawn before retirement. 14. A similar subsidy is offered to contributing low-income workers in the Czech Republic (Vittas 2002) and in Germany's Riester pensions. 15. However, some of the mandatory savings have passed through to the mutual fund sector because pension funds can invest up to a certain ceiling in mutual funds (the ceiling varies between 0 and 15 percent depending on the country). As of December 2001, however, only Argentina (3.3 percent of total assets), Chile (2.6 percent), and Peru (0.5 percent) recorded any pension fund investment in mutual funds. 16. As mentioned already, Chile was the first country to liberalize the voluntary savings market (March 2002), thereby eliminating the restriction on withdrawals (although these withdrawals are now penalized) and permitting workers to place their voluntary savings in a variety of financial instruments (including mutual funds). 17. Pension fund commissions are set on contributions, not assets. A 0.7 percent commission on assets is approximately equivalent to a 15 percent charge of contributions during a worker's career, which is the current level of commissions in Chile. 18. Mexico is the only country in Latin America that permits different com- mission structures in the mandatory pension pillar. 11 The Way Forward THE ISSUE OF PRIVATIZATION FIGURES prominently in the quest for a new development paradigm triggered by economic failures in the developing world. Privatization of government-produced goods and services was, for example, one of the 10 key measures in the erstwhile "Washington Consensus" (see Williamson 1999). In the Latin American region, for which this consensus of policy prescriptions was believed to best apply, structural reforms to social security may well have been the most impor- tant wave of privatization of government services. The region has the longest and richest history of experience in privatization of social security, starting with the reform pioneered by Chile in 1981. Some variant of the Chilean model of social security has been adopted by many of Chile's neighbors in the region, and it has been a serious contender as a reform model in other developing countries and even in some industrial countries. Today it is even considered by some as an alternative to the current U.S. social security system, despite the latter being perhaps one of the best- managed traditional public pension systems in the world. Why This Survey Now? Disappointing economic growth rates, persistently high macroeconomic volatility, and increasing concerns regarding income distribution have led to a reevaluation of the Washington Consensus model in Latin America since the late 1990s. There is growing sentiment that although the policies that composed the original consensus are sound, they are insufficient to address institutional shortcomings in the region. In particular, critics of the Consensus have called for stronger competition policy and finan- cial market regulation to improve the outcomes of privatization and for more targeted efforts at poverty reduction to improve equity. The assessment of regional pension reforms presented in this volume takes place in the context of this larger debate on the course of Latin America's development. 267 268 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA It should not come as a surprise then that the Latin American model of pension reform is being scrutinized--even reconsidered--in countries within the region. The principal concern is the alleged failure of the reforms to increase the coverage of people by social security systems, which was considered an important selling point of reforms when they were initiated. But there are other concerns as well, some of which may reflect a lack of appreciation of the aspects in which the reforms have paid dividends. This book was commissioned by the chief economist of the World Bank's Latin America and the Caribbean Region to provide a balanced assessment of social security reform as a contribution to the debate about the region.1 There is a concern that it is premature to conduct an assessment of pension reform because the experience in most countries is not long enough to permit a reliable investigation. But while economists and pen- sion specialists wait for the reforms to bear riper fruit, these reforms may well be uprooted in some countries. Consider the following developments: · In Peru in late 2002, during the process of rewriting the constitution, articles that would allow affiliates of the private funded system to return to the public PAYG system and to lower the retirement age from 65 to 60 were narrowly defeated. · In Bolivia in late 2002, the government approved measures that would integrate the fund that finances the noncontributory pension bene- fit with the contributory pension funds, a measure that would lead to less than transparent cross-subsidization and could cut the return earned by affiliates on their retirement savings. · In Argentina in 2002, a draft law that would allow workers to switch between the public and the private branches of the pension system was passed in the Lower House--with only one vote against and one absten- tion. By late 2003 the government was working on a reform that would allow workers to return to the old PAYG system. Among the alternatives being weighed are a mixed public­private system with universal coverage to protect the indigent and replacement of the current largely private pen- sion system with a single regimen in which workers would see their con- tributions apportioned between a state-managed pension agency and a private pension system. Similar discontent has emerged in some of the other countries that have adopted the multipillar model with its emphasis on privately man- aged individual accounts. Even in Chile where reforms have been imple- mented with the most vigor--and the new pension model has been in place for the longest period--there are concerns that low participation in funded pensions will keep effective replacement rates low, and thus put mounting pressures on the minimum-income guarantee for affiliates who have completed the minimum required period of contribution to the system. This book aims to provide a balanced and reliable assessment of THE WAY FORWARD 269 country experience with the multipillar model to date, so that any further reform measures are contemplated in an informed manner. What Have Been the Main Benefits of Reform? In light of the concerns above, it is important that the benefits of reforms be made widely known. And, as documented in this volume, the benefits are considerable. First, the reforms are based on a more sustainable social contract where the consumption-smoothing goal is "individualized." The reforms rely on the implementation of individual accounts in which benefits are based on workers' contributions. The costs of financing increased life expectancy are therefore passed on to each generation rather than pooled in a discre- tionary way, as tends to be the case in defined benefit plans. All countries surveyed in this book have introduced such accounts. Second, the aggregate liabilities of governments have fallen. Work com- missioned for this volume reveals that the implicit pension debt has been significantly reduced in most countries that reformed their pension systems. Compared with the hypothetical, counterfactual scenario of no reform, pro- jections show that implicit pension debt (IPD) as a share of GDP in 2001 was lower in the reformed systems by 100 percentage points in Chile, 50 points in Bolivia, and about 25 points in El Salvador, Peru, and Uruguay. The reductions in IPD as a share of GDP are much higher further out in the future: by 2030, for example, the difference between the reform and no- reform scenarios becomes 200 percent of GDP in Bolivia and Chile and 100 percent in El Salvador and Peru (figure 3.1; Zviniene and Packard 2002). Third, there have been important distributional benefits. Besides allowing countries of the region to spend more on public education, health, and social assistance, in all the reforming countries the regressivity of public pension expenditures has been markedly reduced for those who are covered, when measured using (gross of commission and fee) rates of return obtained by wealthier and poorer workers. In some countries, such as Argentina and Chile, reforms helped regressive systems become pro- gressive (figure 5.1, and Zviniene and Packard 2002). Fourth, at least initially reforms led to some improvements in coverage. That is, even though casual analysis shows a fall in contribution rates (see figure 5.5), more careful estimations in two background papers for this book indicate that coverage rates might have been even lower without the reforms (Packard 2001; Valdés-Prieto 2002a). Fifth, the shift to multipillar systems can be credited with setting up a new financial industry that has been a role model for other industries in the region in terms of regulatory oversight. As a paper commissioned for this book argued, the new systems have achieved high standards in asset valuation, risk rating, and disclosure (Yermo 2002a). So whereas the 270 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA direct role of pension reform in increasing national saving is debatable, the indirect effect on saving of improved financial sector functioning is likely to have been positive. Finally, another financial benefit of the reforms has been the rapid growth of a new form of saving in the region. Between 1998 and 2002 the ratio of pension fund assets to GDP rose from 40 percent to 56 percent in Chile, from 2.7 percent to 5.3 percent in Mexico, from 3.3 percent to 11.3 percent in Argentina, from 2.7 percent to 7.7 percent in Colombia, from 2.5 percent to 8.1 percent in Peru, from 1.3 percent to 5.7 percent in Uruguay, from 0.4 percent to 7.4 percent in El Salvador, and from 3.9 per- cent to 15.5 percent of GDP in Bolivia (see table 4.3). Furthermore, insur- ance companies have flourished in their auxiliary role as providers of dis- ability, survivor, and longevity insurance in the new systems. Although it is not obvious how much of this growth in pension fund assets has taken place at the expense of other institutional investors such as mutual funds (which operate on the basis of financial principles similar to those of the pension funds), there should be little doubt that the importance placed on mandatory individual savings accounts has helped financial sector devel- opment. It should be noted, however, that financial sector development can take place effectively in the absence of pension privatization. After all, pension fund assets have also risen impressively in Brazil, a country that has not introduced mandatory private retirement accounts, from US$33 billion in 1994 to $75 billion in 1998 before falling to $53 billion in 2001, and mutual fund assets have grown from slightly over 15 percent of GDP in 1998 to nearly 30 percent in 2001 (see figure 10.2). What Are the Principal Concerns about Reforms? Why is there growing discontent with reforms to social security in Latin America? Some of this discontent may be transitory (arising from sharply lower rates of return on private savings accounts in the past few years compared with rates during much of the 1990s), and some perhaps un- warranted. But it is hard to dismiss all such complaints as baseless. This section considers some potentially valid concerns. First, fiscal concerns are prominent for many policymakers. Although there is no doubt about the long-term fiscal benefits of reform, the short- to medium-term fiscal effects depend greatly on initial fiscal conditions, the extent to which reforms reduce contingent liabilities, and the quality of the reform's implementation. Not all countries have the luxury of embarking on structural reform from a strong fiscal position, but those that take steps to (a) reduce contingent liabilities before making implicit pension debt explicit, and (b) curb fraud and mismanagement in reform implementation can limit transition costs to manageable levels. Finalizing the "first genera- tion" of pension reforms should be a priority for policymakers. THE WAY FORWARD 271 Second, there are concerns over the new structure of the multipillar pension systems. Some countries in the region maintain large, defined benefit PAYG programs that offer earnings-related pensions. In Costa Rica and Uruguay the PAYG plan is mandatory--whereas in Argentina, Colombia, and Peru workers can choose between the defined-benefit PAYG plan and an individual account in the funded defined contribution pillar. These earnings-related, defined benefit PAYG plans will become increasingly costly to finance as life expectancy increases unless policy- makers are able to enact increases in retirement ages or introduce longevity factors into their benefit formulas. At the same time, many countries have not implemented poverty pre- vention pillars for the uncovered population, whereas others have not even done so for the covered population. In Peru, a country that ostensi- bly adopted the multipillar model of old-age income security in 1992, there are understandable concerns that although the system of private mandatory savings accounts was vigorously implemented, the government did not institute the minimum pension guarantee (see box 8.2). And be- ginning in early 2002, when the guarantee component was finally in- stalled, only affiliates who were 55 or older who had contributed for at least 20 years were eligible in the private pillar. It is important to point out that the Peruvian authorities have included the extension of the minimum pension guarantee to all the affiliates of the AFP system who complete the minimum years of contribution as part of the reform agenda. But it is worth noting that it will have taken more than a decade for the compo- nent designed to alleviate old-age poverty to be instituted. No concrete moves have yet been taken toward extending the minimum guarantee. It may well be that the administrative and political demands associated with installation of the private second pillar--which generally has powerful champions in bankers and financiers--actually divert attention from efforts to set up the arguably more important poverty prevention pillar, whose main beneficiaries are the largely unchampioned poor people. Similar concerns have surfaced over Bolivia's universal pension bene- fit, the BONOSOL. Only workers who were 21 years old or older in 1995 are eligible for this benefit, leaving later generations without resort to any mechanism for pooling poverty risk. Nevertheless, as in Peru, a strong political constituency could build over the next few years to de- mand the extension of the benefit to future generations. Programs such as BONOSOL should always be modest; but in Bolivia's current cir- cumstance of debt and high fiscal deficits, restraint is a necessity not an option. Another serious structural concern is the potential inconsistency of reformed systems that aim to pursue poverty prevention and consumption- smoothing objectives through the savings component (i.e., conditioning eligibility for poverty prevention benefits on contribution to second-pillar individual savings accounts). Workers in Chile who earn close to the 272 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA minimum wage are now expected to choose among five different funds with various levels of investment in equities. A rational worker would choose the highest risk portfolio in the knowledge that the government guarantees them a minimum pension if they meet the minimum contribu- tion period. In this book we have argued that the instruments for poverty prevention and consumption smoothing are different and should be kept separate from each other. In particular, access to instruments that pool the risk of poverty should not be conditioned on participation in the savings component. Individual choice over investment is an essential ingredient of the latter, but efficient pooling of poverty risk requires the transfer of investment choice to the underwriter of the risk--the state in this case. Third, there are concerns over the operation of the savings component. Poor and young people may face other pressing demands on their con- sumption and may have limited disposable income for retirement saving. High mandatory contribution rates may, therefore, dissuade these groups from participating in the pension system. Individuals may also find other forms of investment, such as housing or education, equally or even more attractive than the pension funds--especially because of the high ceilings on earnings that are subject to mandatory contributions. It fact, in Peru there is no ceiling and the mandatory contribution rate is levied on affili- ates' full earnings. High ceilings on workers' earnings that are subject to the mandate to save restrict voluntary savings in alternative instruments by even richer workers. If due to a large mandate workers are forced to bor- row to meet important demands (such as housing, education, or health- care), workers can face a further cost represented by the spread between the cost of consumer credit and the return earned on mandatory savings. Policy risk also remains omnipresent in the private funded systems. In Argentina the value of pension fund assets fell sharply when the govern- ment forced pension funds to hold its increasingly risky debt and then defaulted on that debt. The experience of Argentina has brought into clear relief what has always been acknowledged by balanced observers, namely, that "there is little reason to believe that a government that administered a public system poorly would regulate a private system well" (de Ferranti, Leipziger, and Srinivas 2002, p. 42). Argentina's experience shows that even privately administered pension systems are not immune to public policy mistakes. Even in countries such as Chile that have proved over time to be competent regulators of mandatory pension funds, there are worrisome equity-related findings. Three are especially noteworthy: 1. About one-quarter of the cumulative net (of insurance premium) con- tributions of the average worker who retired in 2000 after contributing to the system since its inception in the early 1980s went toward administrative fees (see figure 7.7). The level of commissions changed for administering the funds has prompted critical commentary. As administrative costs have fallen, THE WAY FORWARD 273 later cohorts of workers may do better, but even this positive development raises concerns of intergenerational equity in nascent privately managed mandatory saving systems. Whereas the former PAYG systems transferred wealth from future generations to current ones, these systems have a reverse bias as current generations are forced to pay the set-up costs of financial structures from which future generations would benefit. At least in the Chilean case, earlier cohorts benefited from higher gross returns, but this has not been the case in all Latin American countries. Besides, the experience in Peru indicates that falling operational costs do not necessarily mean lower fees for contributors, and may indeed simply translate into higher profit mar- gins for the fund administrators (see Lasaga and Pollner 2003). 2. The commission structures in countries such as Chile imply that poor people may end up paying a higher share of their salaries in commissions than do wealthy people. Thus the management fees act like a regressive tax. In Chile the ratio of total commissions (administration fees and insurance premiums) to total contributions for the average contributor who retired in 2000 was 28 percent, while for a worker earning the minimum wage it was nearly 33 percent (see figure 7.8). 