Colombia's pension reform : fiscal and macroeconomic effects

WDPF'31- A&. I(6V 314 1 World Bank Discussion Papers Colombia's Pension Reform Fiscal and Macroeconomic Effects Klaus Schmidt-Hebbel  Recent World Bank Discussion Papers No. 247 Improving the Transfer and Use of Agricultural Information:A Guide to Information Trechnology. Willem Zijp No. 248 Outreach and Sustainability of Six Rural Finance Institutions in Sub-Saharan Africa. Marc Gurgand, Glenn Pederson, and Jacob Yaron No. 249 Population and Income Cliatige: Recent Evidence. Allen C. Kelley and Robert M. Schmidt No. 250 Submission and Evaluation of Proposalsfor Private Power Generation Projeas in Developing' Countries. Edited by Peter A. Cordukes No. 251 Supply and Demandfor Finance of Small Enterprises in Ghana. Ernest Aryeetey, Amoah Baah-Nuakoh,Tamara Duggleby, Hemamala Hettige, and William E Steel No. 252 Projectizing the Governance Approach to Civil Service Reform:An Environment Assessment for Preparing a Sectoral Adjustment Loan in the Cambia. Rogerio F Pinto with assiscance from Angelous J. Mrope No. 253 Small Firms Informally Financed: Studiesfrom Bangladesh. Edited by Reazul Islam,J. D.Von Pischke, and J. M. de Waard No. 254 Indicalorsfor Monitoring Poverty Reduction. Soniya Carvalho and Howard White No. 255 Violence Against Women: Tlhe Hidden Hlealth Burden. Lori L. Heise with Jacquehne Pitanguy and Adrienne Germain No. 256 Women's Health and Nutrition: Making a Difference. Anne Tinker, Patricia Daly, Cynthia Green, Helen Saxenian, Rania Lakshndinarayanan, and Kirrin Gill No. 257 Itnproving the Quality of Primary Education in Latin America and the Caribbean: Toward the 21st Century. Lawrence Wolff, Ernesto Schiefelbein, and Jorge Valenzuela No. 258 How Fast is Fertility Declining in Botswana and Zimbabwve? DuncanThomas and Ityai Muvandi No. 259 Policies Affecting Fertility and Contraceptive Use: An Assessment of Twelve Sub-Saharan Countries. Susan Scribner No. 260 Financial Systems in Sub-Saharan Africa: A Cotnparative Study. Paul A. Popiel No. 261 Poverty Alleviation and Social Investment Funds: The Latin American Experience. Philip J. Glaessner, Kye Woo Lee, Anna Maria Sant'Anna, and Jean-Jacques de St. Antoine No. 262 Public Policyfor the Promotion of Family Farns in Italy: The Experience of the Fund for the Fornatiotn of Peasant Property. Eric B. Shearer and Giuseppe Barbero No. 263 Self-Employmentfor the Unemployed: Experience in OECD and Transitional Economies. Sandra Wilson and ArvilV Adams No. 264 Schooling and Cognitive Achievetnents of Chiildren in Morocco: Can the Government Improve Outcomes? Shahidur R. Khandker,Victor Lavy, and Deon Filner No. 265 World Bank-Financed Projects with Community Participation: Procurement and Disbursement Issues. Gita Gopal and Alexandre Marc No. 266 Seed Syslems in Sub-Saharan Africa: Issues and Options. VVenkatesan No. 267 Trade Policy Reform in Developing Countries Since 1985: Review of th e Evidence. Judith M. Dean, Seema Desai, and James Riedel No. 268 Farm Restructuring and Land Tenure in Reforming Socialist Economies: Comparative Analysis of Eastern and Central Europe. Euroconsult and Centre for World Food Studies No. 269 The Evolution of the World Bank's Railway Lending. Alice Galenson and Louis S. Thompson No. 270 Land Reform and Farm Restructuring in Ukraine. Zvi Lerman, Karen Brooks, and Csaba Csaki No. 271 Small Ettterprises Adjusting to Liberalization itt Five African Countries. Ron Parker, Randall Riopelle, and William E Steel No. 272 Adolescent Health: Reassessing the Passage to Adulthood. Judith Senderowitz No. 273 Measurement of Wefare Changes Caused by Large Price Shifts:An Issue in the Power Sector. Robert Bacon No.274 Social Action Programs and Social Funds: Review of Design and Implementation in Sub-Sah aran Africa. Alexandre Marc, Carol Graham, Mark Schacter, and Mary Schmidt No. 275 Investing in Young Children. Mary EmningYoung No. 276 Managing Primary Health Care: Implications of the Health Transition. Richard Heaver No. 277 Energy Demand in Five Major Asian Developing Countries: Structure and Prospects. Masayasu Ishiguro and Takanias Akiyama (Continued on the inside back cover)  3 1 4 S World Bank Discussion Papers Colombia's Pension Reform Fiscal and Macroeconomic Effects Klaus Schmidt-Hebbel The World Bank Washington, D.C.  Copyright C 1995 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W, Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing November 1995 Discussion Papers present results of country analysis or research that are circulated to encourage discussion and com- ment within the development community. To present these results with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use.The boundaries, colors, denomina- tions, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above.The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, with- out asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list (with full ordering information) and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W,Washington, D.C. 20433, U.S.A., or from Publications,The World Bank, 66, avenue d'1ena, 75116 Paris, France. ISSN: 0259-21OX Klaus Schmidt-Hebbel is principal economist in the Macroeconomics and Growth Division of the World Bank's Policy Research Department. Library of Congress Cataloging-in-Publication Data Schmidt-Hebbel, Klaus. Colombia's pension reform: fiscal and macroeconomic effects / Klaus Schmidt-Hebbel. p. cm.-(World Bank discussion papers, ISSN 0259-210X; 314) Includes bibliographical references. ISBN 0-8213-3506-5 1. Pensions-Colombia. I.Title. 11. Series. HD7157.S36 1995 331.25'2'09861-dc2O 95-45621 CIP  iii CONTENTS FOREWORD ..... ........................................... vii ABSTRACT. viii ACKNOWLEDGEMENTS .xi 1. INTRODUCTION I 2. MAIN FEATURES OF COLOMBIA'S PENSION REFORM I The Old Pension System. 2 The New Pension System. 5 The Reform Transition. 7 3. FISCAL IMPLICATIONS OF PENSION REFORM. 9 Fiscal Effects of Colombia's Pension Reform. 9 Pension Reform Deficits and Implicit Pension Liabilities in Other Countries ............................................. 14 4. MACROECONOMIC AND WELFARE IMPLICATIONS OF PENSION REFORM IN OTHER COUNTRIES .15 Pension Reform, Saving, and Growth .15 Pension Reform, Employment, and Growth .17 Capital Market Development and Growth .18 Chile before and after the Pension Reform .19 5. FISCAL AND MACROECONOMIC IMPLICATIONS OF COLOMBIA'S PENSION REFORM .20 Financing Options ......................................... 20 Long-Run Output and Welfare Effects of Colombia's Pension Reform if Financed by Fiscal Contraction .23 Long-Run Effects of Colombia's Pension Reform on Growth and Equity .24 6. CONCLUSIONS AND FUTURE POLICY CHALLENGES .27 REFERENCES .33 ANNEXES .37 l: Model Assumptions for Simulating the Fiscal Effects of Colombia's Pension Reform 2: Macroeconomic and Welfare Effects of Pension Reform: How do they Arise? 3: An Overlapping-Generations Model for Simulating Steady-State Macroeconomic and Welfare Effects of a Pension Reform  iv TABLES 2.1. COLOMBIA: PENSION SYSTEM INSTITUTIONS BEFORE AND AFTER THE REFORM 2.2. COLOMBIA: POPULATION, EMPLOYMENT AND PENSION SYSTEM COVERAGE, 1990- 1993 2.3. COLOMBIA, LATIN AMERICA AND THE OECD: PUBLIC PENSION SCHEME COVERAGE 2.4. COLOMBIA AND THE WORLD: PUBLIC PENSION COVERAGE AND SPENDING 2.5. COLOMBIA: CONTRIBUTION RATES AND BENEFITS OF OLD PENSION SYSTEM 2.6. COLOMBIA: PENSION SYSTEM CONTRIBUTIONS, BENEFITS, SURPLUSES, AND RESERVES, 1992-1994 2.7. COLOMBIA: CONTRIBUTION RATES AND BENEFITS OF NEW PENSION SYSTEM 2.8. COLOMBIA AND THE WORLD: FEATURES OF MANDATORY PENSION SYSTEMS 3.1. FISCAL EFFECTS OF COLOMBIA'S PENSION REFORM: MAIN FEATURES OF EIGHT ACTUARIAL SIMULATIONS 3.2. COLOMBIA: PUBLIC PENSION SYSTEM (SS AND SPP) DEFICITS IN 1994 AND 2025, VARIOUS SIMULATIONS 3.3. COLOMBIA: PUBLIC PENSION SYSTEM DEFICITS BY INSTITUTIONS AND CATEGORIES, 1994-2025, ACTUAL REFORM BASE CASE SIMULATION 3.4. COLOMBIA: CUMULATIVE PUBLIC PENSION SYSTEM (ISS AND SPP) EXPLICIT DEBT, BASE-CASE REFORM SIMULATION 3.5. COLOMBIA: CUMULATIVE PUBLIC PENSION SYSTEM (SS AND SPP) EXPLICIT AND IMPLICIT PAYG DEBT, BASE-CASE REFORM SIMULATION 3.6. COLOMBIA: CUMULATIVE PUBLIC PENSION SYSTEM (ISS AND SPP) DEBT AND PRIVATE PENSION SYSTEM (AFPs) ASSETS AT YEAR 2025, VARIOUS SIMULATIONS 3.7. CHILE: PENSION REFORM DEFICITS, 1981-2030 3.8. COLOMBIA, CHILE, AND SEVEN OECD COUNTRIES: NET LIABILITIES OF PUBLIC PENSION SYSTEMS 4.1. LONG-RUN OUTPUT AND WELFARE EFFECTS OF PENSION SYSTEMS AND REFORMS 4.2. LONG-TERM GROWTH EFFECTS OF PENSION SYSTEMS IN A REPRESENTATIVE ECONOMY 4.3. SELECTED COUNTRIES: REAL INVESTMENT RETURNS FOR PUBLICLY AND PRIVATELY MANAGED PENSION FUNDS, 1980s 4.4. CHILE: PENSION REFORM AND OVERALL PERFORMANCE, 1979-1992 5.1. COLOMBIA: PUBLIC SECTOR DEFICIT AND DEBT, 1990-1993 5.2. COLOMBIA: LONG-RUN MACROECONOMIC AND WELFARE EFFECTS OF A FISCAL CONTRACTION-FINANCED PENSION REFORM, THREE SIMULATION RESULTS 5.3. COLOMBIA AND LATIN AMERICA: EMPLOYMENT STRUCTURE, 1980-1992 5.4. COLOMBIA: PAYROLL CONTRIBUTION RATES 5.5. COLOMBIA: DISTRIBUTIONAL EFFECTS OF OLD PENSION SYSTEM FIGURES 3.1. COLOMBIA: CONTRIBUTORS TO THE MANDATORY PUBLIC AND PRIVATE PENSION SYSTEM, ALTERNATIVE SIMULATIONS, 1994-2025 3.2. COLOMBIA: MANDATORY PUBLIC PENSION SYSTEM ISS PLUS SPP) DEFICITS FOR SELECTED SIMULATIONS, 1994-2025  v 5.1. LIFE-CYCLE INCOME AND CONSUMPTION: PAYG AND FF FOR COLOMBIA. SIMULATION I 5.2. LIFE-CYCLE INCOME AND CONSUMPTION: PAYG AND FF FOR COLOMBIA. SIMULATION 3 5.3. LIFE-CYCLE INCOME AND CONSUMPTION: PAYG AND FF FOR COLOMBIA. SIMULATION 3   vii FOREWORD Colombia's 1994 reform of its pension system -- the fifth in Latin America and the Caribbean -- starts a gradual replacement of its existing state-managed and pay-as-you-go scheme by a privately- managed and fully-funded system. As elsewhere in developing and industrialized economies, the objective of such comprehensive overhauls of old-age security arrangements is to correct the blatant inefficiencies and inequities that characterize conventional systems. By adopting a multi-pillar pension system aimed at protecting the old and promoting growth, Colombia puts into practice the recommendations set forth in the World Bank's 1994 Policy Research Report on "Averting the Old Age Crisis. " This World Bank Discussion Paper assesses the likely fiscal and macroeconomic implications of Colombia's pension reform, drawing from Colombia's structural features, the international experience, and recent research on pension systems carried out at the World Bank and elsewhere. The study provides estimations for the temporary (but protracted) public deficits caused by the reform, analyzes the government's options for financing the deficits, and performs simulations of the likely long-term growth and welfare effects when the reform is financed by a fiscal contraction. The paper argues that in order to reap the reform's potential efficiency and equity gains, a number of complementary reforms should be carried out to improve the working of labor markets, capital markets, and the role of the government in its complementary provision of pension services. Lyn Squire, Director Policy Research Department  viii ABSTRACT In 1994 Colombia started replacing its state-run and pay-as-you-go (PAYG) pension system by a privately-run and fully-funded scheme. This study analyzes prospective fiscal and macroeconomic implications of this reform. It compares the features of Colombia's old and new pension system, puts them into the broader international context, and looks at the reform transition. Numerical simulations for the government's reform transition reveal implicit PAYG debt levels and corresponding reform transition deficits that are high relative to other countries, considering that Colombia's old pension system was characterized by low coverage, low system maturity, and a young population. Simulation results show that output could increase by 10% due to higher future saving caused by financing the pension deficit by a fiscal contraction -- but this would occur only in the very long term. Sooner and possibly larger gains could be reaped from higher employment and production in formal sectors, and the development of capital markets spurred by the reform. In addition, Colombia's new pension system -- that includes a redistributive pillar targeted at the poor -- is potentially more equitable than the old scheme. To reap these efficiency and equity benefits, however, the Colombian government would have to adopt complementary reforms. They include giving the private fully-funded pension pillar a commanding role, supporting the development of capital markets, and bolstering formal-sector employment by the reduction of deadweight tax burden of non-pension programs that are currently financed by payroll taxes on labor.  