3. Some random intergenerational inequity is inherent in a system of defined contributions even when it matures, because of the uncertainty associated with annuitization. Chileans retiring between 1988 and 2001 with identical accumulated balances could end up with annual pension differentials of more than 20 percent (figure 7.9). Such volatility is natu- rally a legitimate concern in a publicly sponsored system designed to reduce uncertainty during retirement. Risk management considerations are also a cause of concern in the new Latin American systems. International diversification of pension fund port- folios is heavily constrained or even prohibited, whereas exposure to public debt is excessive, given the extent of default risk in much of Latin America. Workers outside Chile cannot choose their investment portfolio and are thus unable to adjust the asset allocation of their individual accounts to their risk and time preferences. The retirement stage is also subject to some possibly unnecessary constraints. Annuities are mandatory at retirement in some countries and in others they cannot be combined simultaneously with sched- uled withdrawals. Deferred and variable annuities markets are also insuffi- ciently developed. Finally, and perhaps most importantly, in many of the reform countries the coverage of social security systems has stagnated at levels that seem un- acceptable to many Latin Americans (see chapter 5). Although structural pension reform is not the only factor influencing participation, stagnant coverage ratios are a major concern for reforms that were expected to ex- tend access to formal protection to a wider segment of the population. A large portion of affiliates (more than 30 percent in Chile according to our estimates) may not even qualify for the minimum pension guarantee of 274 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA PAYG or funded systems. Many more affiliates may only qualify for the minimum guaranteed benefit. Those who do not qualify for the minimum pension, together with those who are not affiliated with any system, have generally only rationed social assistance benefits to look forward to in old age. The mass of uncovered informal workers is less likely to take to the streets in protest than are formal sector beneficiaries whose benefits are cut, but people without coverage nonetheless form a large constituency of dissatisfaction with pension reform. Because increased coverage was one of the objectives of the multipillar reforms, it is understandable that lack of progress in this area has raised discontent. It is reasonable to ask whether this preoccupation is legitimate. It can be argued on the basis of cross-country evidence (such as that presented in figure 11.1) that coverage is greater in higher-income countries. The only sustainable way for countries in Latin America to increase coverage is to focus on policies that increase economic growth rather than social security coverage. But two caveats should give us pause: 1. There is considerable variation in coverage rates within reasonably narrow income groups. Thus, for example, countries with per capita Figure 11.1 Coverage Can Vary at Similar Stages of Economic Development 120.0 force 100.0 labor Nicaragua of 80.0 Uruguay Chile 60.0 Costa Argentina percentage Rica Mexico as 40.0 Colombia ES Ecuador 20.0 DR Peru Coverage, Bolivia 0 0 5 10 15 20 25 30 Per capita income, in thousands of dollars Source: Palacios and Pallares-Miralles (2000). THE WAY FORWARD 275 incomes between US$2,000 and $3,000 have coverage ratios ranging from almost 0 to about 50 percent of the labor force. Again, countries with income levels of about $7,000 to $8,000 have coverage rates ranging from less than 10 percent to more than 60 percent. Although part of this variation results from the socialist pasts of some countries, institutional variation cannot be ignored. Such variation exists within Latin America, as well. Witness coverage rates of 62 percent in Chile versus 22 percent in Costa Rica (both upper-middle-income countries), or 25 percent in El Salvador versus 11 percent in Peru (both lower-middle-income countries).3 2. There may be reason to believe that access to social security-- unemployment benefits, antipoverty programs, and retirement income programs--can promote openness and other growth-oriented economic policies, since they help in dealing with the higher volatility that ac- companies these economic reforms. In Latin America there are espe- cially good reasons to believe this (de Ferranti et al. 2000). In opinion polls income insecurity among workers and retirees is usually one of the top concerns of the electorate. Democratic governments legiti- mately consider it a policy priority. What Is the Way Forward? The analysis in this book--a blend of the economics of insurance, detailed review of the experience in Latin America, simulations based on admin- istrative data, and econometric analysis of individual and household surveys--leads to several conclusions. First, and most important, the poverty prevention pillar should get a lot more attention than it has in Latin America during the past decade. This role of government may be even more important than that of pro- viding a safe savings instrument at a reasonable price; the behavior of Chilean contributors reported in chapter 8 appears to reveal a preference for the minimum pension guarantee over the mandatory savings instru- ment. The poverty prevention role of government only increases in im- portance with economic development--as the likelihood of poverty in old age declines, the fundamentals of insurance make pooling of this risk across individuals more, not less, appropriate (see chapter 6). A govern- ment mandate is necessary for this type of defined benefit system because private insurance markets are unlikely to cover the risk of poverty. When funded from general revenues (in contrast with earmarked payroll taxes) and not conditioned on contributions, such a plan is sometimes called a zero pillar. Regardless of what it is called, such a system can permit gov- ernments to play a less ubiquitous role in other forms of old-age income security. Although most countries have some form of minimum pension benefit, some countries (Chile, Colombia, El Salvador, Mexico, and Peru 276 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA when it is implemented) have tied eligibility to the contribution record in the savings component. These countries should consider reopening the debate on the need for a poverty prevention pillar offering benefits that keep households out of poverty and that is based on pure pooling princi- ples (a PAYG defined-benefit plan). Also needed is further analysis of whether universal, subsistence-level pension benefits are more suitable than targeted benefits for low- and middle-income countries with high levels of informality and weak public governance. If contributory systems are retained they should be complemented with means-tested social assis- tance, which is not yet the case in four of the countries surveyed. Second, it should be emphasized that although first-pillar programs will provide a minimum pension to those people who are unfortunate or un- wise, the mainstay for earnings replacement during old age--that is, mech- anisms to help households smooth consumption and cover the loss of earnings capacity while living--should be saving. For most workers it should consist of plans that closely link benefits to contributions and life expectancy and involve no redistribution of benefits or pooling of longevity risk across generations. The design of defined contribution plans is better suited for the earnings replacement function because those plans uncouple the contributions of individuals in one generation from the ben- efits obtained by previous generations. It should be noted also that defined contribution plans can be as effective at pooling longevity risk within the same generation as are defined benefit plans. In all Latin American coun- tries one option at retirement is to purchase an annuity that offers such protection. The move from pure defined-benefit PAYG systems to defined contribution individual accounts as the mainstay of retirement income se- curity is therefore clearly an improvement that conforms to the simplest principles of the economics of insurance because it leads to a better match- ing of instrument to covered risk, thereby increasing the role of saving at the expense of pooling. For this reason alone the reforms in Latin Amer- ica are worth preserving and strengthening. But one should also acknowl- edge the progress made in countries such as Brazil that have moved from pure defined benefits to a hybrid that results in a tighter link between how much workers contribute and how much they receive as pensions. Coun- tries such as Colombia, Costa Rica, Peru, and Uruguay that have retained PAYG systems that provide generous earnings-related pensions should also carefully consider reforms--such as introducing longevity factors in benefit formulas that will improve the actuarial fairness and solvency of their remaining public plans. Third, more attention should be paid to the size of the mandatory savings pillar relative to country-specific circumstances. High contribution rates and similarly high ceilings on earnings subject to the mandate can dis- courage workers from participating and may leave little space for the growth of the third pillar. This is most likely in developing countries for poorer and younger workers who have other urgent competing demands THE WAY FORWARD 277 on their disposable income. As long as a broad and fiscally sustainable poverty prevention pillar is in place, one option is to lift the obligation to save for workers below a certain income level. Contribution rates could also be made age-related so that workers contribute more into the savings system as they approach retirement, although the work incentives of older workers should be considered. Participation by richer workers could also be encouraged by lowering earnings ceilings. These measures would allow greater flexibility in mandatory saving for old age and could foster the de- velopment of the voluntary pensions market. Fourth, the usefulness of second-pillar pensions in making a transition from overly generous PAYG systems and, even more so, in providing an initial boost to capital and insurance markets should be put into context. Some countries do not have such needs, which makes mandatory saving plans redundant (see box 11.1). Countries such as Brazil that have reasonably well-developed capital markets may choose to change the parameters of their public PAYG pension systems rather than switch to a mandatory funded program. Moving to a system of notional defined con- tributions is also an option. Conversely, countries such as Paraguay that have not reformed their PAYG pensions and wish to rejuvenate or develop their financial sector could seriously consider mandatory saving plans. But although workers are likely to respond to improved incentives to con- tribute at the margin, such countries should be warned not to expect a second pillar to increase coverage on its own, and should be advised to institute a robust poverty prevention pension program, however modest. In general, to the extent that governments have limited resources to administer or supervise pension systems, the priority should be the poverty prevention pillar rather than mandatory saving plans. Fifth, the twin goals in countries that already have well-regulated private second pillars should be to institute equally well-run and fiscally sustainable public poverty prevention programs and to lower the costs of mandatory saving plans. Although there have been several attempts to reduce the commissions in second-pillar pensions through regulatory measures and to offer a wider array of savings instruments that differ in their risk and return features, commission rates are still not as low as they could be. In Bolivia commission rates are between two-thirds and one-half of those in other Latin American countries, despite the relative immaturity and small size of the system. Countries can benefit from evaluating this and other experiences in cost management that range from fully central- ized models to fully contestable markets where competition is the time- tested ingredient for efficiency. Between these two extremes there are arrangements such as occupational plans in which employers act as inter- mediaries between workers and pension providers (a practice well ex- tended in OECD countries), and systems with centralized account man- agement and record keeping but open competition in asset management, as in Sweden. Policymakers in the region have also balked at increasing 278 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Box 11.1 The Role of the Second Pillar Despite the clear message in Averting the Old Age Crisis (World Bank 1994) that all three pillars--the "mandatory tax-financed public pillar that has primary responsibility for redistribution," the "mandatory funded private pillar (of personal saving or occupational plans) that has primary responsibility for saving," and the "voluntary pillar that pro- vides supplementary protection" (p. 48)--are necessary to ensure against life's risks and uncertainties, it would not be an exaggeration to say that the component that has received most attention in countries of Latin America during the last decade is the second pillar. Because this was the novel component (most countries in the region already had some social assistance pensions and voluntary retirement savings), the preoccupation with mandatory saving plans is understandable. There are two other pos- sible rationales for this intense attention to mandatory saving schemes: 1. Shifting from unsustainable defined benefit plans. Most reforming countries started from a situation of fiscally unsustainable PAYG sys- tems. Politically it may have been difficult to renege on these generous promises. The second pillar may have changed the basis of old-age in- come security from overly generous defined benefit programs to defined contribution plans, which are by definition fiscally balanced. Of course, privately managed individual accounts are not the only way to change the basis of earnings replacement from defined benefits to defined contribu- tions: notional defined contribution programs are an obvious option (see Fox and Palmer 2001). 2. Fostering capital market development. Most of these countries had underdeveloped private financial markets, so the second pillar was seen as an instrument for jump-starting financial sector development. Although there are alternative ways to encourage private financial markets to begin offering financial instruments for long-term savings, it may well be that sec- ond-pillar pension plans are an effective instrument for doing so. In either case, the role of a mandatory, privately managed individual savings accounts appears to be more transitory in nature than suggested in Averting the Old Age Crisis. The most important reasons for a perma- nent privately administered second pillar are in fact a general distrust of the ability of governments to manage such accounts and the belief that people are generally irrational or ill-informed about their retirement needs. Based on Latin America's experience in the last decade--which provides little in the way of assurance that the three central dilemmas posed in Averting the Old Age Crisis (see box 1.1) are not serious con- tradictions--mandating private savings is seen as a transitional device in this volume. The simplest way to highlight the differences between Avert- ing the Old Age Crisis and this book is to consider what each would recommend for countries such as the United States. Whereas Averting THE WAY FORWARD 279 Box 11.1 (continued) would recommend a sizable system of privately managed mandatory savings accounts for the United States, the reasoning in this report would not lead to that conclusion. Of course this does not mean that countries should never institute a second pillar even when conditions warrant. But it would be equally un- wise to advocate a second pillar in every country that wishes to move the mainstay of old-age income security from pooling (i.e., defined-benefit PAYG systems) to saving (e.g., defined contribution plans), even if the ca- pacity to effectively regulate such a system existed. Other instruments such as notional defined contribution accounts may serve the purpose equally well or better. In fact, where the obligations of the PAYG system are exceptionally high, mandatory savings accounts may be eschewed because of macroeconomic concerns. competition between mandatory and voluntary pension providers. But all governments must squarely face the question: At what stage of financial sector development is it safe to stop cradling this infant industry created by structural pension reform? It is difficult to translate this into elapsed time. Chile appears to have arrived at this point after two decades of suc- cessful implementation of the second pillar: preferential tax treatment of retirement saving is being extended to voluntary contributions to approved (non-AFP) providers, which has led to a sharp increase in the number of contributors. In countries such as Argentina it may be neces- sary to increase competition to restore faith in the system. At the same time it should be noted that increased competition is only effective in lowering costs and improving performance to the extent that it is accompanied by measures to improve the disclosure of fees and performance, as well as the financial education of workers. Sixth, some risk management aspects of the second pillar are also in dire need of a second generation of reforms. Investment regulations are generally overly prescriptive. In some countries there is still a dangerous inclination to divert pension funds toward government securities of low credit rating and a lack of consideration of the virtues of international diversification. The consumption-smoothing objective that underpins the savings pillar also requires individual portfolio choice. Although consid- erations over the financial literacy of the population may delay this reform in the second pillar, it would be wise to permit it at least for voluntary pensions, in which mainly better informed individuals participate. Greater flexibility may also be needed in the retirement stage so that workers may better hedge, diversify, and insure investment and longevity risk. 280 KEEPING THE PROMISE OF SOCIAL SECURITY IN LATIN AMERICA Conclusion Given the importance of reducing the risk of poverty in old age, in this volume we call for greater attention to the poverty-prevention pillar. While individual savings should be a mainstay of consumption smoothing, we also call for a reassessment of the relative size of pillars two and three. But we do not argue that there is never a role for a mandatory savings component within a comprehensive old-age income security system. Almost anything that helps the move from pooling to saving would improve pro- grams designed to replace earnings given rising longevity. But the argu- ments for large mandatory savings pillars appear to be mainly transitional in nature. This reasoning assumes that countries such as Chile, Colombia, El Salvador, Mexico, and Peru follow the advice offered in this book and introduce separate poverty prevention pillars based exclusively on pooling principles. The reality is that most countries in the region started from a situation of high and unsustainable rates of replacement promised by PAYG systems; politically the second pillar may be a convenient tool for ratcheting down these expectations. Reforms have been a step in the right direction but can go further in reducing remaining liabilities of the earlier PAYG system--and thereby create fiscal space for a poverty pre- vention pillar and reduce the size of the second pillar, to make room for the third. To see the second pillar as a large permanent part of the pension sys- tem is to assume individuals are irrational--or, not to have faith in the ability of people to plan their own futures--and to assume that govern- ments have the capacity to vigorously regulate these growing pools of money--that is, to have a lot of faith in governments to behave in an even-handed manner. There is little evidence that either assumption is jus- tified; the coincidence of both in a single country is likely to be rare. Even assuming widespread improvidence in the population (systematic under- estimation of consumption needs in old age until it is too late) rather than irrationality implies a much smaller and perhaps less permanent second pillar. As longevity increases and poverty decreases in the region, the conse- quent need for individualization of pension systems becomes clearer, and the proper role of government becomes of paramount concern. As we have argued in this book, the desirable principles of a pension system--fiscal sustainability, access, equity, incentives to participate, choice, and poverty prevention--do not map into universal policy packages, just as Rodrik (2003) has noted with regard to economic policies in general. The need to reform the region's old single-pillar PAYG pension systems is clear. But just as fiscal, financial, social, and political conditions differ greatly across countries in the region, successful approaches to securing income and pre- venting poverty in old age will differ too. THE WAY FORWARD 281 Notes 1. The 16 background papers commissioned for this study are available online at www.worldbank.org/keepingthepromise. 2. The rationale behind this regulation was to eliminate disincentives for older workers who were still contributing to the public earnings-related plan to move into the private system of individual accounts. 3. Data on coverage ratios are calculated from recent household surveys and reported in figure 1.1, rather than those from Palacios and Pallares-Miralles (2000) reported in figure 11.1, which uses data from the mid-1990s. Technical Annex. Assumptions of PROST Simulations The PROST (Pension Reform Options Simulation Toolkit, developed by the World Bank) simulations presented in chapter 3 impose a uniform set of macroeconomic and certain systemic assumptions, in addition to coun- try-specific assumptions based on each case, to isolate the impact of struc- tural reforms. Although the use of a uniform set of macroeconomic as- sumptions would seem a strange choice when simulating reforms in a group of countries as diverse as El Salvador, Mexico, and Uruguay, it is not without precedent in cross-country simulation analysis (see Holzmann, Palacios, and Zviniene 2001). The assumptions that are common to all country cases are presented in table TA.1. Details particular to each coun- try case are presented in table TA.2. In addition to the parameter assumptions documented in table TA.1 (and those listed in table TA.2), there are assumptions behind the simu- lated pension outcomes used in the equity analysis that have to be kept in mind when analyzing our results. Calculations of Internal Rates of Return Zviniene and Packard (2002) calculated internal rates of return from repre- sentative affiliates' "investments" in the formal retirement security system-- Table TA.1 Assumptions Common to All Country Cases GDP growth 3 percent. Inflation 0 percent. Market interest, discount rate 5 percent. Interest in "deaccumulation" or "payout" phase 4 percent. Interest on PAYG investments 3 percent. Wage bill is set at a constant portion of GDP. Full compliance with retirement age increase: everyone contributes if not allowed to retire. Collection rate of contribution revenue 100 percent. For sustainable benefit we assume that all benefits (old age, disability, survivors, and so forth) are reduced by the same percentage. All demographic data are taken from World Bank population database (see Bos et al. 1994). No coverage expansion or contraction. Recognition bonds are not included in implicit pension debt calculations. Minimum pension guarantees are modeled where applicable. PAYG-pillar benefits, income thresholds, contribution ceilings, and minimum pensions are indexed to nominal wage growth. Fees are paid by affiliates to funded second pillars, as in Devesa-Carpio and Vidal-Meliá (2002). 283 284 Table TA.2 Country-Specific Details, Assumptions, and Sources of Data Country case Reforms modeled Assumptions Data sources Chile Introduction of a funded second Administrative costs of PAYG pil- Wage distribution, pension distri- pillar lar as a percentage of contribu- bution, beneficiary numbers, cur- tions are assumed to be 5 percent rent spending, current average (reliable data were not available). length of service data from the Average length of service will con- Pension Normalization Institute tinue to be 23/38 (women/men). (Instituto de Normalización Pre- visional) and the Superintendency of Pension Fund Administrators (Superintendencia de Admin- istradoras de Fondos de Pen- siones) Peru Introduction of a funded second Average length of service stays con- Wage distribution, pension distri- pillar stant at 23.8 years. bution, beneficiary numbers, cur- Increase in contribution rate from Administrative costs of PAYG pil- rent spending data from the Pen- 9 percent to 13 percent in PAYG lar as a percentage of contribu- sion Normalization Office second pillar and 12.4 percent in tions are assumed to be 5 percent (Oficina de Normalización Previ- funded second pillar (reliable data were not available). sional) and the Banking and TECHNICAL Insurance Superintendency (Superintendencia de Banca y Seguros) ANNEX TECHNICAL Colombia Introduction of a funded second Average length of service is 20 Beneficiary numbers, current pillar years (mandated minimum). spending, administrative costs of Increase in retirement age from Only the Institute of Social Security PAYG data from local sources 55/60 to 62/57 (men/women) in (Instituto de Seguro Social) is Wage distribution, switching pat- year 2014 modeled (33 percent of all pen- terns data from Kleinjans (2001) ANNEX sion liabilities); other special regi- mens for civil servants make up the rest. Argentina Introduction of a funded second Administrative costs of PAYG pil- Wage distribution, pension distri- pillar lar as a percentage of contribu- bution, beneficiary numbers, cur- Decrease in contribution rate for tions are assumed to be 5 percent rent spending data from the employers from 16 percent to 7 (reliable data were not available). National Social Security Adminis- percent in PAYG second pillar Average length of service since tration (Administración Nacional 2008 is assumed to be 30 years de la Seguridad Social) and the for those who retire at ages Superintendency of Retirement younger than 70 and 15 years for and Pension Fund Administrators those who retire later (mandated (Superintendencia de minimum). Admistradoras de Fondos de Jubiliación y Pensiones) (Table continues on the following page.) 285 Table TA.2 (continued) 286 Country case Reforms modeled Assumptions Data sources Uruguay Introduction of a funded second Average length of service is Wage distribution, pension distri- pillar (more appropriately charac- assumed to be 35 years for those bution, beneficiary numbers, cur- terized as a "tier" because partici- who retire at ages younger than rent spending, administrative pation is determined by income) 70 and 15 years for those who costs of PAYG pillar, switching Equalization of retirement ages for retire later (mandated minimum). patterns data from local sources both genders at age 60 Mexico Introduction of a funded second Average length of service is expected Wage distribution, pension distri- pillar to grow from current 22 years to bution, beneficiary numbers, cur- 25 years by 2012 and then stay rent spending, administration constant (mandated minimum). costs of PAYG pillar data from Male/female ratios in contributor the National Retirement Savings and beneficiary populations are Commission (Comisión Nacional taken from Argentina, which has del Sistema de Ahorra para el a labor force gender structure that Retiro) and the Mexican Social is very similar to Mexico (Mexico Security Institute (Instituto Mexi- 1980: 27 percent female; 1996: 31 cano del Seguro Social) percent female. Argentina 1980: Gender differences in wages data 28 percent female; 1996: 31 per- taken from ENEU (1998) TTECHNICALEC cent female [WBI data]). For H NICAL working-age disabled people we used contributor ratios and for elderly disabled people we used ANNEX old-age pensioner ratios. INFONAVIT and social quota pro- grams are included. Bolivia Introduction of a funded second Administrative costs of PAYG pil- Wage distribution, pension distri- TTECHNICALEC pillar lar as a percentage of contribu- bution, beneficiary numbers, cur- HNICAL tions are assumed to be 5 percent rent spending data from local (reliable data were not available). sources Average length of service is 25 Gender differences in wages data ANNEX years (mandated minimum). from CEPAL (1999) BONOSOL program is modeled. El Salvador Introduction of a funded pillar Average length will continue to be Wage distribution, pension distri- 25/27 years (men/women). bution, beneficiary numbers, cur- rent spending, administrative costs, current average length of service data from local sources 287 288 TECHNICAL ANNEX an indicator that can/be used to assess the likely impact of reforms to the formal pension systems on income equity. When an individual participates in a pension system, he or she contributes during the working years and receives benefits after retirement. This can be seen as an investment process where the individual invests the contribution amount into a fi- nancial instrument and receives returns. The internal rate of return (IRR) evaluates this particular investment in the pension system and gives an idea of how much an individual gets for the contributions he or she made. It can be compared with the market rate of interest to evaluate participa- tion in the system relative to other alternatives. In PROST the profiles of different types of individuals are entered. These profiles vary in terms of gender, starting wage as a percentage of average economy wage, starting age of work, retirement age, and life expectancy. PROST then calculates the IRR for each individual based on the individual's cash flows. When comparing different profiles, the IRR statistic acts as a very good indicator of the redistributional properties of a pension system. For example, it can answer the following question: Who gets more out of the pension system--low-income worker or high-income worker, male or female, young or old? By definition, the internal rate of return is the discount rate for which the net present value of the contributions made and the future benefit pay- ments for the individual until death becomes zero. PROST calculates IRR in the following way: 1. The first step is to calculate the cash flows of the individual for each year. The cash flows consist of the contributions from the individual dur- ing the working years and the benefits received after retirement. Contri- butions are taken as negative because they are cash outflows from the individual, and benefits are taken as positive. The cash flows are then con- verted into real terms by discounting for inflation. 2. PROST then uses the following formula to equate the net present value of the real cash flows at the starting year to zero and solve for the internal rate of return. flowrt , if t start year balancet e balancet 1[1 irr] flowrt if t start year where flowrt is real cash flows at time t Equating balancet to zero at the end of the individual's life gives the irr.1 Affiliate Profiles and Other Assumptions Details on the affiliate profiles used in this analysis are shown in table TA.3. We used gender-specific average retirement ages and average years TECHNICAL ANNEX 289 of contributions from PROST output tables. The average retirement age for men and women was applied to each of the three affiliate profiles that we constructed for each gender (poorer affiliates, average affiliates, richer affiliates). When including years out of the formal labor market (particu- larly when calculating the internal rate of return for representative affili- ated women), we assumed that these years were evenly spread throughout the affiliate's adult pre-retirement life. Survivor benefits were not included in the internal-rate-of-return calculations. Finally, in most of the country cases shown structural reforms are more than several years old (see table 2.1). In most cases pension system data for the years prior to structural reforms are scarce and when available rarely can be found in the level of disaggregation required for our simulation model. In those cases (Argentina, Chile, Colombia, and Peru) the simula- tions presented were based on current data. To project a "no-reform scenario"--the counterfactual to structural re- forms--the current number of beneficiaries, wage distribution of contrib- utors, and so forth were used and the parameters of the old single-pillar PAYG system were applied. To simulate structural reforms--particularly the introduction of individual retirement accounts--we assumed that reforms were introduced in the base year (the year to which the data ac- tually correspond). However, the switching pattern (i.e., the distribution by gender and age of workers moving from single-pillar to multipillar arrangements) was set to reflect the actual distribution of contributors be- tween the old PAYG and new multipillar systems. To take account of workers' savings in individual accounts from the ac- tual date of reforms to the base year we simulated a one-off transfer of Table TA.3 Assumed Profiles of Representative Affiliated Men and Women Years of Entry contri- Starting Productivity Profile age butions wage growth Mortality Poorer 16 Average 50 percent 50 percent Dies two affiliate 3 of average of average years earlier than average Average 19 Average 100 percent 100 percent 100 percent affiliate of average of average of average Richer 22 Average 150 percent 150 percent Lives two affiliate ­ 3 of average of average years longer than average Note: Average affiliate of each gender is considered representative. Source: Contribution data from each country's social security and pension agen- cies, as in table TA.2 290 TECHNICAL ANNEX funds into the new private pillar. We did this by setting the contribution rate (as a percentage of wages) to the new individual accounts in the base year equal to the actual statutory contribution plus a percentage of wages that reflected average accumulated assets in each country. In this way we simulated a one-off contribution into individual accounts equal to the av- erage accumulation of savings, divided by average wages. The value of recognition bonds was also adjusted for the actual date of reforms. Although this technique is admittedly second-best, past applications have rendered simulation results very similar to those of other authors. Note 1. PROST calculates the life expectancy of the individual using the individ- ual's mortality multiplier indicated in the input file and the mortality tables of the general population. Appendix. Background Papers for Keeping the Promise Author Title Focus area Oliver Azuara The Mexican Defined Con- Country case tribution System: Perspec- tive for Low Income Workers Abigail Barr and Preferences, Constraints, and Labor, coverage Truman Packard Alternatives under Peru's Pension System Abigail Barr and Revealed Preferences and Labor, coverage Truman Packard Self Insurance: Can We Learn from the Self- Employed in Chile? Federico Escobar Pension Reform in Bolivia: Country case A Review of Approach and Experience Norbert Fiess Pension Reform or Pension Macroeconomics Default? A Note on Pen- sion Reform and Country Risk Truman Packard Is There a Positive Incentive Labor, coverage Effect from Privatizing Social Security? Evidence from Pension Reforms in Latin America Truman Packard Pooling, Savings, and Preven- Labor, coverage tion: Mitigating the Risk of Old Age Poverty in Chile Truman Packard, The Reach of Social Security Labor, coverage Naoko Shinkai, and in Latin America and the Ricardo Fuentes Caribbean Rafael Rofman The Pension System and the Country case Crisis in Argentina: Learn- ing the Lessons Salvador Valdés-Prieto Improving Programs That Labor, coverage Mandate Savings for Old Age 291 292 APPENDIX. BACKGROUND PAPERS Author Title Focus area Salvador Valdés-Prieto Justifying Mandated Savings Labor, coverage for Old Age Salvador Valdés-Prieto Social Security Coverage in Labor, coverage Chile, 1990­2001 Juan Yermo Pension Reform and Capital Financial sector Market Development Juan Yermo Delivering Promises in the Labor, coverage, Chilean Funded Pension financial sector System Juan Yermo Pension Reform and Capital Financial sector Market Development Asta Zviniene and A Simulation of Social Secu- Fiscal, equity Truman Packard rity Reforms in Latin America: What Has Been Gained? Note: All papers are available online at the following World Bank Web site: www.worldbank.org/keepingthepromise. Bibliography The word "processed" refers to informally produced works that may not be commonly available through libraries. Acuna, R., and A. Iglesias. 2001. Chile's Pension Reform after 20 Years. Social Protection Discussion Paper 0129, World Bank, Washington, D.C. Agulnik, Phil. 2000. Maintaining Incomes after Work: Do Compulsory Earnings Related Pensions Make Sense? Oxford Review of Economic Policy 16 (1): 45­56. AIOS (International Association of Pension Funds Supervisory Organ- isms; Asociación Internacional de Organismos de Supervisión de Fon- dos de Pensiones). 2002. Boletín Estadístico Número 8 (December). Alderman, H., and C. Paxson. 1992. Do the Poor Insure? A Synthesis of the Literature on Risk and Consumption in Developing Countries. Pol- icy Research Working Paper 1008, World Bank, Washington, D.C. Alessie, Rob, and Arie Kapteyn. 2001. New Data for Understanding Sav- ing. Oxford Review of Economic Policy 17 (1): 55­69. Amadeo, Eduardo, Indermit Gill, and Marcelo Neri. 1998. Do Labor Laws Matter? The Pressure Points in Brazilian Labor Legislation. In Crafting Labor Policy: Techniques and Lessons from Latin America, eds. I. Gill, C. Montenegro, and D. Domeland. London: Oxford University Press. ANSES (National Social Security Administration [Argentina]; Adminis- tración Nacional de la Seguridad Social). 1995. Series Financieras del Sis- tema Nacional de Previsión Social. Período 1962­1994. Buenos Aires. Appelbaum, E., and E. Katz. 1991. The Demand for Children in the Ab- sence of Capital and Risk Markets: A Portfolio Approach. Oxford Eco- nomic Papers 43: 292­304. Arango, Carlos, and William F. Maloney. 2000. Employment, Informal- ity and Poverty in the Argentine Labor Market. In Poor People in a 293 294 BIBLIOGRAPHY Rich Country: A Poverty Report for Argentina. Report 19992-AR, World Bank, Washington, D.C. Arenas de Mesa, Alberto. 2000. Cobertura Previsional en Chile: Lecciones y Desafios del Sistema de Pensiones Administrado por El Sector Pri- vado. Santiago, Chile: Economic Commission for Latin America and the Caribbean. Arenas de Mesa , Alberto, and Hector Hernández Sánchez. 2001. Cober- tura del Sistema de Pensiones de Capitalización Individual Chileno: Di- agnostico y Propuestas. SOCIALIS: Revista Latinoamericana de Política Social 4 (April): 93­124. Arenas de Mesa, Alberto, and Mario Marcel. 1999. Fiscal Effects of So- cial Security Reform in Chile: The Case of the Minimum Pension. Pa- per presented at the APEC Second Regional Forum on Pension Fund Reforms, Vina del Mar, Chile. Aroca Gonzalez, Patricio, and William F. Maloney. 1999. Logit Analysis in a Rotating Panel Context and an Application of Self Employment Decisions. Policy Research Working Paper 2069, World Bank, Washington, D.C. Arrau, P., and S. Valdés-Prieto. 2001. Para Desconcentrar los Fondos de Pensiones y Aumentar la Competencia en su Administración. Capital Markets Reform Project Proposal. http://www.cepchile.cl. Arriagada, Ana Maria, and Gillette Hall. 2000. Managing Social Risks in Argentina. Human Development Network, World Bank, Washington, D.C. Processed. Atkinson, Antony B. 1987. Income Maintenance and Social Insurance. In Handbook of Public Economics, eds. A. Auerbach and M. Feldstein. Amsterdam: North-Holland. Atkinson, Antony B., and John Hills. 1989. Social Security in Developed Countries: Are There Lessons for Developing Countries? Discussion Paper WSP/38, London School of Economics. Atkinson, Antony B., and Joseph Stiglitz. 1980. Lectures in Public Eco- nomics. New York: McGraw-Hill. Attanasio, Orazio P., and James Banks. 2001. The Assessment: Household Saving--Issues in Theory and Policy. Oxford Review of Economic Pol- icy 17 (1): 1­19. Auerbach, Alan J., and Laurence J. Kotlikoff. 1987. Dynamic Fiscal Pol- icy. Cambridge, U.K.: Cambridge University Press. Azuara, Oliver. 2003. The Mexican Defined Contribution Pension Sys- tem: Perspective for Low Income Workers. Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. BIBLIOGRAPHY 295 Baker, Dean, and Debayani Kar. 2002. Defined Contributions from Workers, Guaranteed Benefits for Bankers: The World Bank's Ap- proach to Social Security Reform. Centre for Economic Policy Re- search, London. Processed. Baker, D., and M. Weisbrot. 