ix ACKNOWLEDGEMENTS I thank Ulpiano Ayala and Hermann von Gersdorff for the initial motivation to write this paper, subsequent discussions, and valuable comments to a first draft. I am also very indebted to Loredana Helmsdorff and Rodrigo Cifuentes, who kindly performed the simulations for sections 3 and 5, respectively. I benefitted from useful discussions with Alberto Carrasquilla, Eduardo Lora, Beatriz Marulanda, Dolly Ovalle, Mauricio Perfetti, Fanny Santamarfa, Beatriz Helena Torres, Jose Darfo Uribe, and M6nica Uribe in Bogota, and valuable comments received from Suman Bery, Juan Luis Londofio, and other participants at a World Bank seminar. I thank Ariel Fiszbein for kindly providing data.   1. INTRODUCTION Colombia's congress enacted a deep reform of the pension system in December 1993. Contributions to new private pension funds started in April 1994. The reform comprises a gradual and partial substitution of Colombia's current state-run pay-as-you-go (PAYG) system by a privately-run fuliy- funded (FF) that could be the dominating pension pillar in the near future. The required institutional, fiscal and financial changes are large, and so are the potential efficiency and equity gains if the reform succeeds. This study analyzes the prospective fiscal and macroeconomic effects of the pension reform.' Section 2 focuses on the features of Colombia's old and new pension system, puts them into the broader international context, and looks at the reform transition. The next section reports numerical simulation results for the government's reform transition deficits and their capitalized value. Sensitivity analyses are performed for alternative economic and systemic assumptions. Colombia's projected figures are compared to Chile's experience after its radical 1981 pension reform and to implicit PAYG pension liabilities in OECD countries. The positive effects of pension reform on growth and welfare occur through three channels: more saving, higher formal-sector employment and production, and development of capital markets. These three links are analyzed in section 4, drawing from analytical results and simulation exercises for other countries. A brief look at Chile's post-reform experience illustrates the empirical dimension of the potential macroeconomic effects of pension reform. Section 5 focuses on Colombia again by discussing its options for financing the pension reform deficit, reporting simulation results for the long-run output and welfare effects of a reform financed by fiscal contraction, and making a qualitative assessment of the overall consequences of pension reform. A concluding section summarizes the main findings and identifies some challenges faced by the Colombian government to ensure that the potential efficiency and equity gains of pension reform are realized. 2. MAIN FEATURES OF COLOMBIA'S PENSION REFORM Colombia social insurance system has been beset by all the ills of state-managed pay-as-you-go (PAYG) schemes in other developing and OECD countries. For the last 50 years, Colombia's pension system has been characterized by uneven contributions and benefits across different pension sub-systems, large transfers to the first generations of pensioners, weak relations between contributions and expected ' For the most complete and recent evaluation of pension systems across the world see World Bank (1994b). For a survey of issues for the design and reform of pension systems see Arrau and Schmidt- Hebbel (1994). Comparative studies of Latin American pension systems and reforms include Mesa-Lago (1978, 1993), Mc Greevey (1990), and Fundaci6n Friedrich Ebert de Colombia (1992). The initial reform proposal by the government of Colombia (that includes actuarial financial projections) is in Ministerio de Trabajo y Seguridad Social de Colombia (1993a). Among other studies on Colombia's pension reform in anticipation of the December 1993 law are C. L6pez (1992), H. L6pez (1992), Contralorfa General de la Republica (1992), Ocampo (1992), Ramfrez (1992), Zuleta (1992), and Lora, Zuleta and Helmsdorff (1993). A comprehensive account of the Colombian pension reform process, its objectives and outcomes is found in Ayala (1995). Lora and Helmsdorff (1995) present an evaluation of the financial prospects of ISS and the private pension funds based on the I)ecember 1993 law.  2 pension benefits, low coverage of Colombia's labor force, inefficient public management, very poor funding of pension liabilities, and rising fiscal transfers to cover cash deficits. However, as a result of its early stage of demographic and systemic maturation, as well as low coverage, most of these deficiencies have been quantitatively smaller in Colombia than in other countries. This makes Colombia's 1994 pension reform -- the fifth after those in Chile, Argentina, Mexico, and Peru -- even more remarkable. It shows the farsightedness of the government in anticipating the long-term unsustainability of the old pension scheme due to its serious fiscal, labor-market, and macroeconomic consequences that would have been suffered in the future. This section focuses on the features of the old and the new pension system, puts them into the international context, and looks at the reform transition. The Old Pension System High pension system fragmentation... Colombia started a social security system for central-government workers with the establishment of the Caja Nacional de Previsi6n (Cajanal) in 1946, that centralized existing occupational retirement plans and extended coverage to all government workers. Other national pension funds and "cajas" were created subsequently for central-government employees working in specialized agencies or professions in response to their successful lobbying efforts for separate, i.e., more beneficial pension regimes. Additional separate cajas and pension funds were started by regional governments for departmental and municipal workers, with legal statutes ranging from established pension funds to simple pension accounts on the books of local governments. A final segment within the public-sector pension system is comprised by state-owned enterprise (SOE) funds established by public-sector firms for their own workers with a self-determined structure of contributions and benefits. Mandatory old-age insurance for private-sector workers -- by far the largest group of contributors -- was formalized only in 1967 by the previously existing Instituto de Seguros Sociales (ISS). Hence at the time of the reform in late 1993 Colombia's fragmented pension system is comprised by six broad categories of pension plans and institutions (see Table 2.1). The first three categories are pension institutions for government workers (Cajanal, 55 other national pension funds and cajas, and 991 regional pension funds and cajas);2 a fourth category comprised by decentralized special-government and SOE pension funds; lSS, the mandatory pension system for private-sector workers; and a final group of pension plans that are sponsored by financial-sector firms (insurance companies) for voluntary old-age saving by private and public-sector workers and by private firms for their own employees (occupational or employer-based schemes). 2 The precise number of national and regional pension funds changes frequently. The 55 national pension funds in 1993 are individually identified in Sub-Direcci6n de Desarrollo Social (1994) and the 991 regional pension funds ate those identified by the 1989-90 Social Security Census summarized in Ministerio de Trabajo y Seguridad Social (1993c).  3 ... with low coverage, little contributions but generous benefits... Colombia's large diversity of pension institutions stands in marked contrast to the low coverage of its labor force. The three government-sector systems (called Sistema PNiblico Pensional or SPP3) comprise 800,000 contributors and 532,000 pensioners, and the dominating ISS comprises 3.425 million contributors and 265,000 pensioners (Tables 2.1 and 2.2). The total number of contributors in 1993 (4.225 m.) reached barely 29.6% of Colombia's labor force or 31.8% of total employment.4 Low coverage reflects large evasion -- not only of pension contributions but of payroll taxes at large as well as profit, income, and indirect taxes. Colombia's 29.6% pension-system coverage of labor force is indeed low by international standards. In Latin America and the Caribbean coverage is 38.3 % (Table 2.3) while in middle-income countries it averages 34.8% (Table 2.4).5 Inferring from cross-country regressions relating pension system coverage to per-capita income, Colombia's labor force coverage is significantly lower than the 43 % figure which would correspond to the country's income level. (World Bank, 1994b, Figure 1.7). As opposed to most other pension schemes in the world, contributions paid by government employees to SPP were generally zero under the old pension system. Private-sector workers, however, were required to contribute a (low) 6.5% of their wages to ISS (Table 2.5). Benefits provided by the old pension system also varied widely between different regimes (and were also more favorable to public-sector workers) but were on average generous. As in many other countries, the first generations after pension systems are started benefit the most. However, their gains are large are not only relative to their low costs but also in comparison to other countries (Table 2.8). Standard retirement ages under Colombia's old pension scheme were 50-55 (for women-men) for public- sector and 55-60 for private-sector workers (Table 2.5). Standard replacement rates were defined as pension payments relative to average nominal wages 2 years before retirement and varied between 75% for public-sector and 45-90% for private-sector workers, subject to low numbers of minimum years of contributions. However, wide-spread special pension regimes provided even better conditions than those of the standard benefits package, such as frequently granted early retirement and replacement rates well in excess of 100%. In Cajanal alone 17 different pension regimes coexisted in 1993, while many national, regional and SOE-sponsored cajas for public-sector workers offered their own old-age insurance packages. All six components of Colombia's old pension system -- excluding a few retirement plans offered by private-sector retirement plans for voluntary savers -- were defined-benefit programs, with 3 Throughout this paper SPP is defined as excluding the three special public-sector pension funds for oil workers, teachers, and armed-forces personnel exempted from the pension reform. 