2002. The Role of Social Security Privatiza- tion in Argentina's Economic Crisis. Washington, D.C.: Center for Economic and Policy Research. Barnes, K., and W. Cline. 1997. Spreads and Risks in Emerging Markets Lending. Working Paper 97-1. Institute of International Finance, Washington, D.C. Barr, Abigail, and Truman Packard. 2000a. Revealed and Concealed Pref- erences and Self Insurance: Can We Learn from the Self Employed in Chile? Discussion Paper 53, Department of Economics, Oxford Uni- versity, London. ------. 2000b. A Tool for Measuring Risk Aversion and Time Preferences in the Field. World Bank, Washington, D.C. Processed. ------. 2002a. Revealed Preferences and Self Insurance: Can We Learn from the Self-Employed in Chile? Policy Research Working Paper 2754, World Bank, Washington, D.C. ------. 2003. Preferences, Constraints, and Alternatives to Coverage Un- der Peru's Pension System. Background Paper for the Regional Study on Social Security Reform, World Bank, Washington, D.C. Barr, Nicholas.1992. Economic Theory and the Welfare State: A Survey and Interpretation. Journal of Economic Literature 30 (2): 741­803. ------. 1998. The Economics of the Welfare State, 3d ed. London: Ox- ford University Press. ------. 2000. Reforming Pensions: Myths, Truths, and Policy Choices. Working Paper WP/00/139, Development Issues Division, Interna- tional Monetary Fund, Washington, D.C. ------. 2001. The Welfare State as Piggy Bank. London: Oxford Univer- sity Press. Barrientos, Armando. 1996. Pension Reform and Pension Coverage in Chile: Lessons for Other Countries. World Development 15 (3): 309­22. ------. 1997. Pension Reform and the Individual Capitalization Pension Scheme Stampede in Latin America, Working Paper 17, University of Hertfordshire Business School, U.K. ------. 1998a. Pension Reform, Personal Pensions, and Gender Differ- ences in Pension Coverage. World Development 26 (1): 125­37. ------. 1998b. Pension Reform in Latin America. Aldershot, U.K.: Ashgate. 296 BIBLIOGRAPHY ------. 1999a. Retirement, Household Income, and Economic Vulnera- bility of Older Groups. Working Paper 20, University of Hertfordshire Business School, U.K. ------. 1999b. Economic Risks, the Labor Market and Older Workers in Latin America. Working Paper 21, University of Hertfordshire Business School, U.K. Beck, T., A. Demirguç-Kunt, and R. Levine. 1999. A New Database on Fi- nancial Development Structure. Policy Research Working Paper 2146, World Bank, Washington, D.C. Beck, T., and R. Levine. 2001. Stock Markets, Banks, and Growth: Cor- relation or Causality. Policy Research Working Paper 2670, World Bank, Washington, D.C. Becker, Gary S., and Casey B. Mulligan. 1997. The Endogenous Determi- nation of Time Preference. Quarterly Journal of Economics 112 (3): 729­58. Becker, Gary S., and Kevin M. Murphy. 1988. A Theory of Rational Ad- diction. Journal of Political Economy 96 (4): 675­700. Becker, Gary, and N. Tomes. 1976. Child Endowments and the Quality and Quantity of Children. Journal of Political Economy 84 (4): 143­62. Beltrão, Kaizô Iwakami, Sonoê Sugahara Pinheiro, and Francisco Eduardo Barreto de Oliveira. 1999. A População Rural E A Previdência Social No Brasil: Uma Análise Com Ênfase Nas Mudanças Constitucionais. Institute of Applied Economic Research (Instituto de Pesquisa Econômica Aplicada), Brasilia, Brazil. Processed. Bernheim, Douglas, B. 1993. Is the Baby-Boom Generation Preparing Ad- equately for Retirement? New York: Merrill Lynch. Bertranou, Fabio, and Rafael Rofman. 2002. Providing Social Security in a Context of Change: Experience and Challenges in Latin America. In- ternational Social Security Review 55 (1): 67­82. Bertranou, Fabio, C. Solorio, and W. Van Ginneken, eds. 2002. Pensiones No-Contributivas y Asistenciales: Argentina, Brazil, Chile, Costa Rica y Uruguay. Santiago, Chile: International Labour Organization. Blanchflower, D., and Andrew J. Oswald. 1991. Self Employment and Mrs. Thatcher's Enterprise Culture. Discussion Paper 30, Centre for Economic Performance, London School of Economics. -----. 1998. What Makes an Entrepreneur? Journal of Labor Economics 16 (1): 26­60. Blau, David M. 1985. Self-Employment and Self-Selection in Developing Country Labor Markets. Southern Economic Journal 51 (2): 351­63. ------. 1987. A Time-Series Analysis of Self-Employment in the United States. Journal of Political Economy 95: 445­67. BIBLIOGRAPHY 297 Blommestein, H. J. 1998. Impact of Institutional Investors on Financial Markets. In Institutional Investors in the New Financial Landscape. Paris: OECD. ------. 1997. Institutional Investors, Pension Reform, and Emerging Se- curities Markets. Working Paper 359, Inter-American Development Bank, Washington, D.C. Bodie, Z. 1989. Pension Funds and Financial Innovation. Working Paper 3101, National Bureau of Economic Research, Cambridge, Mass. Bos, E., M. T. Vu, E. Massiah, and R.A. Bulatao. 1994. World Population Projections 1994. Estimates and Projections with Related Demo- graphic Statistics, Washington, D.C.: World Bank and Johns Hopkins University Press. Bourguignon, François, Martin Cicowiez, Jean-Jacques Dethier, Leonardo Gasparini, and Pierre Pestieau. What Impact Would a Minimum Pension Have on Old Age Poverty? Evidence from Latin America. Paper Pre- sented at the Keeping the Promise of Old Age Security Conference, Bogotá, Colombia, June 22­23, 2004. Brooks, Sarah, and Estelle James. 2001. The Political Economy of Pension Reform. In New Ideas About Old Age Security: Toward Sustainable Pension Systems in the 21st Century, eds. Robert Holzmann and Joseph Stiglitz. Washington, D.C.: World Bank. Burchardt, Tania, and John Hills. 1997. Private Welfare Insurance and Social Security: Pushing the Boundaries. New York: Joseph Rowntree Foundation. Butelmann, Andrea, and Francisco Gallego. 1999. Household Savings in Chile. Santiago: Division of Economic Research, Central Bank of Chile. Campbell, J. Y., J. F. Cocco, F. J. Gomes, and P. J. Maenhout. 1999. In- vesting Retirement Wealth: A Life Cycle Model. Working Paper 7029, National Bureau of Economic Research, Cambridge, Mass. Carranza, Eliana, and Eduardo Morón. 2003. Diez Años del Sistema Pri- vado de Pensiones (1993­2003): Avances, Retos y Reformas. Lima, Peru: Universidad del Pacífico. Carroll, Christopher D. 1992. The Buffer Stock Theory of Saving: Some Macroeconomic Evidence. Brookings Papers on Economic Activity 2: 61­135. Case, Karl, John Quigley, and Robert Shiller. 2001. Comparing Wealth Effects: The Stockmarket Versus the Housing Market. Working Paper 8606, National Bureau of Economic Research, Cambridge, Mass. CASEN (National Socioeconomic Characterization Survey; Caracteri- zación socioeconómica nacional). 2000. MIDEPLAN, Santiago, Chile. 298 BIBLIOGRAPHY Catalán, M. 2003. Pension Funds and Corporate Governance in Develop- ing Countries: What Do We Know and What Do We Need to Know? Pension Primer Paper, World Bank, Washington, D.C. Catalán, M., G. Impavido, and A. B. Musalem. 2000. Contractual Savings or Stock Market Development: Which Leads? Policy Research Work- ing Paper 2421, World Bank, Washington, D.C. CEPAL (Economic Commission for Latin America and the Caribbean; Comisión Económica para América Latina y el Caribe). 1995. La Medicion de los Ingresos en la Perspectiva de los Estudios de Pobreza: El caso de la Encuesta CASEN de Chile: 1987­1994. Working Paper LC/R. 1604, Santiago, Chile. Cerda, Luis, and Gloria Grandolini. 1998. The 1997 Pension Reform in Mex- ico. Policy Research Working Paper 1933, World Bank, Washington, D.C. Chaloupka, Frank. 1991. Rational Addictive Behaviour and Cigarette Smoking. Journal of Political Economy 99 (4): 722­42. Chamorro, Claudio. 1992. La Cobertura del Sistema de Pensiones Chileno. Thesis 107, Economics Institute, Catholic Pontificate University of Chile. Cheyre, H. 1991. La Prevision en Chile Ayer y Hoy. Center for Public Studies, Santiago, Chile. Coate, Stephen. 1995. Altruism, the Samaritan's Dilemma, and Govern- ment Transfer Policy. American Economic Review 85: 46­57. Cohen, Wilbur J., and Milton Friedman. 1972. Social Security: Universal or Selective. Washington, D.C.: American Enterprise Institute. Cole, William E., and Richard D. Sanders. 1985. Internal Migration and Urban Employment in the Third World. American Economic Review 75: 481­94. Corbo, Vittorio, and K. Schmidt-Hebbel. 2003. Efectos macroeconómicos de la reforma de pensiones en Chile. Study prepared for the Pension Fund Administrators' Industrial Association, Santiago, Chile. Processed. Corsetti, G. 1994. An Endogenous Growth Model of Social Security and the Size of the Informal Sector. Journal of Economic Analysis 9 (1): 57­76. Corsetti, G., and K. Schmidt-Hebbel. 1997. Pension Reform and Growth. In The Economics of Pensions: Principles, Policies, and International Experience, eds. S. Valdés-Prieto. Cambridge, U.K.: Cambridge Uni- versity Press. Cortazar, R. 1997. Chile: The Evolution and Reform of the Labor Mar- ket. In Labor Markets in Latin America: Combining Social Protection with Market Flexibility, eds. S. Edwards and N. Lustig. Washington, D.C.: Brookings Institution Press. Cowell, F. D. 1985. Tax Evasion with Labor Income. Journal of Public Economics 26 (1): 19­34. BIBLIOGRAPHY 299 Cox, Donald, Zekeriy Eser, and Emmanuel Jimenez. 1998. Motives for Private Transfers over the Life Cycle: An Analytical Framework and Evidence from Peru. Journal of Development Economics 55:57­80. Cox-Edwards, Alejandra. 2000. Pension Projections for Chilean Men and Women: Estimates from Social Security Contributions. California State University, Long Beach. Processed. Cuesta, Jose A. 2000. Social Transfers, the Household, and the Distribu- tion of Incomes in Chile. PhD diss., University of Oxford. Cuesta, J., R. Holzmann, and T. Packard. 1999. Extending Coverage in Multi-Pillar Pension Systems: Constraints and Hypothesis, Preliminary Evidence and Future Research Agenda. Paper prepared for World Bank New Ideas about Old-Age Security Conference, Washington, D.C. Cunningham, Wendy, and William F. Maloney. 2002. Re-examining the Informal Sector. In Brazil Jobs Report 24408-BR, 40­60. Washington, D.C.: World Bank. Davis, E. P. 2001. Institutional Investors, Corporate Governance, and the Performance of the Corporate Sector. Economic Systems 26 (3): 203­29. ------. 1998. Regulation of Pension Fund Assets. In Institutional In- vestors in the New Financial Landscape. Paris: OECD. Deaton, Angus. 1989. Savings in Developing Countries. World Bank Eco- nomic Review 3 (Suppl): 61­96. ------. 1991. Saving and Liquidity Constraints. Econometrica 59 (5): 1221­248. ------. 1997. The Analysis of Household Surveys: A Microeconomic Ap- proach to Development Policy. Washington, D.C.: World Bank and Johns Hopkins University Press. de Ferranti, David, Danny Leipziger, and P. S. Srinivas. 2002. The Future of Pension Reform in Latin America. Finance and Development (39)3. de Ferranti, David, Guillermo Perry, Indermit Gill, and Luis Servén. 2000. Securing Our Future in a Global Economy. Latin American and Caribbean Flagship Regional Study, World Bank, Washington, D.C. Demarco, Gustavo, and Rafael Rofman. 1998. Supervising Mandatory Funded Pension Systems: Issues and Challenges. Social Protection Dis- cussion Paper 9817, World Bank, Washington, D.C. Demirguç-Kunt, Asli, and D. Levine. 1999. Bank-Based and Market- Based Financial Systems: Cross-Country Comparisons. Policy Research Working Paper 2143, World Bank, Washington, D.C. Devesa-Carpio, José E., and Carlos Vidal-Meliá. 2002. The Reformed Pension Systems in Latin America. Social Protection Discussion Paper 0209, Human Development Network, World Bank, Washington, D.C. 300 BIBLIOGRAPHY Devesa, J. E., M. Martinez, and C. Vidal. 2000. Análisis y valoración de los sistemas de pensiones reformados en Latinoamérica. WIP-EC-11- 2000, Instituto Valenciano de Investigaciones Economicas, Valencia, Spain. Diamond, Peter A. 1993. Privatization of Social Security: Lessons from Chile. Working Paper 4510, National Bureau of Economic Research, Cambridge, Mass. ------. 1998. The Economics of Social Security Reform. Working Paper 6719, National Bureau of Economic Research, Cambridge, Mass. Diamond, Peter A., and James A. Mirrlees. 1978. A Model of Social In- surance with Variable Retirement. Journal of Public Economics 10: 295­336. ------. 1982. Social Insurance with Variable Retirement and Private Sav- ings, Working Paper 296, Massachusetts Institute of Technology, Cam- bridge, Mass. ------. 1986a. Payroll-Tax Financed Social Insurance with Variable Re- tirement. Scandinavian Journal of Economics 88 (1): 25­50. ------. 1986b. Insurance Aspects of Pensions. In Pensions, Labor and In- dividual Choice, eds. David A. Wise. Cambridge, Mass: National Bureau of Economic Research and University of Chicago Press. Disney, Richard, Robert Palacios, and Edward Whitehouse. 1999. Indi- vidual Choice of Pension Arrangement as a Pension Reform Strategy. Working Paper W99/18, Institute for Fiscal Studies, London. Disney, Richard, and Edward Whitehouse. 1999. Pension Plans and Re- tirement Incentives. Social Protection Discussion Paper 9924, Human Development Network, World Bank, Washington, D.C. Disney, Richard F., and Edward Whitehouse. 2002. The Economic Well- Being of Older People in International Perspective: A Critical Review. Luxembourg Income Study Working Paper 306. Dolton, P. J., and G. H. Makepeace. 1987. Interpreting Sample Selection Effects. Economics Letters 24: 373­79. Donkers, Bas, and Arthur van Soest. 1999. Subjective Measures of House- hold Preferences and Financial Decisions. Journal of Economic Psy- chology 20: 613­42. Dulitzky, Daniel. 1998. Social Security Reforms, Retirement Plans, and Saving Under Labor Income Uncertainty. PhD diss., Massachusetts In- stitute of Technology, Cambridge, Mass. Easterly, William. 2001. The Elusive Quest for Growth. Cambridge, Mass.: MIT Press. ECH (Longitudinal Household Survey; Encuesta Continua de Hogares). 1999. La Paz, Bolivia. BIBLIOGRAPHY 301 ECH (Longitudinal Household Survey; Encuesta Continua de Hogares). 2000. La Paz, Bolivia. ECLAC/CELADE (Economic Commission for Latin America and the Caribbean/Demography Center for Latin America and the Caribbean (Centro Latinoamericano y Caribeño de Demografía). 1998. Latin America Population Projections 1970­2050. Demographic Bulletin 62 Santiago, Chile: ECLAC/CELADE. Edwards, Sebastian. 1986. Are Devaluations Contractionary? Review of Economics and Statistics 68 (3): 501­08. Edwards, Sebastian, and Alejandra Cox-Edwards. 2000a. Economic Re- forms and Labor Markets: Policy Issues and Lessons from Chile. Work- ing Paper 7646, National Bureau of Economic Research, Cambridge, Mass. ------. 2000b. Economic Reforms and Labor Markets: Policy Issues and Lessons from Chile. Economic Policy 30: 183­229. Edwards, Sebastian, and Nora Lustig. 1997. Labor Markets in Latin America: Combining Social Protection with Market Flexibility. Wash- ington, D.C.: Brookings Institution Press. EHM (Household Sampling Survey; Encuesta de hogares por muestreo). 2000. Caracas, Venezuela. EHPM (Multipurpose Household Survey; Encuesta de hogares de propósitos múltiples). 1998. San Salvador, El Salvador. EHPM (Multipurpose Household Survey; Encuesta de hogares de propósitos múltiples). 2000. San Salvador, El Salvador. Ehrlich, Isaac, and Gary Becker. 1972. Market Insurance, Self-Insurance and Self-Protection. Journal of Political Economy 80: 623­48. ------. 2000. Market Insurance, Self-Insurance, and Self-Protection. In Foundations of Insurance Economics: Readings in Economics and Fi- nance, eds. G. Dionne and S. Harrington. Boston: Kluwer Academic Publishers. Ehrlich, Isaac, and Hiroyuki Chuma. 1990. A Model of the Demand for Longevity and the Value of Life Extension. Journal of Political Econ- omy 98 (4): 761­82. Eichengreen, Barry, and Ashoka Mody. 1998. What Explains Changing Spreads on Emerging-Market Debt: Fundamentals or Market Senti- ment? Working Paper w6408, National Bureau of Economic Research, Cambridge, Mass. Elter, Doris. 1999. Sistema de AFP Chileno: Injusticia de un Modelo. San- tiago, Chile: LOM Publishing. ENAHO (National Household Survey; Encuesta Nacional de Hogares). 2000. Lima, Peru. 302 BIBLIOGRAPHY ENEU (National Urban Employment Survey; Encuesta nacional de em- pleo urbano). 1998. Mexico City, Mexico. ENEU (National Urban Employment Survey; Encuesta nacional de em- pleo urbano). 2001. Mexico City, Mexico. ENFT (National Laborforce Survey; Encuesta Nacional de la Fuerza de Trabajo). 2000. Santo Domingo, Dominican Republic. ENH (National Household Survey; Encuesta nacional de hogares). 1999. Bogotá Colombia. ENMV (National Living Standards Measurement Survey; Encuesta Na- cional sobre Medición de Niveles de Vida). 1999. Managua, Nicaragua. EPH (Permanent Household Survey; Encuesta Permanente de Hogares). 2002. Buenos Aires, Argentina. Epps, T. W., and K. Singleton. 1986. An Omnibus Test for the Two-Sam- ple Problem Using the Empirical Characteristic Function. Journal of Statistical Computation and Simulation 26: 177­203. Escobar, Federico. 2003. Pension Reform in Bolivia: A Review of Ap- proach and Experience. Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. Evans, D., and L. Leighton. 1989. Some Empirical Aspects of Entrepre- neurship. American Economic Review 79: 519­35. Eyzaguirre, N., and F. Lefort.1999. Capital Markets in Chile: A Case of Successful International Financial Integration. In Chile: Recent Policy Lessons and Emerging Challenges, eds. G. Perry and D. M. Leipziger. Washington, D.C.: World Bank Institute. Faccio, M., and M. A. Lasfer. 2001. Institutional Shareholders and Cor- porate Governance: The Case of UK Pension Funds. Working Paper 11/01, Center for Research on Pensions and Welfare Policies, Torino, Italy. Fafchamps, Marcel, and S. Lund. 2000. Risk-Sharing Networks in Rural Philippines. Oxford University. Processed. Feldstein, M. 1985. The Optimal Level of Social Security Benefits. Quar- terly Journal of Economics 100: 303­19. Feldstein, Martin, and Stephanie Seligman. 1981. Pension Funding, Share Prices, and National Savings. Journal of Finance 36 (4): 801­24. FIAP (International Federation of Pension Fund Administrators). 2002. Boletin No. 12. Santiago, Chile. Fields, Gary. 1974. Rural­Urban Migration, Urban Unemployment and Under-employment, and Job Search Activities in LDCs. Journal of De- velopment Economics 2: 165­88. ------. 1990. Labor Market Modeling and the Urban Informal Sector: Theory and Evidence. In The Informal Sector Revisited. Paris: OECD. BIBLIOGRAPHY 303 Fiess, Norbert. 2003. Pension Reform or Pension Default? A Note on Pension Reform and Country Risk. Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washing- ton, D.C. Fiess, Norbert, William F. Maloney, and Rashmi Shankar. 2000. The Informal Sector, Wage Rigidities, and Real Exchange Rates. Poverty Reduction and Economic Management Department, World Bank, Washington, D.C. Processed. Fiorito, Riccardo, and Flavio Padrini. 2001. Distortionary Taxation and Labor Market Performance. Oxford Bulletin of Economics and Statis- tics 63 (2): 173­96. Fischer, B. 1998. The Role of Contractual Savings Institutions in Emerging Markets. In Institutional Investors in the New Financial Landscape. Paris: OECD. Fox, Louise, and Edward Palmer. 2001. New Approaches to Multipillar Pension Systems: What in the World Is Going On? In New Ideas about Old Age Security: Toward Sustainable Pension Systems in the 21st Century, eds. Robert Holzmann and Joseph Stiglitz. Washington, D.C.: World Bank. Fuchs, Victor R. 1986. The Health Economy. Cambridge, Mass.: Harvard University Press. Gallego, F., and N. Loayza. 2000. Financial Structure in Chile: Macro- economic Developments and Microeconomic Effects. Paper presented at the World Bank Financial Structures and Economic Development Conference, Washington, D.C. Genicot, G., and D. Ray. 2000. Endogenous Group Formation in Risk Sharing Arrangements. Department of Economics Working Paper, New York University. Processed. Getler, Paul. 2000. Insuring the Economic Costs of Illness. In Shielding the Poor: Social Protection in the Developing World, ed. Nora Lustig. Washington, D.C.: Brookings Institution Press and the Inter-American Development Bank. Gill, Indermit, and Nadeem Ilahi. 2000. Economic Insecurity, Individual Behavior, and Social Policy. Office of the Chief Economist, Latin Amer- ica and the Caribbean Region, World Bank, Washington, D.C. Gill, Indermit, and Claudio Montenegro. 2001. Is Chile's Labor Market Serving the Poor as Well as It Does the Rich? In Crafting Labor Policy: Techniques and Lessons from Latin America, eds. I. Gill, C. Montenegro, and D. Domeland. London: Oxford University Press and the World Bank. Gill, Indermit, Truman Packard, Anita Schwarz, and Juan Yermo. 2001. Brazil: Critical Issues in Social Security. Country Study, World Bank, Washington, D.C. 304 BIBLIOGRAPHY Grandolini, Gloria, and Luis Cerda. 1998. The 1997 Pension Reform in Mexico. Policy Research Working Paper 1933, World Bank, Washing- ton, D.C. Gruber, Jonathan. 1995. The Incidence of Payroll Taxation: Evidence from Chile. Working Paper 5053, National Bureau of Economic Re- search, Cambridge, Mass. Gruber, Jonathan, and James M. Poterba. 1993. Tax Incentives and the Decision to Purchase Health Insurance: Evidence from the Self Em- ployed. Working Paper 4435, National Bureau of Economic Research, Cambridge, Mass. Guiso, L., and M. Paiella. 2000. Risk Aversion, Wealth, and Financial Markets Imperfections. Working Paper, Ente Luigi Einaudi for Mone- tary, Banking, and Financial Studies and the Bank of Italy Research De- partment, Rome. Haque, Nadeem Ul, Manmohan S. Kumar, Mark Nelson, and Donald J. Mathieson, 1996. The Economic Context of Indicators of Developing Country Creditworthiness. Working Paper 9619, International Mone- tary Fund, Washington, D.C. Harris, John R., and Michael P. Todaro. 1970. Migration, Unemployment and Development: A Two-Sector Analysis. American Economic Re- view 60 (March): 126­42. Hayashi, Fumio, Joseph Altonji, and Laurence Kotlikoff. 1996. Risk-Shar- ing Between and Within Families. Econometrica 64 (2): 261­94. Hemmer, Hans R., and C. Mannel. 1989. On the Economic Analysis of the Urban Informal Sector. World Development 17 (10): 1543­552. Hernández, L., and E. Walker. 1993. Estructuras de financiamiento cor- porativo en Chile (1978­1990). Public Studies 51, Center for Public Studies, Santiago, Chile. Hirshleifer, John, and John Riley. 1979. The Analytics of Uncertainty and Information--An Expository Survey. Journal of Economic Literature 17: 1375­421. Hoddinott, John. 1992. Rotten Kids or Manipulative Parents: Are Chil- dren Old Age Security in Western Kenya? Economic Development and Cultural Change 40 (3): 545­66. Hogan, John J. 1987. Pensions and the Self Employed. Irish Banking Review Summer: 18­26. Holzmann, Robert. 1997. Pension Reform, Financial Market Develop- ment, and Economic Growth: Preliminary Evidence from Chile. Work- ing Paper WP/96/34, International Monetary Fund, Washington, D.C. ------. 1998. Financing the Transition to the Multi-Pillar. Social Protection Discussion Paper 9809, World Bank, Washington D.C. BIBLIOGRAPHY 305 Holzmann, Robert, and Steen Jorgensen. 1999. Social Protection as Social Risk Management: Conceptual Underpinnings for the Social Protection Strategy Paper. Social Protection Discussion Paper 9904, Human Development Network, World Bank, Washington, D.C. Holzmann, Robert, Robert Palacios, and Asta Zviniene. 2001. Implicit Pension Debt: Issues, Measurement and Scope in International Per- spective. Pension Primer Paper, World Bank, Washington, D.C. Holzmann, Robert, Truman Packard, and Jose Cuesta. 2000. Extending Coverage in Multi-Pillar Pension Systems: Constraints and Hypotheses, Preliminary Evidence and Future Research Agenda. In New Ideas About Old Age Security, eds. R. Holzmann and J. Stiglitz. Washington, D.C.: World Bank. Holzmann, Robert, and Joseph Stiglitz, eds. 2001. New Ideas about Old Age Security: Toward Sustainable Pension Systems in the 21st Century. Washington, D.C.: World Bank. Hubbard, R. G., J. Skinner, and S. P. Zeldes. 1993. The Importance of Pre- cautionary Motives in Explaining Individual and Aggregate Savings. Working Paper 4516, National Bureau of Economic Research, Cam- bridge, Mass. ------. 1994. Expanding the Life-Cycle Model: Precautionary Savings and Public Policy. American Economics Association Papers and Pro- ceedings 84 (2): 174­79. Hurd, Michael. 1990. Research on the Elderly: Economic Status, Retire- ment, Consumption and Saving. Journal of Economic Literature 28 (2): 565­637. Hurd, Michael, Daniel McFadden, and Li Gan. 1998. Subjective Survival Curves and Life Cycle Behavior. In Inquiries in the Economics of Ag- ing, ed. David Wise. Chicago: University of Chicago Press and National Bureau of Economic Research. Hurd, Michael D., and Kathleen McGarry. 1997. The Predictive Validity of Subjective Probabilities of Survival. Working Paper 6193, National Bureau of Economic Research, Cambridge, Mass. IADB (Inter-American Development Bank). 2000. Social Protection for Equity and Growth. Washington, D.C.: Poverty and Inequality Advi- sory Unit, Sustainable Development Department, Inter-American De- velopment Bank. Iglesias, A. 1998. The Impact of Pension Reform on Capital Markets: The Chilean Experience. Paper presented at the Conference on Capital Market Development in Emerging and Transition Economies: Trends and Challenges, Washington, D.C. ------. 2000. Pension Reform and Corporate Governance: Impact in Chile. Revista ABANTE 3 (1): 109­41. 306 BIBLIOGRAPHY ILO (International Labour Organization). 1993. Resolution II Concerning Statistics of Employment in the Informal Sector. Presented at the 15th International Conference of Labour Statisticians, Geneva. www.ilo.org/public/english/120/res/infsec.htm. ------. 1996. Labor Overview: Latin America and the Caribbean, No. 3. Geneva: ILO. ------. 1998. Labor Overview: Latin America and the Caribbean, No. 5. Geneva: ILO. ------. 1999. Labor Overview: Latin America and the Caribbean, No. 6. Geneva: ILO. ------. 2000. World Labor Report. Geneva: ILO. Impavido, Gregorio, Alberto R. Musalem, and Thierry Tressel. 2001a. Contractual Savings, Capital Markets, and Firms' Financing Choices. World Bank Financial Sector Development Department, Washington, D.C. Processed. ------. 2001b. Contractual Savings Institutions and Banks' Stability and Efficiency. Policy Research Working Paper 2751, World Bank, Wash- ington, D.C. Impavido, G., A. Musalem, and D. Vittas. 2002. Contractual Savings in Countries with a Small Financial Sector. In Globalization and Finan- cial Systems in Small Developing Countries, eds. James Hanson, Patrick Honohan, and Giovanni Majnoni. Washington D.C.: World Bank. Insinga, Ralph. 1998. The Retirement Outlook for Chile's Independently Employed Workers: A Case Study of Attitudes and Saving Behavior. LC/R 1840, Economic Commission for Latin America and the Caribbean, Santiago, Chile. International Insurance Council. 1998. Latin America: An Insurance Ref- erence Guide. Washington, D.C.: International Insurance Council. International Monetary Fund (IMF). 2003. International Financial Statistics. Washington, D.C. James, E. 1996. Protecting the Old and Promoting Growth: A Defense of Averting the Old Age Crisis. Policy Research Working Paper 1570, World Bank, Washington, D.C. ------. 1997. Pension Reform: Is There a Tradeoff Between Efficiency and Equity? Policy Research Working Paper 1767, World Bank, Wash- ington, D.C. ------. 1999. Coverage Under Old Age Security Systems and Protection for the Uninsured: What Are the Issues? Paper presented at the Inter- American Development Bank Conference on Social Protection, Wash- ington, D.C. BIBLIOGRAPHY 307 James, E., A. Cox-Edwards, and R. Wong. 2003. The Gender Impact of Pension Reform: A Cross-Country Analysis. Policy Research Working Paper 3074, World Bank, Washington, D.C. James, E., G. Ferrier, J. Smalhout, and D. Vittas. 1998. Mutual Funds and Institutional Investments: What Is the Most Efficient Way to Set Up In- dividual Accounts in a Social Security System? Paper presented at the NBER Conference on Social Security, Bordeaux, France. James, E., J. Smalhout, and D. Vittas. 2001. Administrative Costs and the Organisation of Individual Account Systems: A Comparative Perspective. In Private Pension Systems: Administrative Costs and Reforms. Private Pension Series. Paris: OECD. Kamin, Steven B., and Karsten von Kleist. 1999. The Evolution and Determinants of Emerging Market Credit Spreads in the 1990s. In- ternational Finance Discussion Paper 653, Federal Reserve Board of Governors, Washington, D.C. Karpoff, J. M., P. H. Malatesta, and R. A. Walkling. 1996. Corporate Governance and Shareholder Initiatives: Empirical Evidence. Journal of Financial Economics 42: 365­95. Kleinjans, Kristin J. 2001. The Colombian Pension System After the Re- form of 1994: An Evaluation. Department of Economics, University of Pittsburgh. Processed. Knight, F. 1921. Risk, Uncertainty, and Profit. New York: Houghton- Mifflin. Kotlikoff, Laurence J., William Samuelson, and Stephen Johnson. 1988. Consumption, Computation Mistakes and Fiscal Policy. AEA Papers and Proceedings. American Economic Review 78 (2): 408­12. Krebs, Tom, and William F. Maloney. 1999. Quitting and Labor Turnover: Microeconomic Evidence and Macroeconomic Consequences. Policy Research Working Paper 2068, World Bank, Washington, D.C. Laboul, A. 1998. Private Pension Systems: Regulatory Policies. Ageing Working Paper 2.2, OECD, Paris. Laffont, Jean-Jacques. 1979. Aggregation and Revelation of Preferences. Amsterdam: North-Holland. Laibson, David I., Andrea Repetto, and Jeremy Tobacman. 1998. Self- Control and Saving for Retirement. Brookings Papers on Economic Activity 1: 91­196. La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. W. Vishny. 1998. Law and Finance. Journal of Political Economy 106 (6): 1113­155. Lasaga, Manuel, and John Pollner. 2003. Peru's Private Pension System: An Analysis of Its Evolution, Current Market Structure and Recom- mendations for Reform. Background Paper for Analytical and Advisory 308 BIBLIOGRAPHY Assistance to the Government of Peru, Social Protection Sector Unit, Latin America and the Caribbean Regional Office, World Bank, Wash- ington, D.C. Lawrence, Emily C. 1991. Poverty and the Rate of Time Preference: Evi- dence from Panel Data. Journal of Political Economy 99 (1): 54­77. Lazear, Edward P. 1986. Incentive Effects of Pensions. In Pensions, Labor and Individual Choice, ed. David A. Wise. Chicago: University of Chicago Press and National Bureau of Economic Research. Lefort, F., and E. Walker. 2000a. Pension Reform and Capital Markets: Are There Any (Hard) Links? Working Paper, Catholic University of Chile, Santiago. ------. 2000b. The Effects of Economic and Political Shocks on Corpo- rate Governance Systems in Chile. Revista ABANTE 2 (2): 183­206. Levenson, Alec R., and William F. Maloney. 1998. The Informal Sector, Firm Dynamics and Institutional Participation. Policy Research Work- ing Paper 1988, World Bank, Washington, D.C. Levine, R. 1997. Financial Development and Economic Growth: Views and Agenda. Journal of Economic Literature 32 (2): 688­726. Levine, Ross, Norman Loayza, and Thorsten Beck. 2000. Financial Inter- mediation and Growth: Causality and Causes. Journal of Monetary Economics 46(1):31­77 (August). Levine, R., and S. Zervos. 1998. Stock Markets, Banks, and Economic Growth. American Economic Review 88 (June): 537­58. Lewis, W. 1954. Economic Development with Unlimited Supply of Labor. Manchester School of Economics and Social Studies 22: 139­91. Lindbeck, Assar, and Mats Persson. 2003. The Gains from Pension Re- form. Journal of Economic Literature 41 (March): 74­112. Litchfield, Julie. 2001. Updated Income Distribution and Poverty Mea- sures for Chile: 1987­1998. Background Paper 1. In Poverty and In- come Distribution in a High Growth Economy: The Case of Chile 1987-1998. Washington, D.C.: World Bank. Lucas, Robert E. B., and Oded Stark. 1985. Motivations to Remit: Evi- dence from Botswana. Journal of Political Economy 93 (5): 901­17. Macias, Osvaldo, and Jorge Tarzijan. 1994. Origen y Magnitud de la Eva- sion en el Sistema de Pensiones Chileno Derivado de la Capitalizacion Individual. Fiscal Policy Series Economic Commission for Latin Amer- ica and the Caribbean/United Nations Development Programme, San- tiago, Chile. Maloney, William F. 1998a. The Structure of the Labor Market in Devel- oping Countries: Time Series Evidence and Competing Views. Policy Research Working Paper 1940, World Bank, Washington, D.C. BIBLIOGRAPHY 309 ------. 1998b. Are Labor Markets in Developing Countries Dualistic? Policy Research Working Paper 1941, World Bank, Washington, D.C. ------. 1999. Does Informality Imply Segmentation in Urban Labor Mar- kets? Evidence from Sectoral Transitions in Mexico. World Bank Eco- nomic Review 13 (2): 275­302. ------. 2000. Minimum Wages in Latin America: A Note. Poverty Re- duction and Economic Management Department, World Bank, Wash- ington, D.C. Processed. ------. 2003. Informality Revisited. Policy Research Working Paper 2965, World Bank, Washington, D.C. Maloney, William F. and Jairo Nuñez. 2001. Measuring the Impact of Minimum Wages: Evidence from Latin America. Policy Research Working Paper 2597, World Bank, Washington, D.C. Manski, C. 1989. Anatomy of the Selection Problem. Journal of Human Resources 24 (3): 343­60. Marcouiller, D.,V. Ruiz de Castilla, and C. Woodruff. 1997. Formal Mea- sures of the Informal Sector Wage Gap in Mexico, El Salvador, and Peru. Economic Development and Cultural Change 45 (2): 367­92. Marquez, Gustavo, and Carmen Pages. 1998. The Ties That Bind: Pro- tecting Employment and Labor Trends in Latin America. Paper pre- sented at the annual meeting of the Board of Governors, Inter-Ameri- can Development Bank, Cartagena, Colombia. Marshall, John R. 1976. Moral Hazard. American Economic Review 66: 880­90. Mastrángelo, Jorge. 1999. Políticas Para la Reducción de Costos en los Sistemas de Pensiones: El Caso de Chile. Financing Development Series. Santiago, Chile: CEPAL/ECLAC. Maturana, G., and E. Walker. 1999. Rentabilidades, Comisiones, y De- sempeño en la Industria Chilena de Fondos Mutuos. Estudios Públicos 73 (Summer): 293­334. Mesa-Lago, Carmelo. 1991. Social Security and Prospects for Equity in Latin America. Discussion Paper 140, World Bank, Washington, D.C. ------. 2000. Social Assistance, Pensions and Health Care for the Poor in Latin America and the Caribbean. In Shielding the Poor: Social Protection in the Developing World, ed. Nora Lustig. Washington, D.C.: Brookings Institution Press and Inter-American Development Bank. ------. 2001. La Cobertura de Pensiones de Seguridad Social en America Latina Antes y Después de la Reforma Previsional. Socialis: Revista Latinoamiricana de Política Social 4 (April): 17­27. 310 BIBLIOGRAPHY Min, Hong G., and Jung-goo Park. 2000. How the Republic of Korea's Financial Structure Affects the Volatility of Four Asset Prices. Policy Research Working Paper 2327, World Bank, Washington, D.C. Mincer, Jacob A. 1974. Schooling, Experience and Earnings. New York: Columbia University Press. Ministerio de Economía. 2000. Caracterización y Evolución del Gasto Público Social. Dirección de Gastos Sociales Consolidados, Secretaría de Política Económica, Ministerio de Economía, Buenos Aires. ------. 2002. Series de Gasto Público Consolidado por Finalidad-función (1980­2001). Dirección de Gastos Sociales Consolidados, Secretaría de Política Económica, Buenos Aires. Mitchell, Olivia S. 1998. Building an Environment for Pension Reform in Developing Countries. Social Protection Discussion Paper 9803, Hu- man Development Network, World Bank, Washington, D.C. Montenegro, Claudio. 2001. Wage Distribution in Chile: Does Gender Matter? A Quantile Regression Approach. Washington, D.C.: World Bank Development Research Group. Processed. Moody's Investor Service. 1998. Moody's Approach to Analyzing Pension Obligations of Corporations. Moody's Investor Service, New York, N.Y. Morande, Felipe G. 1992. Dynamics of Real Asset Prices, the Real Ex- change Rate, Trade Reforms and Foreign Capital Inflows: Chile, 1976­ 1989. Journal of Development Economics 39 (1): 111­39 Morande, Felipe, and Raimundo Soto. 1992. Una Nota sobre la Con- strucción de Series de Precios de Activos Reales: Tierra y Cases en Chile (1976­1989). Revista de Analisis Economico (Chile) 7: 169­77. Murgai, R., P. Winters, E. Sadoulet, and A. de Janvry. 2000. Localized and Incomplete Mutual Insurance. University of New England Work- ing Paper Series in Agricultural and Resource Economics 2000-5. New South Wales, Australia. Processed. Nesbitt, S. L. 1994. Long-Term Rewards from Shareholder Activism: A Study of the CalPERS Effect. Journal of Applied Corporate Finance 6: 75­80. OECD (Organisation for Economic Co-operation and Development). 1997. Insurance Guidelines for Economies in Transition. Paris: OECD Centre for Co-operation with the Economies in Transition. ------. 1998. Maintaining Prosperity in an Ageing Society. Paris. ------. 1999. Institutional Investors' Statistical Yearbook 1998. Paris. ------. 2001. Insurance Regulation and Supervision in Asia and Latin America. Paris. BIBLIOGRAPHY 311 Orszag, Peter, and Joseph Stiglitz. 2001. Rethinking Pension Reform: Ten Myths About Social Security Systems. In New Ideas About Old Age Se- curity: Toward Sustainable Pension Systems in the 21st Century, eds. R. Holzmann and J. Stiglitz. Washington, D.C.: World Bank. Packard, Truman. 1997. Adjustment, Migration and Economic Informal- ity in Latin America. M.Sc. Economics for Development, Extended Es- say, Queen Elizabeth House, University of Oxford, U.K. ------. 2001. Is There a Positive Incentive Effect from Privatizing Social Security? Evidence from Pension Reforms in Latin America. Policy Re- search Working Paper 2719, World Bank, Washington, D.C. ------. 2002. Pooling, Savings and Prevention: Mitigating the Risk of Old Age Poverty in Chile. Background Paper for the Regional Study on So- cial Security Reform, LAC Region, World Bank, Washington, D.C. Packard, Truman, Naoko Shinkai, and Ricardo Fuentes. 2002. The Reach of Social Security in Latin America and the Caribbean. Background Pa- per for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. Pages, Carmen, and Claudio Montenegro. 1999. Job Security and the Age Composition of Employment: Evidence from Chile. Washington, D.C. Inter-American Development Bank. Palacios, Robert J. 1996. Averting the Old Age Crisis: Technical Annex. Policy Research Working Paper 1572, World Bank, Washington, D.C. Palacios, Robert J., and Montserrat Pallares-Miralles 2000. International Patterns of Pension Provision. Social Protection Discussion Paper 0009, Human Development Network, World Bank, Washington, D.C. Paredes, Ricardo, and Jorge Nino. 2001. Proyecto Contra la Evasión, Re- bajas de Impuestos y Compensaciones. Economy and Administration 140 (April/May). Pemberton, James. 1997. Protecting the People from Themselves: Switch- ing from Universal to Means-Tested Pensions When Some People Are Myopic. Discussion Papers in Economics and Management 387, Uni- versity of Reading, U.K. ------. 1999. Pension Reform, Latin American Style: A National Free Lunch? Discussion Papers in Economics and Management 414, Uni- versity of Reading, U.K. Perry, Guillermo, and Luis Servén. 2003. The Anatomy of a Multiple Crisis: Why Was Argentina Special and What Can We Learn From It? Policy Research Working Paper 3081, World Bank, Washington, D.C. Pinera, Jose E. 1991a. El Cascabel al Gato: La Batalla por la Reforma Previsional. Santiago, Chile: Zig Zag. ------.1991b. Principios y Fundamentos del Sistema Privado de Pen- siones en Chile. In Análisis del Sistema Privado de Pensiones en Chile, 312 BIBLIOGRAPHY ed. Rosario Cruz Ovalle. Santiago, Chile: Ibero-American Congress of the Association of Pension Fund Adminsitrators. ------. 1995. Empowering Workers: The Privatization of Social Security in Chile. Cato Journal 15: 155­66. Poterba, James M. 1988. Are Consumers Forward Looking? Evidence from Fiscal Experiments. AEA Papers and Proceedings. American Eco- nomic Review 78 (2): 413­18. Poterba, James M., Steven F. Venti, and David A. Wise. 1994. Targeted Saving and Net Worth of Elderly Americans. American Economics Re- view Papers and Proceedings 84 (2): 180­85. Pradhan, Menno. 1995. Sector Participation Decisions in Labor Supply Models. Living Standards Measurement Study Working Paper 113, World Bank, Washington, D.C. PRIESO (Encuestas de Previsión de Riesgos Sociales). 