4 This figure of 4.225 million contributors in 1993 excludes the roughly 300,000 contributors to the three special public-sector funds exempted from the pension reform and an estimated 100,000 contributors to private voluntary and/or private occupational plans. 5 The latter figure --as well as subsequent figures for middle-income countries quoted below -- is a weighted average of the group averages for lower-middle and higher-middle income countries reported in Table 2.4.  4 plan sponsors bearing in theory the risk of changes in investment values (in those funds with at least partial funding) or in contributions (in the more typical cases of unfunded or PAYG plans). However, real pension benefits were not effectively protected under the old pension scheme. Frequent legislative and administrative changes in pension plans and variable inflation rates -- that affect the real value of both the pension base at retirement and pension benefit payments during retirement -- changed significantly constant-price pension benefits and hence weakened further the links between individual contributions and expected benefits.6 run as a PA YG system with small financial deficits expected to grow much in the future ... Colombia's old mandatory pension system sponsored by the public sector was operated as a financially unbalanced PAYG system requiring net government transfers to cover its revenue shortfalls. Pension payments by SPP reached 1.42% of GDP in 1993, almost entirely financed by budgetary public- sector transfers to Cajanal, the other national, and the regional pension funds and cajas (Table 2.6). Pension benefits paid by ISS reached 0.84% of GDP in 1993, slightly below 1993 contributions of 0.97% of GDP, implying a small surplus. Colombia's mandatory public pension spending at 2.3% of GDP (or 11 .1 % of government expenditure) is well below the average 4.5% of GDP (or 16.2% of government expenditure) in middle- income countries (Table 2.4). Low coverage is only one of the three reasons for Colombia being an outlier. The other two are that Colombia is still at relatively early stages of population maturation and pension system maturation. Hence its old-age dependency ratio -- the relation between the number of pensioners and contributors -- attains only 12.6%, also much below the average 21.0% observed in Latin America and the Caribbean (Tables 2.2 and 2.3). Coverage of old people -- the ratio between pensioners and persons over 60 -- is only 24.6% in Colombia, contrasting to the average 30.8% in Latin America and the Caribbean. Colombia's ongoing demographic transition and the maturation of its pension system imply that the low 1993 pension deficit is not at all representative of future pension payments shortfalls that would have accrued under the old pension scheme. The simulations in section 3 will show that under the old structure of contributions and benefits a massive rise in the pension deficit would have occurred in the next decades, forcing either increasing government transfers or lower net pension benefits to future generations, or a combination of both. ... and a large shortfall of pension system reserves from implicit pension liabilities. Run as a de-facto PAYG scheme, SPP current pension reserves are close to zero (Cajanal's reserves are actually zero) and ISS pension reserves reach only 1.6% of GDP (Table 2.6), far below its implicit pension liabilities estimated below. In the absence of pension reform, the gap between implicit PAYG debts and explicit reserves would only increase in the next decades until maturity of the population age structure and the pension system were attained. 6 Full indexation of pension benefits to inflation was adopted in 1988 for public-sector pensioners.  5 However, a deteriorating fiscal position is only the most visible cost of Colombia's old pension system. Its design flaws -- common to most other pension regimes in the world -- impose rising efficiency and equity costs. The large transfers to the first generations of pensioners reduces long-term saving and growth. The weak links between payroll contributions and PAYG benefits embedded in a PAYG system with implicit redistribution and large benefit uncertainty contributes to employment and production informality, tax evasion and moral-hazard behavior by contributors, and again, lower growth (see Ayala 1995). Both the lack of funding and public-sector management precludes the positive contribution to capital-market development and growth made by privately-managed FF schemes that invest pension savings in long-term financial instruments. State-managed pension funds with non-transparent cost and benefit structures and strongly subject to political and special-interest pressures show high administrative costs relative to their low-quality services reflected in low steady-state benefits to pensioners. Income redistribution in such systems is often perverse and regressive, seldom reaching the old-age poor. The New Pension System Realization of these costs and inequities of the old system led to the Colombian pension reform. After three years of government preparation and congressional discussion of social security reform, the Colombian Congress passed Law 100 in December 1993. The law provides the general frame for both pension and health system reforms with considerable administrative latitude given explicitly by Congress to the government to issue decrees on essential features of both reforms until July 1994. The government issued a large number of decrees that regulate fundamental aspects of pension reform legislation and implementation. The pension reform introduces fundamental changes to the old mandatory pension scheme.7 Its main elements are: * Significant increase in pension contribution rates paid by all active workers and significant reduction of pension benefits accrued to all younger workers and to those older workers that change affiliation to private pension funds * Substitution of a mixed PAYG - fully-funded (FF) scheme for the existing PAYG system * Introduction of two alternative mandatory pension sub-systems, chosen on an either/or base by contributors: a state-managed defined-benefit partially-funded PAYG scheme run by ISS and a privately-managed defined-contribution FF system comprised by private providers of contribution collection and investment services (pension-fund management firms called Administradoras de Fondos Pensionales or AFPs) and by private providers of pension sen ices and annuity payments (insurance companies) * Transparent income redistribution by starting explicit programs in support of pension benefits for the poor. Voluntary complementary pension plans offered by the financial sector or private employers are not affected by the reform.  6 * Government recognition of complete pension rights accrued to current (1994) pensioners and older contributors and of partial pension rights accrued to younger contributors, implying public- sector repayment of the corresponding implicit PAYG debt through means other than future payroll contributions * Regulation and supervision of investment funds managed by ISS and AFPs * Regulation of reform transition and reform exemptions The institutional set-up for the reform transition lasting through, say, 2060 is complex (Table 2.1). All public-sector workers changing jobs and new labor market entrants are forced to affiliate either with ISS or AFPs. Insolvent national, regional, special government, and SOE-sponsored funds will be closed and their workers can choose between ISS and AFPs. Currently insolvent funds -- the very large majority of all public funds -- are allowed to constitute reserves to reach solvency and hence remain in business. Solvent funds retain their affiliates until the last survivors pass away. Three public-sector pension funds were exempted from this general rule and hence will survive in the long term as unfunded (or partially-funded) PAYG and defined-benefit (DB) plans: those for teachers, oil workers, and the armed forces. Transitional transfer programs financed by the central government (the Fondo Nacional de Pensiones Piublicas) and regional governments (the Fondos Territoriales de Pensiones Pdblicas) are started to pay pensions to pre-reform pensioners during the reform transition. All private-sector workers can choose between the new ISS and AFPs. Redistribution through the pension system is made transparent and explicit by two new central government transfer programs (Table 2.1). The Fondo Nacional de Solidaridad Pensional (FNSP) subsidizes contributions by lower-income contributors -- a distributive subsidy intended to extend labor- force coverage -- and will be financed by a payroll tax of 1 % charged to higher-income contributors. The Minimum Pension State Guarantee provides a capital subsidy to poor people that have contributed for at least 10 years but have not been able to accumulate a retirement capital required for financing an annuity equivalent to a minimum pension. Contribution rates have been raised for all pension sub-systems to 11.5% in 1994, 12.5% in 1995, and 13.5% in 1996 and thereafter (Table 2.7). The 13.5% paid to AFPs is the sum of 10% for pension contributions and 3.5% for invalidity and survival insurance premia and fund administration costs. All contributors with a wage base exceeding 4 minimum wages pay an additional 1 % contribution to finance FNSP. Colombia's new pension payroll tax rates are slightly higher than those in three other Latin American countries that have reformed their pension systems (Argentina, Chile, and Pern), exceed average contribution rates in Latin America and the Caribbean (10.5%), and are well below those in Eastern Europe and the Former Soviet Union (25.5%) and the OECD (16.3%) (Table 2.8). However, what really matters for forward-looking contributors is not the absolute size of payroll taxes but how closely they are related to expected benefits. Colombia's pension reform establishes a close link between pension contributions and benefits, at least for those workers that choose the fully-funded AFP system, and abstracting from the 1 % payroll surtax on higher-income earnings. As opposed to most other reforms in Latin America, however, the generous current old-pension benefit package for those women (men) of age 35 (40) or older in 1994 who choose to affiliate with ISS  7 has not been modified (Table 2.7). Also the defined-benefit formula for people choosing [SS is unchanged if they are of age 35 (40) or more. This continuation of low retirement ages and generous DB formulae for middle-aged and older cohorts introduces a strong (albeit temporary) bias in favor of ISS affiliation by these age groups, particularly women. Retirement ages for younger contributors at ISS