2000. Social Risk Management Survey. Washington, D.C.: World Bank. ------. 2002. Social Risk Management Survey. Washington, D.C.: World Bank. Pritchett, Lant, and Lawrence H. Summers. 1993. Wealthier Is Healthier. Policy Research Working Paper 1150, World Bank, Washington, D.C. Queisser, M. 1998a. Pension Reform: Lessons from Latin America. Policy Brief 15, OECD Development Centre, Paris. ------. 1998b. The Role of Pension Funds in the Stabilization of the Do- mestic Financial Sector. Washington, D.C.: World Bank. Processed. ------. 1998c. The Second Generation Pension Reforms in Latin Amer- ica. Paris: OECD Development Centre. Rabin, Matthew, and Richard Thaler. 2001. Anomalies: Risk Aversion. Journal of Economic Perspectives 15 (1): 219­32. Ramos, J. 1999. La Experiencia de las Administradoras de Fondos de Pen- siones en el Financiamiento de las Empresas a través del Mercado de Capitales en el Perú. AFP Integra, Lima. Processed. Rees, H., and A. Shah. 1986. An Empirical Analysis of Self-Employment in the U.K. Journal of Applied Econometrics 1: 95­108. Reisen, H. 1997. Liberalising Foreign Investments by Pensions Funds: Pos- itive and Normative Aspects. Technical Paper 120, OECD Develop- ment Centre, Paris. Reisen, Helmut, and J. Bailliu. 2000. Do Funded Pensions Contribute to Higher Aggregate Savings? A Cross-Country Analysis. In Pensions, Savings and Capital Flows, ed. Helmut Reisen. Paris: OECD. Revilla, Ernesto. 2002. The Cost of Social Security Reform in Bolivia. IADB, Washington, D.C. Processed. BIBLIOGRAPHY 313 Robson, Martin T., and Colin Wren. 1999. Marginal and Average Tax Rates and the Incentive for Self Employment. Southern Journal 65 (4): 757­73. Rodriguez, L. Jacobo. 1999. In Praise and Criticism of Mexico's Pension Reform. Policy Analysis 340, Cato Institute, Washington, D.C. Rodrik, Dani. 2003. Growth Strategies. Kennedy School of Government, Harvard University, Cambridge, Mass. Processed. Rofman, Rafael. 1999. El Costo Laboral como Explicación del Desempleo en la Argentina: Un Análisis de los Efectos de las Reducciones en las Contribuciones Patronales sobre el Desempleo entre 1994 y 1999. Jor- nadas Argentinas de Población. ------. 2000. The Pension System in Argentina: Six Years After the Re- form. In Pension Reform Primer, eds. R. J. Palacios and E. Whitehouse. Washington, D.C.: World Bank. ------. 2002. The Pension System in Argentina: Learning the Lessons. Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. Rofman, Rafael, and Gustavo Demarco. 1999. Collecting and Transfer- ring Pension Contributions, Social Protection Discussion Working Pa- per 9907, World Bank, Washington, D.C. Rofman, Rafael, Gustavo Stirparo, and Pablo Lattes. 1997. Proyecciones del Sistema Integrado de Jubilaciones y Pensiones. Special Studies 12, Superintendency of Retirement and Pension Fund Administrators, Buenos Aires. Rojas, E. 1999. The Long Road to Housing Reform. Lessons from the Chilean Experience. Sustainable Development Department Best Prac- tices Series. Washington, D.C.: Inter-American Development Bank. Rosenzweig, M. R. 1988. Labor Markets in Low-Income Countries. In Handbook of Development Economics, Vol. 1, eds. H. Chenery and T. N. Srinvasan. Amsterdam: North-Holland. ------. 2001. Savings Behavior in Low Income Countries. Oxford Re- view of Economic Policy 17 (1): 40­54. Rubalcava, L., and O. Gutierrez. 2000. Políticas para canalizar mayores recursos de los fondos de pensiones hacia la inversion real en México. Financing Development Series. New York: United Nations. SAFJP (Superintendencia de AFJP). 2002. Memoria Trimestral 31. Buenos Aires. SAFP (Chilean Pension Fund Administrators' Superintendency; Superin- tendencia de Administradoras de Fondos de Pensiones de Chile). 1998. Santiago, Chile. ------. 1999. El Sistema Chileno de Pensiones. Santiago, Chile: SAFP. 314 BIBLIOGRAPHY ------. 2000. Buletin Estadistico del Sistema Chileno de Pensiones. San- tiago, Chile. Salomon Smith Barney. 1998. Private Pension Funds in Latin America. Salomon Smith Barney Latin America Equity Research. New York. Samwick, A. 1997. Discount Rate Heterogeneity and Social Security Re- form. Working Paper 6219, National Bureau of Economic Research, Cambridge, Mass. Schmidt-Hebbel, K. 1997. Pension Reform, Informal Markets and Long- Term Income and Welfare, Working Paper 04, Central Bank of Santiago, Chile. Schmidt-Hebbel, K. 1998a. Does Pension Reform Really Spur Productiv- ity, Saving, and Growth? Working Paper 33, Central Bank of Chile, Santiago, Chile. ------. 1998b. Pension Reform and Labor Market Adjustment. Processed. Schmidt-Hebbel, Klaus. 1999. Chile's Pension Revolution Coming of Age. (English version) In Gesetzliche Alterssicherung, Reformerfahrungen im Ausland: Ein systematischer Vergleich aus sechs Landern (Reform- ing the Pension System: What Germany Can Learn from Other Coun- tries). Cologne: Deutsches Institut für Altersvorsorge. Schmidt-Hebbel, K., and Luis Servén. 1996. Hacia una Menor Inflacion en Chile: Contraccion Monetaria Bajo Expectativas Racionales. In Analisis Empirico de la Inflacion en Chile, eds. F. Morande and F. Rosende. Santiago, Chile: Latin American Institute of Doctrine and So- cial Studies and Pontifica Universidad Catolica de Chile. Schreiber, Sven. 2001. Estimating the Fiscal Cost of the Minimum Pension Guarantee in Chile. Paper presented at the International Institute of Public Finance meeting, August 27­30, 2001, Linz, Austria. Schubert, R., M. Brown, M. Gysler, and H. W. Branchinger. 1999. Fi- nancial Decision-Making: Are Women Really More Risk-Averse? American Economic Review 89 (2): 381­85. Schwarz, A. 1998. Comment to the Shift to a Funded Social Security Sys- tem in Argentina. In Privatizing Social Security, ed. Martin Feldstein. Chicago: University of Chicago Press. Serrano, Carlos. 1999. Social Security Reform, Income Distribution, Fis- cal Policy and Capital Accumulation. Policy Research Working Paper 2055, World Bank, Washington, D.C. Shah, Hemant. 1997. Towards Better Regulation of Private Pension Funds. Policy Research Working Paper 1791, World Bank, Washington, D.C. Skully, M., and D. Vittas. 1991. Overview of Contractual Savings Institu- tions, Working Paper 605, World Bank, Washington, D.C. Smith, M. P. 1996. Shareholder Activism by Institutional Investors: Evi- dence from CalPERS. Journal of Finance 51: 227­52. BIBLIOGRAPHY 315 Snyder, James M., Jr., and Irene Yackovlev. 2000. Political and Economic Determinants of Government Spending on Social Protection Programs. Background Paper, World Bank Latin America and Caribbean Studies, Washington, D.C. Srinivas, P. S., and J. Yermo. 1999. Do Investment Regulations Compro- mise Pension Fund Performance? Evidence from Latin America. Latin America and Caribbean Region Viewpoint Series. Washington, D.C.: World Bank. ------. 2000. Risk Management Through International Diversification: The Case of Latin American Pension Funds. Philadelphia: Wharton School, University of Pennsylvania. ------. 2002. Risk Management Through International Diversification: The Case of Latin American Pension Funds. In Innovations in Retire- ment Financing, eds. O. S. Mitchell, Z. Bodie, P. B. Hammond, and S. Zeldes. Philadelphia: University of Pennsylvania Press for the Pension Research Council. Srinivas, P. S., Edward Whitehouse, and Juan Yermo. 2000. Regulating Private Pension Funds' Structure, Performance and Investments: Cross- country Evidence. Social Protection Discussion Paper 0113, World Bank, Washington, D.C. Stark, Oded. 1990. A Relative Deprivation Approach to Performance In- centives in Career Games and Other Contests. Kyklos 43: 211­27. Stark, Oded. 1995. Altruism and Beyond: An Economic Analysis of Transfers and Exchanges within Families and Groups. Cambridge, U.K.: Cambridge University Press. St. John, Susan, and Larry Willmore. 2001. Two Legs Are Better Than Three: New Zealand as a Model for Old Age Pensions. World Devel- opment 29 (8): 1291­305. Summers, Lawrence H. 1989. Some Simple Economics of Mandated Bene- fits. American Economic Review Papers and Proceedings 79 (2): 177­83. Taubman, Paul, and Michael L. Wachter. 1986. Segmented Labor Mar- kets. In Handbook of Labor Economics, Vol. 2, eds. O. Ashenfelter and R. Layard. Amsterdam: North-Holland. Taylor, Mark, P. 1996. Earnings, Independence or Unemployment: Why Become Self Employed? Oxford Bulletin of Economics and Statistics 58 (2): 253­65. Thaler, Richard H. 1994. Psychology and Savings Policies. American Eco- nomic Review Papers and Proceedings 84 (2): 186­92. Thoenissen, Christoph. 1998. Pensions, Longevity and the Pursuit of Leisure: An Analysis of Pension Reform and Ageing in an Overlapping Generations Model with Endogenous Labor Supply. Discussion Paper in Economics and Management 385, University of Reading, U.K. 316 BIBLIOGRAPHY Thomas, J. 1996. The New Economic Model and Labor Markets in Latin America. In The New Economic Model in Latin America and Its Impact on Income Distribution and Poverty, ed. Victor Bulmer- Thomas. London: Institute of Latin American Studies, University of London. Thompson, Lawrence H. 1998. Older and Wiser: The Economics of Pub- lic Pensions. Washington, D.C.: Urban Institute Press. Titelman, Daniel. 2000. Reformas al Sistema de Salud en Chile: Desafios Pendientes. Economic Commission for Latin America, Financing De- velopment Series, 104, Santiago, Chile. Tokman, V., ed. 1992. Beyond Regulation: The Informal Economy in Latin America. Geneva: World Employment Program, International Labour Organization. Tokman, Victor, and Daniel Martinez. 1999. Costo Laboral y Competi- tividad en el Sector Manufacturero de America Lanita, 1990­1998. Re- vista de la CEPAL 69 (December): 53­70. Torche, Aristides, and Gert Wagner. 1997. Prevision Social: Valoración Individual de un Beneficio Mandatado. Cuadernos de Economia 103: 363­90. Truglia, V. 2000. Can Industrialized Countries Afford Their Pension Sys- tem? Washington Quarterly (Summer): 201­11. ------. 2002. Sovereign Ratings and Aging Societies. Paper presented at Rosenberg Institute of Global Finance Conference on Financing Global Aging, Waltham, Mass. http://www.brandeis.edu/global/rosenberg_pa- pers/truglia_paper.pdf. Uthoff, Andras. 1997. Baja Cobertura de la Seguridad Social en America Latina: Un Problema de Incentivos o de Exclusion Social? Paper pre- sented at the Social Security, Micro-enterprise and Self Employment in Latin America seminar organized by World Bank, ILO, CEPAL, and International Association of Social Security, San Jose, Costa Rica. ------. 1998. Pension Funds, the Financing of Transition Costs and Fi- nancial Markets Development: Lessons from the Chilean Privatization Reform. Santiago, Chile: CEPAL/ECLAC. ------. 2001. La Reforma del Sistema de Pensiones en Chile: Desafios Pendientes. Economic Commission for Latin America, Financing De- velopment Series, 112, Santiago, Chile. Uusitalo, Roope. 1999. Homo Entreprenaurus? Working Paper 205, Gov- ernment Institute for Economic Research, Helsinki. Valdés-Prieto, Salvador, ed. 1997. The Economics of Pensions: Principles, Policies and International Experience. Cambridge, U.K.: Cambridge University Press. BIBLIOGRAPHY 317 ------. 1998. Administrative Costs in a Privatized Pension System. Paper presented at the Pensions Systems Reform in Central America Confer- ence, Cambridge, Mass. ------. 2002a. Improving Programs That Mandate Savings for Old Age. Background Paper for the Regional Study on Social Security Reform, World Bank, Washington, D.C. ------. 2002b. Justifying Mandated Savings for Old Age. Background Pa- per for the Regional Study on Social Security Reform, World Bank, Washington, D.C. ------. 2002c. Social Security Coverage in Chile, 1990­2001. Back- ground Paper for the Regional Study on Social Security Reform, World Bank, Washington, D.C. Van De Ven, Wynand P.M.M., and Bernard M. S. Van Praag. 1981. The Demand for Deductibles in Private Health Insurance. Journal of Econometrics 17: 229­52. Vittas, Dimitri. 1996. Pension Funds and Capital Markets. Financial Sec- tor Development Note 71, World Bank, Washington, D.C. ------. 1998a. Regulatory Controversies of Private Pension Funds, Policy Research Working Paper 1893, World Bank, Washington, D.C. ------. 1998b. Institutional Investors and Securities Markets: Which Comes First? Paper presented at the World Bank Conference on Development in Latin America and the Caribbean, San Salvador, El Salvador. ------. 1999. Pension Reform and Financial Markets. Discussion Paper 7, Harvard Institute for International Development, Cambridge, Mass. ------. 2002. Policies to Promote Saving for Retirement: A Synthetic Overview. Policy Research Working Paper 2801, World Bank, Wash- ington D.C. Von Gersdorff, Hermann. 1997. Pension Reform in Bolivia: Innovative Solutions to Common Problems. Policy Research Working Paper 1832, World Bank, Washington, D.C. Wahal, S. 1996. Pension Fund Activism and Firm Performance. Journal of Financial and Quantitative Analysis 31: 1­23. Walker, E. 1998. The Chilean Experience Regarding Completing Markets with Financial Indexation. Working Paper 29, Central Bank of Chile, Santiago. Walker, Eduardo, and Fernando Lefort. 2001. Pension Reform and Cap- ital Markets: Are There Any (Hard) Links? In Pension Reform Primer, eds. R. J Palacios and E.Whitehouse. Washington, D.C.: World Bank. ------. 2002. Pension Reform and Capital Markets: Are There Any (0) Links? Social Protection Discussion Paper. 0201, World Bank, Wash- ington, D.C. 318 BIBLIOGRAPHY Whitehouse, Edward. 2000. How Poor Are the Old? A Survey of Evidence from 44 Countries Social Protection Discussion Paper 0017, Human Development Network, World Bank, Washington, D.C. ------. 2001. Administrative Charges for Funded Pensions: Comparison and Assessment of 13 Countries. In Private Pensions Systems: Admin- istrative Costs and Reforms. Private Pensions Series. Paris: OECD. Williamson, John. 1999. What Should the Bank Think about the Wash- ington Consensus? Paper prepared as a background to the World De- velopment Report 2000. World Bank, Washington, D.C. Willmore, Larry. 2001a. Three Pillars of Pensions: Is There Really a Need for Mandatory Contributions? In OECD Private Pensions Conference, 385­97. Private Pension Series. Paris: OECD. ------. 2001b, Universal Pensions in Low-Income Countries. Paper pre- pared for a meeting of International Association of Pension Fund Su- pervisory Organisms, San Jose, Costa Rica. Processed. Wodon, Quentin. 2000. Poverty and Policy in Latin America and the Caribbean. Technical Paper 467, World Bank, Washington, D.C. Woidtke, T. 2001. Agents Watching Agents? Evidence from Pension Fund Ownership and Firm Value. Journal of Financial Economics 63: 99­31. World Bank. 1994. Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth. Washington, D.C.: World Bank and Ox- ford University Press. ------. 1996a. Argentina: Provincial Pension Reform Adjustment Loan Project. President's Report P6967, Loan 4116, Washington, D.C. ------. 1996b. Mexico: Contractual Savings Development Program Pro- ject. President's Report P7018, Loan 4123. Washington, D.C. ------. 1998. Mexico: Second Contractual Savings Development Pro- gram (CSDP II) Adjustment Loan Project. President's Report P7239, Loan 4343. Washington, D.C. ------. 2000. Nicaragua: Pension and Financial Market Reform Techni- cal Assistance Credit Project. Project Appraisal Document, Report 19680, Credit 3344. Washington, D.C. ------. 2001. Poverty and Income Distribution in a High Growth Econ- omy: The Case of Chile 1987­1998. Report 22037-CH, Poverty Reduction and Economic Management Sector Unit, Latin America and the Caribbean Regional Office, Washington, D.C. ------. 2002. Brazil: The New Growth Agenda. Volume I: Policy Brief- ing. Report 22950-BR, Brazil Country Management Unit, Latin Amer- ica and the Caribbean Region, Washington, D.C. BIBLIOGRAPHY 319 Yamada, G. 1996. Urban Informal Employment and Self-Employment in Developing Countries: Theory and Evidence. Economic Development and Cultural Change 44: 289­314. Yermo, Juan. 2001. Simulation of Minimum Pension Distributions Based on PRIESO Survey Data. Background Note for the Regional Study on Social Security Reform, World Bank, Washington, D.C. ------. 2002a. Pension Reform and Capital Market Development. Back- ground Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. ------. 2002b. The Performance of Funded Pension Systems in Latin America. Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. ------. 2002c. Delivering Promises in the Chilean Funded Pension Sys- tem. Background Paper for the Regional Study on Social Security Re- form, LAC Region, World Bank, Washington, D.C. Zurita, Salvador. 1994. Minimum Pension Insurance in the Chilean Pen- sion System. Revista de Analisis Economico 9 (1): 105­26. Zviniene, Asta, and Truman Packard. 2002. A Simulation of Social Secu- rity Reforms in Latin America: What Has Been Gained? Background Paper for the Regional Study on Social Security Reform, LAC Region, World Bank, Washington, D.C. Index Note: f indicates figures, n indicates notes (nn more than one note), and t indicates tables. A and contribution rates, 194n11, accumulation stage 232, 264n3 defined contribution formula and investment choices, 157, for, 31 159­60, 159f7.12 and life insurance industry, 246, see also elderly; old age 257 agency risk, 127b7.1, 129, 237, and return rates, 130­43 261­62 administrative costs, 32­33, American depository receipts 272­73 (ADRs), 60, 72, 87n6 commissions to cover, 144 annuities funded systems, 130 inflation-indexed, 245­46 intergenerational impact of, 148, and life insurance industry, 162n16 256 Peru, 234f10.1 and longevity risk, 199, 224n1, reforms to decrease, 232­39, 245 264nn7­10 and personal pension plans, see also commissions and 250­51b10.2 commission rates scheduled withdrawal, 150, 151, admistradora de fondos de 151f7.4 pensiones. see pension Argentina, 70, 82, 103, 105n7, 214 fund administrators (AFPs) and bond market, 76, 83 ADRs. see American depository Capital Markets Law, 65b4.1 receipts (ADRs) commission structure, 162n10 adverse selection, and annuities and coverage rates, 99, 100 market, 152, 163n21 features of reforms in, AFORE system, 208b9.1 23­25t2.1 AFPs. see pension fund and fiscal issues, 5, 44, administrators (AFPs) 47­48b3.2, 262 age, 187 funds portfolio, 38n12 and accumulated wealth, 201, inflation-indexed securities, 204f9.2 74­75b4.2 321 322 INDEX Argentina (continued) B investments Banco Hipotecario, 87n10 choices of, 160 banking sector in government bonds, 87n3 investment of pension funds in, in pension funds, 59­60, 80 67 returns on, 140 pension funds impact on and risk, 132, 155 efficiency and stability of, and IPDs, 40f3.1, 41 78­80 mandatory savings plans, reforms to, 66b4.1, 76­80, 31­32 87nn9­11 pensions return on deposits, 136, 138, AFPs, 33, 239 140f7.4 choices, 4, 5 bankruptcies, insurance companies, deficits, 41, 42t3.1, 43­44, 257, 258f10.6 43f3.2, 54n3 Barr, Nicholas, 11, 49, 99, 121 guarantees, 33, 211 and contribution density, 101 noncontributory poverty and economic growth, 46 prevention, 216t9.3 on macroeconomics of pensions, and policy risk, 133­35b7.3, 110­11b6.1 272 and tier vs. pillar plans, 15n1 portfolios, 137­38b7.4, 138­39 Barrientos, Armando, 167, 170 ratio of variable commission Becker, Gary, 109, 113b6.2, 114, rates to contributions, 146, 123n1 147f7.6 benefits reforms forms of, 150, 151t7.4, 151­53, reversal of regressive nature 154f7.9, 163n21 of, 90­91 link to contributions, 22, 98­99 to social security, 20­21 relationship to participation taxonomy of, 22, 28, 37n7 rates, 103, 104t5.2 women and reform gains, relationship to wages, 164n22 91 Bertranou, Fabio, 225n7 switch to PAGY systems, 15n6 Bismarkian PAYG systems, 225n3 transition debt, 69­70 Bolivia, 8, 30, 76, 83, 145 venture capital and real estate BONOSOL program, investment, 84 210­11b9.2, 271 voluntary savings plans, 35t2.4 cash-flow gap, 45f3B1 Arrau, P., 237 commission structure, 144, assets and asset classes 162n11 data on, 175 coverage rates, 99 and investment of pension distribution choices, 152, 153 funds, 59­60, 61t4.1 fiscal abuse by government, 5 management of, 70­71 impact of reforms on women, Averting the Old Age Crisis, 1, 2, 3 92 and coverage rates, 97­98b5.2 interest rate spreads, 79, 79f4.2 and economic growth, 46, 49 investment of pension funds, pillars of, 278b11.1 