are increased by 2 years and their DB formula differs slightly from the standard DB formula of the old pension scheme (cf. tables 2.7 and 2.5). For all workers -- young and old -- affiliating with AFPs, retirement ages are 60 (62) or any earlier age at which retirement savings finance a pension annuity of at least 110% of the minimum wage. Colombia's benefit structure comprised by its new (steady-state) retirement ages and minimum years of required contributions is very similar to that prevalent in the reformed Latin American economies, but is substantially less generous than that offered in most other, typically unreformed, developing countries (Table 2.8). Finally, Colombia's steady-state retirement ages are only slightly lower while its required years of contributions are higher than in the OECD countries, where life expectancy is well above Colombia's. The Reform Transition Some of the permanent and transition features of Colombia's pension reform are worth to look at because of their implications for the country's fiscal position and macroeconomic outlook. A useful distinction for understanding the pension reform transition and how the implicit PAYG debt is made explicit is between pension rights of current (pre-reform) pensioners and those accrued to current workers in lieu of their affiliation to the old system. Payments of pensions to current pensioners will be done through the newly established national and regional pension funds and those few surviving funds and cajas deemed to be solvent. This will imply explicit transfers from the central and regional governments as long as the last current pensioner survives -- for at least three decades from now on. The second and larger part of the implicit PAYG debt are past pension rights accrued to current (1994) workers. Following the Chilean precedent, pension recognition bonds (bonos de reconocimiento pensionales) will be issued to those currently active workers that choose to shift affiliation from any pre- 1994 pension institution to AFPs or from any pre-1994 pension institution other than ISS to the ISS. Recognition bonds will be issued by the pension institutions were workers were affiliated before shifting to AFPs or ISS. They are debt instruments that make explicit DB pension rights accrued to workers in their old pension funds -- proportional to the number of years of past affiliation -- and that mature at the date of retirement. Recognition bonds will be issued by the central government for pension rights accrued at Cajanal or the national pension funds, by the regional governments for their regional pension funds, and by public and private enterprises for their own funds.8 With regard to institutions, Cajanal is expected to be phased out by 1995 while all other insolvent national pension funds and cajas could be closed in 3-4 years. The insolvent regional pension funds should be phased out in a similar time frame. Although their pension liabilities are the responsibility of 8 The formula for calculation of pension bonds is established by Law 100-1993 and 1994 Ministry of Finance decrees. However, establishing their values is a lengthy and difficult process -- that in Chile took up to 10 years -- requiring establishment of past employment and pension fund histories.  8 the corresponding regional governments, it is anticipated that the central government will play a significant role in bailing them out. In accordance to its new role, ISS has been converted from a government agency to an autonomous and corporatized SOE (empresa industrial y comercial del estado). ISS will continue as a PAYG pension fund in the sense that current pensions will be paid by current capital income and current contributions; hence there is no one-to-one link between contributions and future pension benefits at the individual level. However, the degree of ISS funding -- the ratio of reserves to pension liabilities -- should increase substantially in the future as a result of new affiliations by active workers bringing with them pension recognition bonds. Two years ago Colombia started a system of privately-managed unemployment insurance funds (fondos de cesantfa) that invest mandatory payroll contributions and pay out their capitalized value at labor severance or quit. These funds have been a lab test for the privately-managed pension funds or AFPs that started operation in April 1994. AFPs are receiving contributions from new affiliates since June 1994. Affiliation to AFPs is growing from 0.319 m. in late June to 0.553 m. in late September 1994, the latter figure equivalent to 13% of total pension contributors in late 1993. At retirement AFP contributors withdraw their capitalized pension savings to buy a pension annuity, a deferred pension or programmed pension withdrawals from insurance companies. Hence while active-life pension plans with AFPs are based on defined contributions, retirement-life pension benefits are actuarially-fair defined-benefit plans. The recently established Intendency for Pension Funds in the Superintendence for Banks is mandated with regulation and supervision of both 1SS and AFPs in all facets related to pension contribution collection, establishment and management of pension funds, and investment of pension fund assets. As has happened in Chile since 1980, one could anticipate a positive externality of FF and privately-managed AFPs on the development of capital market instruments and its regulation. In fact, recent reforms of Colombia's financial system in 1990 and 1992 were motivated in part by the subsequent pension reform. The coexistence of two different pension systems potentially enriches the set of consumer choices allowing contributors to decide between defined-contribution (DC) and DB systems. However, because the DB feature is embedded in a state-run PAYG (or partially-funded) pension system, the costs of such a scheme are perpetuated. in fact, considerable financial and economic uncertainties arise from the right conferred to contributors (subject to certain restrictions) to shift every three years their affiliation -- back and forth -- between ISS and AFPs. While currently older cohorts have a clear incentive to remain affiliated with their current pension funds or re-affiliate with ISS, the choices of younger and future generations will be strongly influenced by the expected risk-benefit profile of ISS relative to that of the AFPs. The degree of ISS funding and its political and financial autonomy from the government will be paramount in contributor decisions.  9 3. FISCAL IMPLICATIONS OF PENSION REFORM Colombia's reform -- like other pension reforms in Latin America -- combines two very different changes: a reduction in net benefits paid to future pensioners by the existing PAYG system, and a gradual substitution and partial privatization of the existing PAYG scheme by a dual PAYG-FF system. The first reform component reduces pension system deficits and the implicit PAYG debt of the government incurred vis-a-vis future pensioners and improves the financial position of public pension institutions. The second reform component entails a regime change that makes explicit the currently implicit PAYG debt as current PAYG pensioners are paid off and past PAYG pension liabilities accrued to workers are paid to those shifting to a FF scheme. During this extended horizon the making explicit of PAYG debt is reflected by reform transition deficits incurred by the government. This section reports numerical simulation results for the government's reform transition deficits and their capitalized value. Sensitivity analyses are performed for alternative economic and systemic assumptions. Finally, Colombia's projected figures are compared to those of Chile's post-reform experience and to implicit PAYG liabilities in the major OECD economies. Fiscal Effects of Colombia's Pension Reform Colombia's pension reform has significant consequences for the financial position of public pension institutions and therefore will impinge on official government finance indicators. This does not necessarily mean that overall public finance positions are affected by the reform. As discussed in subsequent sections, if the pension reform is financed by issuing public debt, the latter simply substitutes implicit PAYG so that overall (i.e. correctly measured) government debt indicators do no change. As mentioned above, the reform combines two different changes: the reduction in net benefits paid to future pensioners by the existing PAYG system, and the substitution of the existing PAYG pension scheme by a dual PAYG-FF system. The opposite effects of these two changes on Colombia's public finances are disentangled next for a better understanding of their very different nature. This section reports numerical simulation results of the fiscal effects of Colombia's pension system changes. They are based on an actuarial model that computes annual pension revenues, expenditures, surpluses, and outstanding net liabilities for the three components of Colombia's current mandatory pension system: (i) the government pension system (SPP) comprised by Cajanal, the national, and the regional pension funds and cajas as well as their successor transfer programs (FNPP and FTPP), (ii) the public Instituto de Seguros Sociales (ISS), and (iii) the new privately-managed pension funds or AFPs.' 9 It should be noted that the projections of contributors and pensioners (and hence of deficits and outstanding debts) for SPP include all public-sector workers that will re-affiliate with ISS now or in the future, while the corresponding projections for ISS exclude the latter group of workers. Hence this unrealistic but useful simplifying assumption affects the projected composition but not the overall levels of worker affiliation, deficits, and debts of the total public pension system (comprised by SPP plus ISS).  10 The simulations performed for this study were requested for this study through the Ministry of Finance of Colombia and are reported in detail by Helmsdorff (1994). The simulation model and its applications follow similar exercises reported in Ministerio de Seguridad Social y Trabajo (1992, 1993a) and Lora, Zuleta, and Helmsdorff (1992). There are two major differences between the model used here and its predecessors. First, this model embodies the features of the actual pension reform reflected in Law 100-1993 (Ministerio de Seguridad Social y del Trabajo, 1993b) as opposed to those of preliminary reform projects and proposals embedded in the other studies."0 Second, this study allows for a series of step-wise simulations that identify the separate contribution of demographic, systemic-reform, and economic factors to transition deficit and debt levels. However, both the model used here and its predecessors are partial-equilibrium accounting frameworks, devoid of behavioral economic content, that treat relevant macroeconomic variables -- such as growth and interest rates -- as pre-determined. However, the simplicity of these frameworks make them useful devices to provide actuarial projections of fiscal deficits and debt levels for a given macroeconomic environment. The assumptions on coefficient values and simulation features are summarized in Annex 1. Eight simulations are performed below. Table 3.1 summarizes their main features. The simulations follow a sequence starting with a no-reform scenario to follow with a step-wise introduction of relevant features and necessary assumptions related to Colombia's pension reform. Some variations of the base scenario test for the sensitivity of the results to alternative assumptions on economic variables and the speed of private system affiliation. The main economic variables shaping the fiscal consequences of pension reform are the real rates of interest, GDP growth, and wage growth. Uncertainty related to pension system affiliation is much higher in Colombia's reform than, for instance, in the cases of Chile or Argentina, because of Colombia's low labor force coverage and lower pension system and demographic maturity, and most important, because the new pension law allows for re-affiliation between the AFPs and the ISS. Alternative time profiles for the numbers of contributors to ISS, SPP, and AFPs are considered first. Figure 3.1 shows projected contributors to each pension system component for three different scenarios: (i) constant labor-force coverage and no reform (simulation 1), (ii) gradual and large rise in labor-force coverage of the pension system from 29.6% in 1994 to 46.6% in 2025, for simulations with no reform (simulation 2) and with partial reform (simulation 3), and (iii) actual (full) reform case, with a gradual and large rise in labor force coverage of the pension system from 29.6% in 1994 to 46.6% in 2025 (simulations 4-6). 10 Ministerio de Seguridad Social y del Trabajo (1992) reports one actuarial partial-equilibrium simulation of net surpluses and assets for ISS, SPP and private pension funds (AFPs) for the pension reform. Ministerio de Trabajo y Seguridad Social (1993a) provides actuarial partial-equilibrium simulations of net surpluses and assets for ISS with and without pension reform and for AFPs under the pension reform, for the projection period 1992-2020 and under alternative sets of economic and demographic assumptions. Lora, Zuleta and Helmsdorff (1993) report alternative simulations for ISS and AFP net assets under different reform scenarios.  11 Under simulations 1-3, SPP contributors grow slightly from 800,000 contributors in 1994 to 850,000 in 2025." With constant labor-force coverage in simulation 1, ISS grows from 3.8 m. in 1994 to 6.1 m. in 2025. However, when coverage is raised all additionally covered workers affiliate with ISS so that ISS grows massively to 10.7 m. contributors in 2025 under simulations 2 and 3.12 Finally, under the reform-case simulations 4-6, the shift of affiliation from ISS and SPP to AFPs implies that AFP contributors increase from 600,000 at the end of 1994 to 10.1 m. in 2025, at the expense of declining contributors at ISS (with only 600,000 in 2025) and at SPP (with 200,000 in 2025). The projected shift of affiliation to AFPs is massive but gradual in the base case, reflecting the assumption that the risk- return frontier offered by the AFPs dominates that of ISS during the length of the transition horizon. Strong AFP growth in the reform base case is consistent with the experience observed during the initial reform months and the forecasts by market participants. However, the large difference in institutional affiliation between simulations 2-3 and 4-6 is subject to the uncertainty of any long-term projection of institutional shares of the population covered by a mandatory pension system. To address this uncertainty, simulations 7 and 8 -- not summarized in figure 3.1 -- show the sensitivity of the base- case results to alternative speeds of affiliation from ISS and SPP to AFPs. Let's now turn to the fiscal effects of the pension reform. Table 3.2 reports first and last-year deficits of ISS and SPP during the simulation horizon 1994-2025 and figure 3.2 shows the complete time profile of annual deficits of the mandatory public pension system (ISS and SPP combined) for simulations 1-4. Under the constant-coverage no-reform simulation 1, the total public pension system deficit rises from 1.1 % of GDP in 1994 to 3.5% of GDP in 2025. This reflects what would have happened to Colombia's public PAYG scheme in the absence of any reform -- an explosive deficit path as a result of pension system maturation and a graying population.'3 The counter-factual scenario of no-reform but increasing pension system coverage implies postponing to the future the deficits embedded in the old PAYG system. When affiliation to ISS grows strongly during 1994-2025, a larger base of contributors sustains over a longer period the exponentially growing number of ISS pensioners. Therefore the time profile of ISS deficits (and of total deficits in figure 3.2) is shifted down, but toward the end of the projection horizon deficits start to grow strongly, reaching 2.99% of GDP in 2025. The (still counter-factual) scenario 3 combines an increase in coverage with a partial pension reform. The differences in levels and time profile of deficits with simulation 2 are staggering. Raising both contribution rates and retirement ages has a massive and permanent effect on the financial position of both ISS and SPP. Total pension deficits decline immediately -- in 1994 -- to 0.71 % of GDP and continue falling subsequently to reach a bottom level of 0.15% of GDP in 2017 and 2018, to start rising " These figures exclude the roughly 300,000 contributors to the three public pension funds exempted from the reform: teachers, oil workers, and armed forces personnel. 12 A significant increase in labor force coverage as a result of Colombia's pension reform is highly likely as a result of both the reduction in the pure tax component of pension contributions and a partial subsidy to pension contributions by low-income groups. " The aggregate deficit of ISS and SPP hides a quite different financial picture in each institution. While SPP shows deficits in all years and simulations, ISS shows (declining) surpluses until 2007 in simulation 1, 2015 in simulation 2, 2024 in simulation 3, and 1995 in scenario 4, and deficits thereafter.  12 again, attaining 0.88% of GDP in 2025. This confirms the conclusion -- observed in other pension reform cases such as Chile -- that raising contribution rates and retirement ages has a large impact in reducing pension system deficits and hence the size of the implicit PAYG debt. However, as opposed to Chile where retirement ages were raised for younger and older cohorts of contributors, the increase in retirement ages affects only the younger generations. If they had been raised for the older, too, transition deficits would have been lower. Now consider the actual reform base case embedded in simulation 4. Introducing the option of affiliation with AFPs implies that ISS and SPP lose gradually their contributor bases while continuing pension payments to their own current and future pensioners and to those who have moved to AFPs. The time profile of total ISS and SPP deficits starts at 0.88% of GDP in 1994, reaches a peak of 2.61 % in 2013, and then declines to 1.92% of GDP in 2025 (and even further subsequently). A more disaggregate projection of annual deficits by institutions and types of deficits is summarized in Table 3.3. Recognition bond deficits of ISS and SPP arise from honoring at worker retirement the recognition bonds issued to those that have shifted affiliation to AFPs. Recognition bond deficits start only in 2003 for ISS and peak at 0.61 % of GDP in 2020-21. SPP recognition bond deficits start later and are close to zero, as only a small share of SPP contributors are expected to shift to AFPs. Operational pension deficits arise from the shortfall of declining contributions from pension payments to remaining pensioners. Operational pension deficits exceed significantly recognition bond deficits in both institutions. ISS shows slight operational surpluses in 1994-95 that turn subsequently into deficits peaking at 0.84% of GDP in 2013. SPP operational deficits start at a high 0.97% of GDP in 1994 and peak at 1.41 % of GDP in 2011-13. In the reform base-case simulation 4 -- as well as in simulations 1-3 and 7-8 -- real GDP growth (g) is equal to the real interest rate (r). This allows to add up annual deficit ratios to GDP in order to obtain the capitalized deficit ratio or cumulative pension debt ratio to GDP. This value is the present value pension debt ratio to GDP at any year, today or in 2025. Adding up 1994-2025 annual deficit ratios yields the cumulative pension debt ratio to GDP for that period. Total ISS and SPP recognition bond debt incurred during 1994-2025 is 9.0% of GDP, only one sixth of total operational deficit debt at 54.2% of GDP. Hence total explicit pension debt incurred during 1994-2025 under the reform base-case simulation is estimated at 63.2% of GDP (Tables 3.3 and 3.4). It is expected that pension transition deficits of both ISS and SPP will continue beyond 2025, converging toward zero around 2060. SPP (including the transfer programs FNPP and the FTPPs) will be extinguished by then. However, one may anticipate continuation of ISS with a core number of affiliates for the very long term, implying a corresponding implicit PAYG debt associated to this group. For the total pension transition period 1994-2060, explicit debt levels attain 14.4% of GDP for the recognition-bond deficit component and 69.2% of GDP for the operational deficit part. This implies reaching a grand total of 83.6% of GDP of explicit pension debt, that requires an identical amount of explicit government financing. How much of this explicit debt is due to pension system deficits that would have been observed under a partial pension reform and how much is due to making explicit the implicit PAYG debt as people shift to the fully-funded AFPs? The answer lies in cornparing the cumulative explicit pension debts for the partial-reform simulation 3 and the full-reform base-case simulation 4 (Table 3.5). The cumulative 1994-2060 explicit debt under simulation 3 is 24.4% of GDP, falling 59.2% of GDP short of the cumulative explicit debt of the base-case simulation. Hence the implicit PAYG debt of Colombia's  13 pension system (evaluated at contribution and benefit levels of the new system) that is made explicit by the actual reform is 59.2% of GDP. Adding to this figure the remaining steady-state implicit PAYG debt due to the survival of ISS with a core number of affiliates in 2060 and beyond -- estimated at 4.5% of GDP " -- implies a total implicit PAYG debt of 63.7% of GDP. Finally, the total sum of Colombia's explicit pension reform and remaining implicit PAYG debt at 88.1 % of GDP reflects what would have been the total explicit debt if the PAYG system (ISS) had been phased out entirely by the pension reform, as happened in Chile. Let's now return to the other results by comparing the debt levels incurred by SPP and ISS under the eight simulations to and the asset levels accrued to AFP affiliates under simulations 4-8 (Table 3.6). This comparison is limited to the first 32 years of the transition period (1994-2025). Under constant coverage and no reform (simulation 1), the total (ISS and SPP) explicit pension debt is 60.6% of GDP. Increasing ISS coverage (simulation 2) raises ISS contributions for an extended (but temporary) period, leaving ISS with net assets of 1.0% of GDP in 2025 and postponing for subsequent decades the accumulation of net ISS debt. The financial effectiveness of the partial pension reform -- comprising higher contribution