59­60 and reliance on tax base, 223 and IPDs, 40f3.1, 41 INDEX 323 pension assets, 62, 62t4.3 international, 137­38b7.4 pension deficits, 41, 42t3.1, links to mandatory savings, 119 43f3.2 Capital Markets Law, 60, 65b4.1 pension fund managers, 33, 68 capital reserves, 58 portfolios, 62t4.2, 137­38b7.4, Caribbean, and pension funds, 71 138­39 Carranza, Eliana, 233 poverty prevention, 207 cash-flow gap, Bolivia, 45f3B1 reforms to social security in, centralization in pension 20­21, 22, 23­25t2.1, 28 administration, 238 returns on investments, 131 charge ratio, 143­44, 145, 146 transition costs, 44, 45­46b3.1 cumulative, 149, 162n18 transition debt, 69, 70 relationship to wages, 149, and transparency, 87n3 149f7.8, 150 voluntary savings plans, 35t2.4 at retirement, 146­47, 148f7.7, BOLIVIDA, 211b9.2 162nn12­15 bond market, 73 children dollar-linked, 141 child care, 176­77b8.1, 183 interest rate volatility of, 257 influence on contribution rates, see also government bonds 174­75 bond swaps, 134b7.3 as substitute for investment in BONOSOL program, 210­11b9.2, pension systems, 188, 194n13 271 Chile, 54, 69, 70, 83, 143 Brazil banking sector, 76 capital market development, 246 bond market, 73, 75, 79­80 inflation-indexed securities, and charge ratio, 145 74­75b4.2 commissions investment choices, 249­50, ratio of variable rates to 250­51b10.2 contributions, 146, mutual fund industry, 251­52, 147f7.6 254 structure of, 8, 144, 149, noncontributory poverty 162n11 prevention pensions, 216t9.3 contribution density, 101, reforms to retirement system, 180­81f8.1 37n5 and corporate governance role role of financial systems in of AFPs, 71, 72 savings plans, 250­51b10.2 and cost of universal pensions, and stock market liquidity, 81, 222 82 coverage rates, 99­101 distribution choices, 151, C 152­53, 154f7.9, 164nn22 capital accounts, liberalization of, household surveys, 175 82 increase in savings, 119 capital inflows, 132 inflation-indexed securities, capitalization, limits on, 71 74­75b4.2 capital markets insurance sector, 246, 257 development of, 85, 246, intercohort differences in 278b11.1 returns, 135­36, 139f7.3 324 INDEX Chile (continued) reforms interest rate spreads, 79, 79f4.2, features of, 22, 23­25t2.1, 80 28, 38n8 investments fiscal impact of, 44, choices of, 213 47­48b3.2 in corporate bonds, 87n7 gains for women, 91 pension funds, 59, 60, 131 impact on economic growth, returns, 131, 140 49 risk management, 155­60, reversal of regressive nature 242, 243­44t10.3 of, 90­91 and IPDs, 40f3.1, 41 tax reform, 66b4.1 life insurance industry, 259­60 returns on deposits, 140f7.4 and management fees, 146, 147, and risk, 8, 181­83, 194n12 148f7.7 shareholders' rights, 65b4.1 multiple fund system in, 242, Social Risk Management Survey, 264n11 175, 176­77b8.1 mutual fund industry, 254, and stock market liquidity, 82 255f10.4, 259­60, 261 switch to PAGY systems, 15n6 new funds in, 157­58, 157t7.5, venture capital and real estate 163n24 investment, 84 participation choice, 177 voluntary savings plans, 34, participation rates, 15n5 35t2.4, 161n1, 248, PASIS benefits, 225n7 248b10.1, 249, 258­60 pensions Chilean Association of Pension and AFPs, 5­6, 33, 58 Funds, 71 assets, 60, 62, 62t4.3 Cia. de Minas Buenaventura, 82 choices, 4, 5 civil servants, 29 comparison of reformed clearing and settlement systems, pension systems with 64b4.1 Peru, 175, 178t8.2 Cobb-Douglas production function, deficits, 41, 42t3.1, 43, 49 43f3.2 Colombia, 29, 70, 99 impact on stock market, 80 commission structure, 144 minimum pension guarantees, corporate bond market, 83 53, 207, 211, 219 and cost of universal pensions, noncontributory poverty 222 prevention, 216t9.3 creditors' rights, 65b4.1 occupational pension plans, distribution choices, 152­53 237 gains in rates of return, 91 and real returns, 136, 138, government debt markets, 72 140f7.4 inflation-indexed securities, 73, value of annuitizing pension 74­75b4.2 fund balances, 245 interest rate spreads, 79, 79f4.2 portfolios, 137­38b7.4, 138­39 and investment of pension and private sector securities funds, 59­60 markets, 63 and IPDs, 40f3.1, 41 private sector workers, 30 mandatory savings plans, 31 INDEX 325 minimum pension guarantees, life-cycle consumption and 211, 219 savings theory, 194n12 and mortgage bond market, 78 consumption smoothing, 15n9, pension choices, 4, 5 229, 271­72 pension deficits, 41, 42t3.1, 43, goal of, 269 43f3.2 and minimum pension pension fund managers, 33 guarantee, 212, 213 and portfolio risk, 137­38b7.4 and risk management of poverty prevention, 207 mandated savings, 239, reforms to social security in, 242­47, 264n11­12 20­21, 23­25t2.1, 28 contribution density, 177, voluntary savings plans, 34, 193nn7­9 35t2.4 Peru, 184­85f8.4, 185­86 commissions and commission rates, relationship to gender, 179­81, 8, 32­33, 273 193­94nn10­11 caps on, 233­34, 239, threshold for, 179, 180 240­41t10.2 contribution rates, 12, 13, 277 as cost to retirement accounts, in accumulation stage, 31 143, 162n9 age-related, 232, 264n3 flat, 149 coverage gap, 186 insurance products, 260 earnings-linked, 32t2.3, 232, mutual funds, 254, 255f10.4, 264n3 265n17 and eligibility threshold, ratio of to total contributions, 190­91, 194n15 144, 145t7.3, 145­46, employer, 43, 55n4 146f7.6 gender-related, 193n8 reductions in, 12 link to benefits, 22, 96, 98­99 reforms to decrease, 232­39, Peru, 6 240­41t10.2, 264nn7­10 of poorer workers, 211­12 regressive nature of structure, ratio of variable commissions to 149­50, 149f7.8 total contributions, 144, regulation of, 144, 162n10 145­46, 145t7.3, 146f7.6 relationship to pension fund relationship of participation performance, 150 level to, 101, 103, 105nn7­8 transparency in structure of, 260 requirements of, 22, 183­84, variable, 144, 145, 145t7.3 190 competition, 239, 253 and risk aversion, 191, compliance, with reform, 43, 55n4 194­95n16 comprehensive insurance, 15n9, 112, Corbo, Vittorio, 49, 54 113b6.2, 115, 123, 123n1 corporate bond market, consumer price index, and pension development of, 81­82 subsidies, 31 corporate governance, 58, 65b4.1, consumption 119 and contributions to pension Corsetti, G., 99 funds, 168 Costa Rica of elderly, 219 commission structure, 144 inter-temporal, 247, 264­65n13 corporate bond market, 84 326 INDEX Costa Rica (continued) demographics investments changes in, 19­20 choices of, 160 and contribution behavior, investment risk, 155 170­75, 193nn5­6 of pension funds, 59­60, 60 disability insurance, 31, 38n11, in private sector, 67 38n13, 105n4 mandatory individual retirement distribution stage accounts, 237 and performance of new funded noncontributory poverty pillar, 150­53, 154f7.9, prevention pensions, 216t9.3 162­63nn19­22 and pension guarantees, 33 and risk management, 245­47 portfolio floors, 62t4.2 diversification, 135, 138 reforms in, 4, 20­21, 26­27t2.2, of assets, 7 29 in domestic equities, 139­40, taxonomy of reform, 22, 28, 142t7.2 37n7 in foreign equities, 140, 142t7.2 voluntary savings plans, 34, international, 159­60 35t2.4 and risk, 9, 242, 243­44t10.3 country risk, 53 dollar-indexed bonds, 78 coverage ratios, 5­7 Dominican Republic Cox-Edward, Alejandra, 91, 101, mandatory savings plans, 31­33 193n8 minimum pension guarantees, 211 creditors' rights, 65b4.1 private sector workers, 30 credit-to-GDP ratio, 76 reforms to social security in, cross-country comparisons, 54n1 20­21, 26­27t2.2, 28 Cuenta 2 plan, 249, 249f10.2, 259 self-employed workers in, 238 Cuesta, Jose, 167, 170 subsidies for savings, 247 Cuota Social, 149, 208­09b9.1 custodian services, 64b4.1 E Czech Republic, 265n14 EAP. see economically active population (EAP) D earnings DB. see defined-benefit (DB) system and contribution rates, 32t2.3, DC. see defined contribution (DC) 232, 264n3 system loss of capacity for, 110b6.1, death, insurance against risk of, 111, 189 38n13 maximum taxable, 12, 15n10 debt markets, 246 relationship to pension income, default risk, 7, 131, 140, 262 94f5.3 defined-benefit (DB) system, 4, 9, replacement of, 11­12, 276 12, 15n6, 31 Easterly, William, 49 and life expectancy, 129, 161n4 Eastern Europe, multipillar reforms defined contribution (DC) system, in, 2 4, 21, 129 economically active population (EAP) demand-side measures, to reduce and participation rates, 99, l, costs, 234, 240­41t10.2 100f5.4, 105nn4­5 Demirguç-Kunt, Asli, 81 pension system coverage, 6f1.1 INDEX 327 economic development investment in private sector, effects of pension reform on, 46, 67 49­50, 51 and IPDs, 40f3.1, 41, 52 and variation in coverage, 274, mandatory savings plans, 274f11.1, 275 31, 33 economies of scale, and mutual minimum pension guarantees, fund management, 254, 211 256f10.5 pension deficits, 41, 42t3.1, 43, Ecuador, reforms in, 20­22, 43f3.2 26­27t2.2, 28, 29, 37n3, pension fund managers, 33 37n7 and portfolio risk, 137­38b7.4 education, 183 poverty prevention, 207 Edwards, Sebastian, 101 reforms, 20­22, 23­25t2.1, 28, Ehrlich, Isaac, 109, 113b6.2, 114, 30 123n1 impact on women, 92 elderly transition debt, 69 accumulated wealth of, 201, voluntary savings plans, 34, 204f9.2 36t2.4 consumption of, 219 Emerging Market Bond Index, 51, and the Cuota Social, 52, 52f3.3 208­09b9.1 employers growth in share of, 20, 20­21f2.1 contributions to pension plans, impact of population increase on 43, 55n4 pension systems, 200f9.1 employer-sponsored retirement instruments of income security, programs, 251 10t1.1 and pension plans, 234, 237­38, pension coverage, 7f1.2, 103 264nn6­8 poverty prevention among, employment, and contribution 206­11 behavior, 170­75, 193nn5­6 role of government in, enforcement, of contribution 199­205, 224­25nn1­3 mandates, 99 poverty rate when measured by Escobar, Federico, 44 current income, 200­01, Estonia, 238 202­03t9.1, 224n2 explicit debts, 51, 55n7, 132 social insurance coverage of, converted from IPDs, 44, 47b3.2 95­96b5.1 and NDCs, 128b7.2 see also old age extended family, and old-age Electronic Stock Exchange, 70 security, 168­69, 192­93nn3 eligibility threshold, 188, 190­91 El Salvador, 4, 29, 99, 222 F bond market, 75 fee structure, 247 commission structure, 144 Mexico, 260­61, 265n18 corporate bond market, 84 mutual funds, 254, 255f10.4, demographics, 19 265n17 distribution choices, 151, 152, fertility rate, 20f2, 1 153, 164n22 Fiess, Norbert, 44, 51 government debt markets, 72 financial literacy, 247, 260 328 INDEX financial sector and contribution histories, and decisionmaking in 184­86, 193n8 oligopolistic industry, and disparities in rates of return, 68­69 92f5.2, 93­95 importance of reforms in, impact of reforms on, 92, 92f5.2 64­66b4.1 and investment choices, 157 investments in, 63, 63f4.1, 67, Germany, 265n14 67t4.4, 131, 132f7.1 global factors, and reforms in Latin modernization of, 64b4.1, 85 America, 53, 55n11 and policy risk protection, government 133­35b7.3 and BONOSOL program, role in savings plans, 210­11b9.2 250­51b10.2 credibility of, 169 and transparency of costs and and the Cuota Social, 208­09b9.1 benefits of, 247 fiscal issues, 5, 7­8, 262­63 first-pillar pensions and fund withdrawals, 231 argument for, 121, 122 and investment risk, 245 objective of, 9, 37n4 and macroeconomic arguments vs. pillar zero, 205­06 for mandatory savings, see also defined benefit (DB) 116­21 system; poverty prevention management of centralized fiscal impact of reform, 40­44, pension systems, 238­39, 58nn2­3, 59n4 264n10 Argentina, 44, 47­48b3.2 and market insurance, 113b6.2 Chile, 44, 47­48b3.2 pension deficits financed by, 41, simulations of, 44­46, 47b3.2 42t3.1, 43 flat pensions, 206­07, 209­11 regulation of mutual funds, 253 Fondo 1, 156­58 role in investment choices, 260 Fondo 2, 156­58 role in pillars, 9, 10t1.1, 15n8 foreign assets, 140 role in poverty prevention, foreign currencies, 59­60 110­11b6.1, 199­205, foreign debt, 72 224­25nn1­3 foreign investments, 59­60, government bonds, 139 138­39, 242 AFPs as purchasers of, 119, foreign investors, 81­82, 87n12 121 formal sector, 96, 98, 99, 174, dollar-linked, 141 193n6 effects of pension funds on fraud, rise in fraudulent claims, market for, 75­76 45­46b3.1 inflation-indexed, 76, 155 Fuentes, Ricardo, 170, 193n5 investment in, 59­60, 131, fund management industry, 22 132f7.1, 263 and NDC, 127, 128b7.2, 129 G and policy risk, 132 GDP. see gross domestic product wage-indexed, 128b7.2 (GDP) government debt gender default risk, 140 and contribution density, investment in, 63, 63f4.1, 67, 179­81, 193­94nn10­11 67t4.4, 134b7.3 INDEX 329 and leasing bonds, 77 Mexico, 52 rating of, 141, 162n8 and NDCs, 129 weakness of markets for, 72­75 relationship with country risk, gross domestic product (GDP) 51, 55n7 credit as share of, 76 implicit-to-explicit debt conversion, IPDs as share of, 40f3.1, 41, 51 269 improvidence arguments, 116, 152, mutual fund assets as share of, 162­63n19 252 incentives, 12 pension assets as share of, 60, improvement of, 189, 194n14 62, 62t4.3, 71, 270 and participation rates, 101, pension debt as share of, 4, 44, 102t5.1, 103 48, 48f3B.2b and pension guarantees, 207 pension-related deficits as share and voluntary savings plans, 34, of, 45f3B.1 249­50 relationship to Cuota Social, income 208­9b9.1 and disparities in rates of return, stock market capitalization as 92f5.2, 93­95 share of, 80 household, 168 guarantees, 22 and investment choices, 159­60, eligibility threshold for, 188 159f7.12 and insurance company per capita, 105, 105n6 offerings, 246, 257 poverty rate among elderly when see also minimum pension measured by income, guarantees 200­01, 202­03t9.1, 224n2 replacement of, 201, 204, 225n3 H retirement, 126­30 Harrod-Domar growth models, 49 risks to old-age income security, Holzmann, Robert, 41, 44, 80, 126­30, 161nn2­4 167, 170 see also wages households indexation unit, 71­75 and contribution behavior, individual behavior 170­75, 193nn5­6 as argument for mandatory income, 168 savings, 115­16 housing, 129 characteristics relative to financing of, 77­78, 87n9 contribution behavior, as investment choice, 188, 170­75, 193nn5­6 194n13, 230­31 conclusions about, 166 human capital, 231 implications of income security preferences made by, 189­91, I 194­95nn14­16 Iglesias-Palau, A., 71, 80 and low participation rates, Impavido, Gregorio, 78 166­70, 192­93nn3­4 implicit pension debts (IPDs), 40, individualization of social security, 40f3.1, 41, 50, 269 4, 11, 114 converted to explicit debt, 44, individual retirement accounts, 89, 47b3.2 90, 128b7.2, 129, 208b9.1, defaulting on, 132, 135 237 330 INDEX industries, and contribution investments and investment choices, behavior, 170­75, 193nn5­6 272 inflation, 128b7.2, 129, 135b7.3 Brazil, 249­50, 250­51b10.2 inflation-indexed government caveats to, 85­86 bonds, 76, 155, 257 and centralized pension and NDC, 127, 128b7.2, 129 administration, 238, inflation-indexed liabilities, 153 264nn9­10 inflation-indexed securities, 60, 73, and corporate bond markets, 74­75b4.2 81­82 Argentina, 134b7.3 and financial literacy, 247 and investment risk, 242, 245 government bonds, 59­60, 131, private annuities, 245­46 132f7.1 INFONAVIT, 31 government debt and financial informal sector institutions, 67, 67t4.4 employment in, 193n6 government role in, 260 equity implications of, 93­96 and guarantees, 8 and social insurance coverage, home ownership, 188, 194n13, 96b5.1, 97­98b5.2, 98, 230­31 105n2 human capital, 231 infrastructure bonds, 83 and longevity risk management, institutional investors, 63, 67, 78 154­60, 163n24 institutional structures, and and moral hazard, 213­14 coverage, 99, 105n3 restrictions on, 59­60, 87nn3­6 insurance industry, 66b4.1 returns from, 130­43 comprehensive insurance, and risk management, 242, 245 15n9 and stock market liquidity, and pension assets, 62­63 80­83, 87n12 and personal pension plans, and transition debt, 70, 87n7 250­51b10.2 IPDs. see implicit pension debts rates of return, 246 (IPDs) interest-bearing assets, 131 Ireland, 239 interest rate spreads, 79, 79f4.2 intergenerational issues, 162n16, J 273 James, E., 91, 99, 167 administrative costs, 148, 162n16 L commission rates, 8 labor force intermediation, 230, 238 and choice of pension systems, internal rates of return, 92, 92f5.2 187 International Labour Organization, and contribution density, 180 214, 215, 215t9.2 and coverage issues, 96­103, investment floors, 60 105nn2­8 investment risk, 126, 127b7.1, 129 and the Cuota Social, 208­09b9.1 management of, 154­60, 242, decreases in, 19, 20f2.1 243­44t10.3 and improved incentives, 49 and occupational savings plans, new entrants into, 28­29, 30, 264n7 38n9 INDEX 331 participation in pension systems, long-term government bonds, 125, 177 128b7.2 and perceived benefits and loyalty discounts, 144 contribution link, 98­99 lump-sum distributions, 150, and Peru's reformed pension 162­63n19 system, 177, 179b8.2 relationship of participation M rates to mandated macroeconomics contributions, 101, 103, and arguments for mandating 105nn7­8 savings, 116­21 see also workers of pensions, 110­11b6.1 La Porta, R., 65b4.1 and portfolio performance, 141 Lasaga, Manuel, 233 and structural pension reform, leasing bonds, 83 50­54, 55nn6­7, 55nn9­11 Lefort, F., 70, 71, 80 management fees, 7, 8, 146, 147, legislation 148f7.7 and coverage rates, 99 mandatory individual retirement and participation rates, 103 accounts, Costa Rica, 237 letras hipotecarias, 77­78 mandatory pension systems Levine, R., 78, 81 administrative costs of, 232­39 licensing process, insurance burden of, 125­26 companies, 257 destination of contributions, life-cycle consumption and savings 29f2.2 theory, 194n12 mandatory private savings, 5, 14, life cycle needs, meeting of, 22 229­32, 263­64nn1­4 mandatory savings plans, 1, 8, 13, life expectancy, 12, 92­93, 128b7.2 272 and choices at distribution stage, contribution rates and earnings 150­52, 163n20 ceilings, 32t2.3 cost of increases in, 129, 161n4 individual welfare argument for, impact of increase in on pooling 115­16 pension systems, 200f9.1 and life cycle needs, 229­32, information on, 246 263­64nn1­4 rise in, 20­21f2.1 macroeconomic argument for, life insurance industry, 31, 246, 116­21 256­57, 258f10.6 and mutual funds, 265n15 Lima, Peru, PRIESO survey, regulation of, 57­60, 86n1, 184­85f8.4, 185­86 87nn3­6 Lindbeck, Assar, 114 role of, 277, 278­79b11.1 longevity risk, 12, 38n15, 92­93 size of, 12, 227­28, 276­77 and annuities, 199, 224n1 structure and implementation of, and investments, 154­60, 30­33, 38nn10­13 163n24 market concentration, and pension management of, 129, fund managers, 68t4.5 243­44t10.3, 245 marketing, 143 overview, 127b7.1 market insurance, 113b6.2, 114 protection against, 128b7.2 market value of assets, 33, 38n12 332 INDEX Mastrángelo, Jorge, 143, 150 traditional security systems of, Maturana, G., 254 174 maximum taxable earnings, 12, transition debt, 69, 70 15n10 valuation mechanisms, 59, 87n4 means testing, 207, 209 voluntary savings plans, 34, Mesa­Lago, Carmelo, 44 36t2.4, 161n1 Mexican Stock Exchange, 65b4.1 microeconomics, and contribution Mexico, 59, 72, 83, 99 density, 180 bond market, 75­76 military personnel, 29 calculation of fees, 260­61, minimum pension guarantees, 8, 53, 265n18 194n15, 211­14, 225nn4­5 capital market development, 246 Chile, 53, 207, 211, 219 civil servants pension plan, 29 and contribution density commission structure, 144, 149, threshold, 179, 193­94n10 162n11 contributory, 211­14, 225nn4­5 creditors' rights, 65b4.1 and high-risk portfolios, 59, Cuota Social, 208­09b9.1 87n5 distribution choices, 153 mandatory funding protection financial market infrastructure, by, 33 64b4.1 Peru, 271 fiscal burden of pensions, 5 and risk pooling, 224 housing loans, 87n10 Moody's, 50, 55n6 increase in country risk, 51­52, moral hazard 52f33.3 and benefit guarantees vs. inflation-indexed securities, 73, individual choice, 8 74­75b4.2 and distribution choices, 152 investments and fund withdrawals, 231 in mutual funds, 84 and investment choices, 213­14 of pension funds, 60 of overly generous social returns on, 131, 140 assistance, 207 and IPDs, 40f3.1, 41 Peru, 160 life insurance policyholders, 257 relationship to years of mandatory savings plans, 31­32, contribution, 190 33 Morgan Stanley Capital pension deficits, 41, 42t3.1, 43, International Europe, 43f3.2 Australia, Far East index, 140 pension guarantees, 33, 87n5 Morón, Eduardo, 233 portfolios, 62t4.2, 137­38b7.4, mortality rates, 92­93, 168 138­39, 162n7 mortgage bonds and