rates for all and higher retirement ages for younger cohorts -- is reflected by an increase in ISS assets and an increase in SPP debt ratios, each by 18 percentage points of GDP (simulation 3). Hence total public pension debt shrinks from 48% of GDP (simulation 2) to 12.4% of GDP (simulation 3). As discussed above, the actual reform base case (simulation 4) implies an explicit public pension debt level of 63.2% of GDP. Private pension (AFP) system assets accumulated during 1994-2025 attain 52.3% of GDP. It is important to note that this figure is very close to 50.8% of GDP. The latter is the difference between simulation-3 and simulation-4 explicit debt levels, i.e., the implicit PAYG debt made explicit by the shift of pension system contributors to AFPs during the first 37 years of the reform.'5 Simulations 5 and 6 reflect alternative assumptions on growth and interest rates. Under lower GDP and real wage growth rates (simulation 5) than in the base-case (simulation 4), the ratios to GDP of public pension liabilities and private assets reach higher levels. When both interest (r) and growth (g) rates increase, but r still exceeds g (simulation 6), public pension debt ratios rise slightly and private asset ratios rise substantially as compared to the base case. Simulations 7 and 8 report variations based on alternative assumptions on the speed of shift from public to private pension system affiliation. As opposed to changes in economic variables, variations in the velocity of transfer from the public to the private funds affects only the timing but not the long-term pension transition debt. Slowing down (raising) the transfer speed in simulation 7 (simulation 8) reduces '4 For the purpose of pension debt simulations -- and their overall public-sector impact -- it is immaterial how much of this implicit PAYG debt will be explicitly recognized -- and paid -- by the public pension institutions (or their sponsors) that will loose affiliates, being forced to honor recognition bonds and hence to partially fund ISS. The resulting improvement in ISS reserves simply offsets a net asset deterioration in another public-sector institution. 15 The reason for the difference is that the rate of return on private pension assets (5%) is higher than the rate of interest paid on recognition bonds (4%).  14 (raises) public liabilities and private assets at year 2025 but does not have an impact on the levels which would be attained by, say, the year 2080. However, the substantial differences in transfer speed between simulations 4, 7, and 8 have some effects on the debt levels attained in 2025. We conclude that Colombia's pension reform will imply substantial transition deficits. Their values are sensitive to three highly uncertain conditions: the success of Colombia's new mandatory pension system in extending worker coverage, the speed of re-affiliation from ISS-SPP to AFPs (as well as their possible return to the ISS), and macroeconomic performance. On the latter this section has assumed that growth, interest, and wage rates are unaffected by the pension reform itself -- a simplifying assumption of actuarial projections that will be lifted in section 5.2 below. Pension Reform Deficits and Implicit Pension Liabilities in Other Countries How do Colombia's projected pension transition deficits and pension debt levels compare to other countries? We start by looking at the Chilean pension reform case, the single case of a radical pension reform with some relevant post-reform experience. Chile applied a two-stage reform by first reducing substantially its implicit PAYG debt as a result of raising substantially net retirement ages (in 1979) and, second, by substituting its state-run PAYG scheme by a privately-managed FF scheme (in 1981). Actual (i.e. ex post) reform transition deficits increased from 3.2% of GDP in 1982 to a peak of 4.8% of GDP in 1991 (Table 3.7). They are projected to decline gradually thereafter and disappear after 2025-2030. The two components of the pension reform deficit behave very differently, a result of system maturity, demographics and pension reform design. Recognition bond deficits are much lower and peak later than the very substantial operational deficits. There is much less uncertainty related to the projection of future pension deficits in Chile than in Colombia because of Chile's more mature pension system and population structure, its larger population coverage, and its phasing-out of its state PAYG institutions and their transition successors. Comparing Colombia's base-case reform transition deficits with Chile's deficits (cf. Tables 3.3 and 3.7), two differences are apparent. Colombia's deficits are lower and are stretched over a longer time period than Chile's. Both features are reflected by the fact that Colombia's peak pension deficit is only 2.6% of GDP (compared to Chile's 4.8%) and occurs 20 years after the reform start (as compared to 11 years in Chile). The reasons behind this are Colombia's smaller pension system coverage (although a significant increase over actual 1994 coverage is embedded in the latter projection) and its slower and incomplete substitution of the initial PAYG scheme. Comparing both countries' net pension liabilities, Colombia's 86.5% of GDP is significantly lower than Chile's comparable 126% of GDP (Table 3.8). Colombia's figure is also lower than that in 6 of the G-7 countries, where comparable ratios range between 44% of GDP in the U.S. and 251 % of GDP in Canada. Colombia's generally lower figure should not surprise as the OECD countries are characterized by large coverage, mature PAYG schemes, and old populations. The implication of this international comparison for Colombia is that although its pension system is relatively small, the country has incurred in pension liabilities that are sizable relative to other countries, implying a long period of substantial reform deficits. Hence Colombia will have to define the ways by which it will finance its transition deficits. As discussed in the next section, different ways of financing have very different macroeconomic and welfare implications.  15 4. MACROECONOMIC AND WELFARE IMPLICATIONS OF PENSION REFORM IN OTHER COUNTRIES Pension reform -- substitution of a FF privately-managed pension system for a PAYG state- managed system -- can have significant positive effects on growth and welfare. They occur through three channels: more saving, higher formal-sector employment and production, and development of capital markets. These three links are analyzed next, drawing from analytical and simulation results. A more detailed discussion is provided in Annex 2. A brief look at Chile's post-reform experience illustrates the empirical dimension of the potential macroeconomic effects of pension reform. Pension Reform, Saving, and Growth Substituting a FF for a PAYG pension system involves incurring in reform transition deficits like those for the Colombian reform projected in the preceding section. How the government finances transition deficits is the single most important factor in determining the long-run saving and derived growth effects of such a reform. If the government swaps the implicit PAYG debt for explicit new government debt -- and under certain additional conditions (see Annex 2) -- long-term saving and output levels are not affected by the reform. This first choice could literally imply issuing more government debt or else could involve sale of government assets, such as SOEs (through privatization), or government reserves. The important point to realize is that financing the pension reform in this way does not change the government's net asset position and hence does not shift income across different generations, leaving their wealth and saving levels unaltered. The second way of government financing is through fiscal contraction, i.e., by raising taxes or cutting government expenditure. This choice imposes an income loss on all generations that have to pay for higher taxes or that suffer from lower governmnent spending. And it involves an income gain for all future generations -- those living after the entire implicit PAYG debt has been paid off -- freed from PAYG contributions and unaffected from transitional tax payments or expenditure cuts affecting their predecessors. Hence income and saving of the transition generations is cut while income and saving of future generations is raised. The latter's higher saving could feed into larger domestic investment and hence raise long-term output and welfare levels. One should note that this second option combines a "pure" pension reform embodied in the first option with a contractionary fiscal policy. With little country evidence on the saving and growth effects of pension reform, one has to rely on simulation results of those few studies that have assessed the short and long-run fiscal, output, and welfare effects of introducing or substituting mandatory pension systems. These are four simulation studies that quantify the output and welfare level effects of pension systems and reforms when long-run growth is exogenous. Their main results are summarized in Table 4.1. The first study is for the U.S. economy (Auerbach and Kotlikoff, 1987, denoted AK), the second is for a set of representative economies (Arrau and Schmidt-Hebbel, 1993, denoted AS); and the third and fourth are also for a representative economies (Valdes-Prieto and Cifuentes 1993, denoted VC, and Cifuentes and Vald6s- Prieto 1994, denoted CV)."6 16 The focus of our discussion is only on long-run stationary effects, although three of the four studies show results for the entire transition path after a pension system is started or reformed; often short and medium-term effects on all macroeconomic variables differ strongly from the steady-state effects discussed here.  16 Starting PAYG shifts resources from future to current generations. For the U.S., adoption of a PAYG system is estimated by AK to reduce long-term output levels by figures close to 5 %, with minor differences depending on how general taxes are raised (either on general income, wage income, or consumption). The corresponding long-term welfare (wealth equivalent) losses of future cohorts are 5 to 6%. The AS simulations for representative economies distinguish between two demographic scenarios (high and zero population growth) and how the transition deficit is financed (debt or taxes). Consider first the case of high population growth. When the fiscal transition deficit is financed by issuing explicit government debt (i.e., the case of a straightforward pension reform), the implicit PAYG debt is put on government books. Therefore the explicit government debt increases significantly, although this massive debt build-up does not crowd out private investment. The reason is that the reform raises both demand and supply of government debt, as new worker contributions to the FF system are invested in newly issued government debt during the 45 years of fiscal transition deficit. The income tax rate increases with the stock of public debt because higher taxes are required to finance interest payments on the higher explicit debt. While direct intergenerational transfers are ruled out by debt financing, higher income taxation for paying interest on the explicit debt imposes a slight but permanent efficiency cost. Hence long-run saving, capital, investment and output levels are slightly but negatively affected by debt financing. Output at year 110 is I % lower than under the initial PAYG scheme, reflecting the full impact of the modest efficiency loss from higher income taxation. The welfare loss of future steady-state generations is 0.3%, derived from permanently higher income taxation. It is important to note that this loss could be a net gain when labor is supplied elastically and the associated labor market efficiency gain more than offsets the income-tax efficiency loss from higher income taxation. When the transition deficit is financed by taxes, the pension reform is actually combined with a contractionary fiscal policy. This mix hurts tax-paying transition generations and benefits post-transition cohorts. The transfer to future generations raises long-run saving, capital and output levels. However, long-run output gains of a fully tax-financed pension reform are modest. By the year 110, output exceeds the level it would have attained under the old PAYG system by only 3 %. Future generations gain 6.8 % of their wealth as a result of both the transfer from the tax-paying transition cohorts (which pay off the initial implicit PAYG debt) and a small efficiency gain due to slightly lower income taxes. For a stationary population the qualitative results remain unchanged although their size is larger. The reason that the reform effects grow with the old-age dependency ratio is the larger initial PAYG debt, implying stronger efficiency effects and, in the case of tax financing, a larger transfer toward future generations. Consider now the simulation results by VC and CV. As opposed to the two preceding studies, they introduce heterogeneous consumer groups with different degrees of myopia (that is, dissimilar subjective discount rates) in combination with the possibility of credit constraints. The latter hit consumers with high discount rates because of the additional restriction that non-human wealth has to be non-negative at any point in time. The VC study allows to assess the important role played by the group of myopic and credit- constrained individuals when substituting PAYG by FF. Without binding credit constraints the long-term effects of a tax-financed pension reform are modest, similar to the results shown by AS. However, when  17 widespread myopia-cum-credit-constraints is considered, the pension reform boosts (involuntary) saving significantly, so that the long-term output level gain rises 14-fold, from 1.9% to 27.1%. Welfare increases by significantly less because of the involuntary shift of consumption toward the future imposed on credit-constrained myopes. However, the large size of these effects (and those of CV reported next) is due in part to the assumption that FF savings are exempt from income taxation. This assumption -- not made in the two preceding studies -- provides an additional incentive to saving and hence capital formation when adopting a FF system. Finally consider the CV results. that allow to distinguish -- only for the case of binding credit constraints for a group of myopes -- between steady-state effects under two different financing options for the reform transition deficits. When debt-financing is large relative to tax financing (75% and 25%, respectively), the long-term output gain is 7%. When the transition deficit is fully tax-financed, the long- term output gain rises to 21.8%. The latter figure is quite large, a result due in part to two important assumptions: FF savings exemption from income taxation and very low stationary GDP growth (I %). In sum, the simulation results of the four studies report modest to moderate long-term changes in output and welfare levels caused by a pension reform. And in the few cases where long-term percentage effects reach double-digit levels, these effects are only reaped decades after the reform has been initiated. Could larger and permanent growth effects be expected when the structure of employment and production is allowed to respond to pension reform? To this question we turn next. Pension Reform, Employment, and Growth Phasing out PAYG in favor of FF eliminates the pure tax component of pension payroll contributions, i.e., that part of each individual's payroll taxes that is not closely linked to her expected future pension benefits. This pure tax component is large in countries where the pension system is strongly redistributive (and not necessarily toward the old-age poor) and, because of frequent changes in pension system rules and uncertainty about future benefits, people generally perceive a weak link between their payments and their future benefits. Eliminating the PAYG tax component reduces the incentive of payroll tax evasion and could therefore increase overall employment and induce a shift of labor from informal )or tax-evading) sectors to formal (or tax-paying) sectors. Because of generally low aggregate labor supply elasticities it is unlikely that pension reform would raise significantly overall employment. However, a large shift of employment and production from informal to formal-sector activities could be observed in response to the elimination of the pure tax on formal activities. A study by Corsetti and Schmidt-Hebbel (1994) reports simulations of long-run growth effects of a PAYG-FF pension reform. It is based on a two-sector (informal-formal) economy with two overlapping generations, where a PAYG system transfers resources from young to old and imposes a welfare and growth cost by discouraging production in the formal sector (which uses both capital and labor) and spurs production in the informal sector (which uses only labor). Because capital is the only ultimately productive factor in this endogenous-growth model, growth effects of the PAYG distortion are sizeable. Table 4.2 summarizes the steady-state growth simulation results based on a parameterization for a representative economy. Under any mandatory pension contribution rate, growth reaches 3.7% per year under a FF pension system. A PAYG system with a 10% contribution rate that does not distort labor market and production decisions -- but transfers resources to the (first) generations of retired  18 people -- reduces growth to 3.4% as a result of the inter-generational transfer. If in addition PAYG distorts labor decisions and hence shifts production from the formal to the informal economy, growth falls by an additional 0.4 percentage points, to 3.0% per year. We conclude that the potential growth effects of substituting FF for PAYG could be substantial if PAYG is very distortionary, capital is only ultimate productive resource, and the informal sector is less productive than the formal economy. Capital Market Development and Growth The least understood and documented channel by which pension reform could raise growth is through the development of capital markets. However, one can point toward four potential efficiency and growth gains that could be reaped from a radical pension reform. First, privatization of the pension system (of either a FF or a PAYG system) tends to improve the quality and efficiency of pension services. Efficiency arguments in favor of private industries of banking, mutual-funds and insurance services over state-managed financial services also apply here. Casual evidence on the dismal quality of pension services provided by public pension institutions in developing countries can be complemented with more systematic cross-country figures for real rates of investment returns for partially or fully-funded publicly and privately managed pension funds (Table 4.3). That evidence shows a massively superior performance of privately managed pension funds."7 The dismal performance of public funds is not only the result of weaker performance incentives of public fund managers but also of their portfolio bias toward government and SOE securities or public-housing authorities, typically at below-market conditions. Second, pension reform contributes to a concentration of the public sector on (i) more effective redistribution by running a transparent and well-defined transfer program toward the old-age poor, and (ii) more effective regulation and supervision of private pension fund managers. This specialization could improve the quality of both functions and reduce their costs. Third, a FF pension system run by the private sector and endowed with broad choices on investment portfolios contributes to the development of capital markets in general and of long-term securities in particular. Substituting the intergenerational contract implicit in PAYG by an explicit demand for long-term securities tends to boost the development of private equity and securities markets. This could reduce the cost of capital for investment, lengthen the maturity of financial instruments, and hence boost capital formation and growth. Finally, adoption of a FF pension system based on defined-contribution plans with individual accounts could boost the awareness of savers of the need for attaining adequate levels of old-age savings. This reduction in myopia or personal discount rates increases net private saving volumes, complementing any long-term effects reaped from financing transition deficits through fiscal contraction. 17 The net rates of return in table 4.3 are not adjusted for risk. While the ranking of individual pension systems would change if risk-adjusted net returns were computed, it is likely that the general conclusion on the better returns of privately run systems would be upheld.  19 Chile before and after the Pension Reform What does the Chilean experience before and after its 1981 pension reform suggest? Table 4.4 provides a summary of its recent performance. It reports evidence that the relative size of formal labor markets and production sectors has grown in Chile after 1980. In addition, a massive improvement in private saving and overall growth performance took place during the last decade. A closer look at the changes in economic structure, private saving, and GDP growth is warranted."8 Chile's pension reform involved a reduction of overall social security contribution rates from 29.3% to 17%, of which the contribution to the new FF scheme is 10% (table 3.1). The new system provides a close relation between earnings and pension benefits, that was absent under the preceding PAYG regime. The estimated reduction in the pure tax component of pension contributions -- from 16.0% of net wages in 1980 to 6.8% in 1982-1985 and 2.8% in 1990-1992 -- may have contributed to higher net wages, lower gross wages, and higher employment in formal labor markets. Possible efficiency gains in labor markets are suggested by the following changes. The share of independent workers in the labor force (who are not required to contribute to mandatory pension schemes) has declined from 26% before the reform to an average 24.5% after the mid 1980s, signalling an increase of both potential pension contributors and formal labor markets. More direct evidence on the change in the formal-informal structure of labor markets is provided by the significant decline in the relative share of informal-sector employment, from 36.0% in 1980 to 31.1 % in 1990-92. In addition, male labor force participation -- that could reflect the incentive effect of the reform on total male employment -- has increased slightly during the last years. More ambiguous is the behavior of the share of active contributors to pension systems (comprising contributors to both the old and new schemes) in total employment, starting at 62.5% in the early 1980s to decline thereafter and recovering to an estimated 63% in 1993. The growth of private pension funds has contributed to new equity issues and stock market capitalization in Chile. Stock market capitalization has tripled since 1980 to an average 72.2% in the early 1990s. The ratio of total pension fund investments to GDP reached 31. 