securities, 60, poverty prevention, 207 77­78 private sector workers, 30 multipillar systems, equity of, reforms in, 20­22, 23­25t2.1, 28 90­91, 91f5.1 impact on women, 91, 92 Musalem, Alberto R., 78 taxonomy of, 22 mutual fund administrators, 253 and stock market liquidity, 81, mutual fund industry, 66b4.1, 85, 82, 82­83 238, 261 subsidies for savings, 247 experience in Chile, 259­60 tax structure, 66b4.1 investment in, 84 INDEX 333 as retirement savings, 246, see also elderly 264n12 operating costs and voluntary savings plans, of banks, 79­80 251­56, 265nn15­17 Chile, 254 ratio to management fees, 8 N reduction of, 233­34, National Housing Fund for 235­36t10.1 Workers Institute, 87n10 Organisation for Economic National Retirement Savings Co-operation and Commission, Mexico, 87n4 Development (OECD), 69 NDC. see notional defined Orszag, Peter, 99 contribution (NDC) systems oversight, 85 net interest margins, 78, 87n11 net investment returns, 131 P net-present-value calculations, 50 Packard, Truman, 54, 101, 105n8, New Zealand, 211, 222 167 Nicaragua and fiscal impact of reforms, and coverage rates, 99 39­40 reforms to social security in, on minimum pension guarantee, 20­21, 26­27t2.2, 37n3 194 noncontributory pension programs, on pension contributions, 89 213­18, 225nn6­7 risk preferences in Chile, 183 notional defined contribution worker's contribution behavior, (NDC) systems, 127, 170, 193n5 128b7.2, 129, 194n14 Palacios, Robert J., 41, 44, 99, 101, 105n6 O Pallares-Miralles, Montserrat, 99, occupational pension plans, 237, 101, 105n6 250, 250­51b10.2, 264nn6­7 participation rates, 5­6, 273­74 OECD. see Organisation for Chile, 15n5 Economic Co-operation and and education of participants, Development (OECD) 156­57, 156f7.10 old age and extent of private provision, losses associated with, 101, 105n7 110­11b6.1 impact of pension reform on, macroeconomic argument for 189­90, 194n14 mandatory savings for, and incidence of benefits, 103, 116­21 104t5.2 mandatory savings and as indicator of coverage, individual welfare, 115­16 99­101, 105nn4­5 pooling risk to insure against mandatory vs. voluntary, 9­10, poverty, 114­15 10t1.1 risks to income security during, overview, 175­77 126­30, 161nn2­4 for poorer workers, 149f7.8 savings as income security, as reflection of demand, 175­77 112­14 as social inclusion or individual traditional strategies for income choice, 166­70, 192­93nn3­4 security, 168­69, 192­93n3 PASIS benefits, 225n7 334 INDEX pay-as-you-go (PAYG) systems, 4 pension income, relationship to Bismarkian PAYG systems, earned income, 94f5.3 225n3 Pension Reform Options and retirees, 30 Simulation Toolkit (PROST), and rising life expectancy, 39­40, 54n1, 225n8 20­21f2.1 pensions and pension funds, 5, 19, pension debt 119 accumulated after 2001, 42, approach to study of, 2­4 43f3.2 Barr's list of myths about, debt-to-GDP ratios, 4 110­11b6.1 financed by government BONOSOL program, transfers, 41, 42t3.1, 43, 210­11b9.2 43f3.2 Brazil, 250, 250­51b10.2 and fiscal sustainability, 44, 48, choice of by younger workers, 48f3B.2b 187 pensiones graciables, 225n6 costs of, 218­23 pension fund administration, coverage in EAP, 6f1.1 centralization of, 238 and the Cuota Social, pension fund administrators 208­9b9.1 (AFPs), 68, 70, 85, 175 default option, 158, 158f7.11 Argentina, 239 deficits, 41, 42t3.1, 43f3.2 Bolivia, 238 equity limits, 157, 157t7.5 Chile, 33 growth in, 60, 62­67 collective action by, 58 impact on bank efficiency and and commission structure, 144 stability, 78­80 competition among, 233 impact on government bonds and corporate governance, 71 market, 75­76 and Cuenta 2 plan, 249, investment in venture capital 249f10.2 and real estate funds, 84 fees and commissions charges, as investors in government debt, 32­33, 233 63, 63f4.1 as financial risk managers, 53 and new entrants into labor providing advice on choices, 158 force, 28­29, 30, 38n9 as purchasers of government noncontributory compared to bonds, 119, 121 contributory, 217, 217f9.4 and rate of return, 137­38b7.4 operations of industry, 67­72, recordkeeping of funds 87nn7­8 collected, 31­33 participation rates as social regulation of, 57­59, 86n1, exclusion or individual 87n2 choice, 166­70, and voluntary savings plans, 192­93nn3­4 248b10.1 pillar zero vs. pillar one, workers use of AFP accounts, 205­06 5­6 populations receiving, 7f1.2 pension fund investments. see private management of, 126, investments and investment 161n2 choices ratio of assets to GDP, 270 INDEX 335 role in development of bank reforms, 20­21, 23­25t2.1, 28, securities markets, 77­78, 38n8, 177, 179b8.2 87nn9­10 comparison of reformed setting minimum benefits at pension systems with minimum wage, 218­19 Chile, 175, 178t8.2 survey of contribution behavior, return rates, 131, 138, 140 170­75, 193nn5­6 securities three-fund systems, 242, 245, housing-related securities, 77 264n11 inflation-indexed securities, top-up, 206 73, 74­75b4.2 see also poverty prevention limits on private sector per capita income, 105, 105n6 securities, 59­60 Persson, Mats, 114 and private sector securities Peru, 6, 58, 70, 91, 146 markets, 63 administrative costs, 234f10.1 shareholders' rights, 65b4.1 bond market, 75, 78 social insurance coverage gap, civil servants pension plan, 29 96b5.1 commission structure, 162n10 Social Risk Management Survey, contribution density, 101, 175, 176­77b8.1 184­85f8.4, 185­86 and stock market liquidity, 82, corporate bond market, 83 87n12 creditors' rights, 65b4.1 switch to PAGY systems, 15n6 distribution choices, 152­53, value of annuitizing pension 163nn22 fund balances, 245 and distribution of returns, 91 venture capital and real estate household surveys, 175 investment, 84 interest rate spreads, 79, 79f4.2 voluntary savings plans, 36t2.4 investments peso devaluation, 134b7.3 choices, 160 pillar zero, vs. pillar one, 205­06 of pension funds, 60, 80 policy risk, 127b7.1, 129, 131, and IPDs, 40f3.1, 41 261­62, 272 management fees, 8 Argentina, 133­35b7.3 mandatory savings plans, 31 and investment in government measures of risk preference, bonds, 132 186­87, 187f8.5, 188f8.6 political risk, 264n10 participation choice, 177 and government credibility, 169 pensions politics, and BONOSOL program, choices, 4, 5 211b9.2 deficits, 41, 42t3.1, 43, 43f3.2 Pollner, John, 233 and guarantees, 33, 271 pooling arrangements pension fund management, 33 forecasting cost of, 218­23, and portfolios, 137­38b7.4, 225n8 138­39 vs. savings, 9­10, 10t1.1, 11 profits for fund managers, 233 pooling risk, 112­14 ratio of variable commission to insure against old-age rates to contributions, 146, poverty, 114­15 147f7.6 and market failures, 113b6.2 336 INDEX pooling risk (continued) relationship of mandatory and minimum required contributions to age and contributions, 190 income, 232 risks that justify pooling, role of government in, 110b6.1, 121­22 199­205, 224­25nn1­3 portfolios Uruguay, 264n2 asset allocation and portfolio price elasticity, 143, 150 limits, 142t7.2 price indexation, 77 investment in corporate bonds, PRIESO survey, 175, 176­77b8.1, 83 179­86 leasing bonds in, 77 private sector bonds and securities, limits on, 53, 55n10, 59, 60 63, 67, 77­78, 80­84, 87n12 maturities of, 76 private sector workers, 30 and minimum pension privatization, 267 guarantees, 59, 87n5 of pensions, 3b1.1, 28 regulation of, 62t4.2 and use of income from for and risk, 137­38b7.4 pensions, 210­11b9.2 see also pensions and pension PROST. see Pension Reform funds Options Simulation Toolkit poverty, among elderly when (PROST) measured by current income, provident funds, 161n2 200­01, 202­03t9.1, 224n2 publicly managed individual poverty prevention, 11, 14, 152, retirement systems, 127, 161n3, 271­72 128b7.2, 129 assessment conclusions, 275­76 public pension systems, changes to, and BONOSOL program, 29 210­11b9.2 public sector, bond market, 75 and contributory minimum public sector debt, market for, pension guarantees, 211­14, 72­75 225nn4­5 and the Cuota Social, R 208­09b9.1 rates of return, 131 focus on, 201, 204 and AFPs, 58 forecasting cost of pooling Chile intercohort differences in, arrangements for, 218­23, 135­36, 139f7.3 225n8 December 1994 to December and funding of benefits, 230 2002, 130, 130t7.1, 131 generosity and cost to taxpayers, by insurance companies, 246 209, 212f9.3 ratio of return to standard institution of programs for, 277, deviation, 135, 136f7.2, 279 137­38b7.4 noncontributory programs, risk-adjusted, 140 213­18, 225nn6­7 value placed on, 191, options for elderly, 206­11 194­95n16 Peru's reformed pension system, volatility of, 136 177, 179b8.2 real estate, investment in, 84, 188, pillar zero vs. pillar one, 205­06 194n13, 230 INDEX 337 record-keeping functions, 239 progress in Latin America, 4­9, reforms 15nn4­6 assessment conclusions, 275­80 and risk management, 239, assessment overview, 268­69 242­47, 264n11­12 on banking sector, 76­80, simulated fiscal impact of, 87nn9­11 40­44, 58nn2­3, 59n4 benefits of, 34, 36­37, 126, success of, 4­5 269­70 taxonomy of, 20­30, 37n7, comparison of system in Chile 37nn2­5, 38n8 and Peru, 175, 178t8.2 voluntary savings plans, 258­61, concerns about, 270­75 265n18 and coverage issues, 96­103, regulations 105nn2­8 asset management practices, to decrease costs and 70­71 commissions, 232­39, insurance companies, 257, 240­41t10.2, 264nn7­10 258f10.6 features of investment strategies, 53, 59­60, 1980s and 1990s, 23­25t2.1 82, 83, 87nn3­6, 131­32, 1990s and 2000s, 26­27t2.2 161­62n6 in financial sector, 64­66b4.1 of mutual funds, 253, 253­54 and fiscal positions of of pension fund administrators, governments, 7­8 57­59, 68, 86n1, 87n2 impact of risks and portfolio floors, 62t4.2 macroeconomic concerns on, retirement 50­54, 55nn6­7, 55nn9­11 charge ratio at, 146­47, impact on economic growth, 46, 148f7.7, 162nn12­15 49­50, 51 choices at distribution stage, impact on other benefits, 37n1 150­53, 154f7.9, impact on private sector 162­63nn19­22 securities, 80­84, 87n12 and management fees, 146, 147, improvement of equity of, 148f7.7, 162nn18 91f5.1 withdrawal of funds prior to, incentive effects, 101, 102t5.1, 259­60 103 retirement systems, investment in and increase in coverage, housing as substitute for, 97­98b5.2 188, 194n13 and individualization of social Riester pensions, 265n14 security, 11, 114 risk-adjusted returns, 140 inequitable effects of social risk and risk management, 12, 126, security and reform progress, 130, 273, 279 89­93 and choices at distribution stage, and longevity risk management 150­51 of, 129, 243­44t10.3, 245 comparison of, 137­38b7.4 multipillar approach to, 1, 4, diversification of, 9 15n1 faced by poorer households, 168 options to reduce operational government vs. households, 169, costs, 235­36t10.1 193n4 338 INDEX risk and risk management second-pillar pensions, 122, 277, (continued) 278­79b11.1 of mandated retirement savings, accounts funded with defined 239, 242­47, 264n11­12 contributions, 21 and measures of preference in costs for poorer workers, Peru, 186­87, 187f8.5, 149f7.8 188f8.6 objective of, 9, 15n8, 37n4 Mexico's pension reform, reform options to reduce 51­52, 52f33.3 operational costs of, to old-age income security, 235­36t10.1 126­30, 161nn2­4 see also consumption restrictions on, 155 smoothing; contribution and retirement income security rates; mandatory savings systems, 126, 127b7.1 plans strategies for, 110b6.1 securities markets, reforms in, and structural pension reform, 65b4.1 50­54, 55nn6­7, 55nn9­11 self-employed workers, 193n6 see also pooling risk and contributions, 105n3 risk aversion, 129 and mandatory pension plans, and choice of pension system, 237­38 191 and participation choices, 175 measure of, 181­83, 184, and risk preferences, 181­83, 194n12 187f8.5, 188f8.6, 194n12 self-employed in Peru, 188f8.6 self-insurance, 113b6.2 risk pooling self-protection, 113b6.2 as mechanism to mitigate set-up costs, 143 poverty in old age, 201 shareholders, 71 and minimum pension and AFPs, 58 guarantees, 212­13, 224, of companies traded in stock 225n4 market, 119 public options for, 209 rights of, 65b4.1, 81 risk ratings, 50, 51, 59 Shinkai, Naoko, 170, 193n5 Rofman, Rafael, 103, 105n7, 150 simulations of fiscal impact of reform, S 40­44, 58nn2­3, 59n4 Samaritan's dilemma, 95b5.1 interpretations of results, 44­46, Santiago, Chile, 176­77b8.1 47b3.2 Santiago Stock Exchange, 64b4.1, social exclusion, and participation 70, 80 rates, 166­70, 192­93nn3­4 savings social insurance system, coverage costs to affiliates of, 143­50, problems, 95­96b5.1, 96 162nn9­18 Social Risk Management Survey, as mainstay for earnings 175, 176­77b8.1 replacement, 11­12 social security reform. see reforms vs. pooling, 9­10, 10t1.1, 11 Specialized Retirement Fund see also mandatory savings plans Investment Societies, 76 Schmidt-Hebbel, K., 49, 53, 54, 119 Srinivas, P.S., 140 INDEX 339 Standard & Poor's index, 140 three-fund systems, 242, 245, Stiglitz, Joseph E., 99 264n11 stock market thrift savings plans (TSPs), 242 capitalization of, 80 time preference rate, 168, 183 liquidity of and pension fund transaction costs, in securities investment, 80­83, 87n12 markets, 80 subordinated bonds, 83 transition costs, 132, 262 subsidies Bolivia, 44, 45­46b3.1 flat contribution, 208b9.1 financing of, 53, 55n9 and funded systems, 31 transition debt, 69­70, 87n7 for home purchase, 87n9 transparency, 71, 85 and savings plans, 247­48, Bolivia, 87n3 265n14 commission structure, 260 supply-side measures, to reduce and company pension plans, costs, 233­34 234 survivor insurance, 31, 38n11, of costs and benefits of financial 38n13, 105n4 products, 247 swap of bonds, 134b7.3 of fees, 239 Sweden, 238, 239, 264n10 and mutual funds, 253 treasury bonds, 73 T Tressel, Thierry, 78 targeted benefits, 219, 222 Truglia, V., 50 targeted pensions, 206­07, 209­11 TSPs. see thrift savings plans (TSPs) taxation turnover ratios, 80­81, 81f4.3 after-tax returns, 66b4.1 Cuenta 2 plan, 259 U expenditure tax, 247, unidad de fomento (UF), 71­75, 77 264­65n13 unisex mortality tables, 93 and mandatory savings, United Kingdom, 239 31­32 United States, 242, 264n7 and mutual fund savings, 253 universal benefits, 206­07, 209­11, and perceived benefits and 219, 222 contribution link, 98­99 Uruguay, 4, 72, 99 reforms in, 66b4.1 commission structure, 144 tax liability and retirement corporate bond market, 84 savings, 247 demographics, 19 and voluntary savings plans, distribution choices, 152­53 248, 249­50 investments tax-preferred voluntary retirement choices, 160 savings, 34 investment risk, 155 Telmex, 82­83 in mutual funds, 84 third-pillar pensions, 122 of pension funds, 59­60, 60 and defining incentives, 21 in private sector, 67 objective of, 9, 37n4 and IPDs, 40f3.1, 41 see also consumption mandatory savings plans, 31 smoothing; incentives; noncontributory poverty voluntary savings plans prevention pensions, 216t9.3 340 INDEX Uruguay (continued) and costs of pension fund pension deficits, 41, 42t3.1, 43, administration, 148­49, 43f3.2 162n17 pension fund managers, 33 and mandatory savings plans, and pension guarantees, 33 31 portfolios, 62t4.2, 137­38b7.4 maximum taxable, 232 and poverty prevention, 264n2 minimum wages as share of reforms to social security in, average wages, 218­19, 20­22, 23­25t2.1, 28, 29, 218t9.4 37n7 relationship to benefits, 164n22 transition debt, 69, 70 setting minimum pension voluntary savings plans, 36t2.4 benefits at minimum wage, Uthoff, Andras, 66b4.1 218­19 Walker, E., 70, 71, 80, 254 V Washington Consensus model, 267 VAC. see valor adquisitivo wealth, accumulated, 201, 204f9.2 constante (VAC) Whitehouse, Edward, 143, 201 Valdés-Prieto, Salvador, 103, 141, Willmore, Larry, 209, 211 163n21, 194­95n16 withdrawals, restrictions on, age-related contributions, 264n3 265n16 and mandated savings, 116, 237 women valor adquisitivo constante (VAC), contribution histories, 193n8 77­78 losses in rates of return, 92­95 valuation mechanisms, 59, 87n4 Wong, R., 91 Venezuela, life insurance workers policyholders, 257 commissions on pension funds, venture capital, 84 144, 145t7.3 voluntary savings plans, 8, 13, 125, contribution behavior, 170­75, 161n1, 228, 237 193nn5­6, 211­12 Brazil, 250­51b10.2 contribution histories, 183­85 improving of, 247­51, 258­61, costs of participation in second 264­65nn13­14, 265n18 pillar, 149f7.8 and life insurance industry, disposable income of, 253 256­57, 258f10.6 and distribution choices, 150­53 and mutual funds, 251­56, and foreign investors, 82, 87n12 265nn15­17 impact of mandatory savings on structure and implementation of, life cycle needs, 229­32, 34, 35­36t2.4 263­64nn1­4 taxation of, 248 improvident, 230, 232 inequitable impact of pension W benefits on, 93­96 wage employees. see workers and management fees, 146, 147, wage-indexed government bonds, 148f7.7 128b7.2 participation rates as social wages, 162n13 exclusion or individual choice, and cost of providing pension 166­70, 192­93nn3­4 equal to minimum wage, and policy risk protection, 220­21t9.5 133­35b7.3 INDEX 341 poverty prevention options, World Bank 206­11 background papers on pensions, private sector, 30 2­3, 25n2 relationship of wages to coverage rates, 97­98b5.2 mandatory savings, 31 PROST, 39­40, 54n1 and risk management, 154­60, see also Averting the Old Age 245 Crisis and risk preferences, 181­83, 187f8.5, 194n12 Y share who contribute to pension Yermo, Juan, 84, 140 systems, 219 yield, reduction in, 143­44 use of AFP accounts, 5­6 yield curve, 83 see also labor force; self- employed workers Z working-age population, percent of, Zviniene, Asta, 39­40, 41, 44, 21f2.1 54 LATIN AMERICAN DEVELOPMENT FORUM E mpirical analysis of two decades of pioneering pension and social security reform in Latin America and the Caribbean shows that much has been achieved, but that critical challenges remain. In tackling this unfinished agenda, a great deal can be learned from the reform experience of countries in the region. Keeping the Promise, produced by the chief economist's office in the Latin America and Caribbean Region at the World Bank, evaluates policy reforms in 12 countries, points to successes and shortcomings, and proposes priorities and options for future reform. The authors argue that preventing poverty in old age should be the primary objective of public pension schemes, and that this goal can best be achieved by relentless efforts to extend coverage to broader segments of society, and by encouraging (though not necessarily mandating) personal saving. "Although it maintains the fundamental principles of Averting the Old Age Crisis, published by the World Bank in 1994, Keeping the Promise takes a fresh look at the outcome of structural pension reforms implemented in 12 Latin American countries during the 1990s, and proposes new, sometimes bold policies. This is an important and controversial book that must be read by experts on pensions, social security, and Latin America." --Carmelo Mesa-Lago, Distinguished Professor Emeritus of Economics and Latin American Studies, University of Pittsburgh "Keeping the Promise is not a typical World Bank book on pensions. From a welcome emphasis on the individual and on the primary importance of poverty relief, this volume builds on careful analytics and new empirical evidence to document the successes and failures of pension systems in Latin America. The conclusions are well-argued, clear, and--given their origins--striking." --Nicholas Barr, Professor of Public Economics, London School of Economics "Keeping the Promise provides a timely assessment of two decades of pension reform experience--with a wealth of new data, and empirical evaluation of reformed social security systems. Many economists and policymakers will not be persuaded by some of the main conclusions and recommendations--such as the supposed failure to increase coverage, and the call for strengthening a pay-as-you-go defined-benefit scheme for poverty prevention-- but they will welcome the book's critical appraisal. This is required reading for pension specialists and policymakers in Latin America and beyond." --Klaus Schmidt-Hebbel, Chief of Economic Research, Central Bank of Chile "A heavyweight analysis of the Latin American pension revolution, which raises important questions about the optimal scale of compulsory saving when redesigning pension systems." --Paul Wallace, The Economist www.sup.org STANFORD UNIVERSITY PRESS THE WORLD BANK ISBN 0-8047-5181-1