1% in the early 199os and exceed 40% of GDP in 1994. Another major change observed in Chile is the massive increase in private sector saving from levels close to zero in 1979-81 to an average 17.1 % of GDP in 1990-92. This radical departure from the past has made possible both higher investment levels and lower foreign saving inflows. The private consumption share in GDP shows a 10 percentage point decline, from 73% in 1960-1981 to 63% in 1986- 92. Econometric evidence suggests that growth of pension asset holdings may have contributed to the increase in private saving in Chile (Corsetti and Schmidt-Hebbel 1995). 18 However, this look should be qualified by stating that Chile underwent many contemporaneous structural changes and suffered various external shocks during the period of pension reform. This implies that one should be careful -- in the absence of a well-specified framework able to distinguish between different reforms and shocks -- in attributing to much to pension reform alone. For a comprehensive assessment of Chile's performance during the last 20 years see Bosworth, Dornbusch and Laban (1994) and for an excellent account of Chile's pension reform see Diamond and Valdes-Prieto (1994) published in that volume.  20 Per capita GDP growth has risen significantly since the mid-1980s, exceeding 5% per year. Higher factor productivity explains part of this growth spurt. Both the elimination of the pure tax component of PAYG and the deepening of financial markets resulting from pension reform may be having a significant influence on growth. Summing up, Chile has shown significant improvements in labor market formalization, capital market development, private saving, and growth after 1981. However, one should be careful in attributing too quickly a large contribution to pension reform in these results, as that requires controlling for other intervening factors, including reforms in other policy areas. Hence we only conclude tentatively that pension reform is a possible explanation -- among others -- for some of the macroeconomic and factor market improvements observed in Chile since 1980. 5. FISCAL AND MACROECONOMIC IMPLICATIONS OF COLOMBIA'S PENSION REFORM Colombia's GDP growth reached an average annual 3.5% during the 1980s, exceeding by much the depressed 1.4% average GDP growth rate in the rest of Latin America and the Caribbean."9 This better-than-average result was largely due to Colombia's more consistent and careful macroeconomic policy stance. But such a performance is still insufficient to lead to significant improvements in per-capita living standards and to overcome poverty. This realization led the Colombian government to initiate deep structural reforms since 1990, including trade reform, liberalization of the domestic financial sector and international capital flows, public sector reform, and limited privatization of SOEs. At the same time, public sector activity and spending were reoriented toward social sectors and poverty elimination. Part of the latter reorientation is reflected by the 1993 pension and health reforms. While Colombia has faced strong fluctuations in its external terms of trade during this period of reforms, it is benefitting from major recent oil discoveries in Cusiana and Cupiagua. The present value of 1993-2005 net income flows of the oil discoveries is estimated at US$ 15.2 b. (or 26.1 % of 1994 GDP), of which the government is entitled to 81 % or an estimated US$ 12.4 b. (equivalent to 21.2% of 1994 GDP). Therefore a major macroeconomic challenge facing Colombia in the near future is to manage its oil bonanza in a way that is consistent with pursuing a sustainable fiscal policy and a structural reform program that includes a significant pension reform. Against this background this section reviews Colombia's options for financing its pension reform deficit, reports simulation results for the long-run output and welfare effects of a reform financed by fiscal contraction, and makes a qualitative assessment of the overall consequences of pension reform. Financing Options Colombia is facing a long period of pension reform deficits that start at 0.9% of GDP in 1994 and peak at 2.6% in 2013, falling gradually thereafter to disappear by, say, the year 2060. The '9 For recent reviews of Colombia's macroeconomic, growth, and equity performance and perspectives see, among others, Garcfa (1995), Montenegro (1995), Posada and Gaviria (1995), and Uribe (1995).  21 capitalized value of pension deficits incurred until 2060 could reach a total of 83.6% of GDP, according to the base-case reform simulations summarized above (Tables 3.3 and 3.4). While pension deficits will occur during a very long time horizon during which many other shocks will impinge on public finances, it is sensible to identify now possible financing sources for the significant and protracted pension reform deficits arising from making explicit the government's implicit debt incurred in the past by its PAYG pension system. Obviously this decision has to be taken as part of the government's overall financial programming that requires deciding on the size and financing of its overall officially recorded deficit. A look at Colombia's fiscal stance reveals a picture of a growing public sector (as measured by expenditure and revenue shares in GDP) that, however, has been able to maintain a roughly balanced budget and low levels of indebtedness (Table 5. 1). The deficit of the consolidated non-financial public sector (comprising central and regional governments as well as the surplus of SOEs) was close to zero in 1990-92 and a small surplus was recorded in 1993. Domestic debt of the central government is very low, standing at 6.0% of GDP in 1993, and the external medium and long-term public-sector debt stands at 25.7% of GDP in 1993. These fiscal conditions are much more favorable than those in most other OECD and developing countries, including those prevailing in Chile during the first 5 years after the start of its pension reform (cf. Table 4.4). Following on the discussion in section 4. 1, Colombia's government faces two fundamental choices in deciding how to finance its pension reform deficit: (a) Paying off the implicit PAYG debt by a combination of issuing domestic debt, using future net income from oil discoveries, using revenue from SOE privatization, and cutting government capital expenditure; or (b) reducing its non-pension fiscal deficit by raising taxes or reducing current expenditure. As discussed above, the first option is the "purest" form of pension reform. It leaves the government's net total wealth or asset position (including explicit and implicit liabilities) unchanged even though the official government deficit records an increase by the amount of the pension deficit. The second option, however, avoids an increase in the official government deficit although it implies an increase of net total (explicit and implicit) government wealth. This alternative also implies an income transfer from current and future transition generations (those hurt by higher taxes or lower government expenditure) toward long-term future generations freed from PAYG transfers (which would have been paid if the old PAYG system were maintained) or higher taxes/lower public expenditure (which would have been required to finance the shortfall of returns as a result of lower non-pension net assets under the first option). In deciding among these choices -- or any combination of them -- the following criteria should be considered. (i) Resources Available from New Oil Discoveries and Privatization. While not all resources from oil discoveries and privatization could be used for paying for pension deficits (due to budgetary and political pre-commitments and restrictions), it is useful to compare the upper bounds of potentially available resources with the pension resource needs. Adding the total estimated present value of net income accruing to the government from the oil discoveries (21.2% of GDP) to a rough estimate of the present  22 value of privatization revenue (10.0% of GDP)' yields a figure of 31.2% of GDP, that falls short of total long-term pension financing requirements (83.6% of GDP). Still, the figure of 31.2% of GDP is equivalent to the capitalized value of total pension deficits during the first 18 years of reform (1994-2011, cf. Table 3.3). (ii) Financial Effects of Issuing Explicit Domestic Debt. If all government revenue from oil discoveries and privatization were used for paying off part of the pension debt, there would still be a sizable amount -- 52.4% of GDP -- requiring other sources of financing. What effect would be felt in domestic financial markets -- particularly on interest rates -- if that amount were entirely financed by issuing domestic debt, hence raising the outstanding domestic debt from 6.0% in 1993 to 58.4% of GDP by, say, 2060? In answering this question it is important to keep in mind that issuing explicit domestic debt for paying off implicit government debt does not have first-round macroeconomic and financial effects if financial-market participants see through this debt swap, i.e., do not suffer from fiscal illusion. If financial market participants, including the new AFPs, realize that the government's net wealth position is unaffected by a debt-financed pension reform that simply increases both the supply of government debt and the demand for government paper (through the investment of AFP resources), interest rates on government debt (and hence domestic interest rates at large) should be unaffected. This seems to have happened in the aftermath of the Chilean reform throughout the 1980s when explicit government debt grew rapidly while PAYG pension debt was repaid, but real interest rates did not change much (see Table 4.4). However, some degree of fiscal illusion should not be ruled out a priori, justifying a careful initial use of debt financing.2" (iii) Intergenerational redistribution, savng and growth effects offinancing the pension deficit by reducing the non-pension deficit. As discussed above, a pension reform combined with a contractionary fiscal policy is what the second option is about. It involves a resource transfer from transition toward future generations, contributing in this way to higher long-term saving and income. Therefore wider considerations of fiscal sustainability and intergenerational equity should determine how much pension deficit financing should rely on fiscal contraction. This decision will also be influenced by its effects on long-term saving and output, estimated in the next section. (iv) Iffiscal contraction is used and the non-pension deficit is cut, which taxes should be raised and which expenditure programs should be cut? Answering to this question is beyond the scope of this paper -- it is part of the government's ongoing evaluation of fiscal policy and the relative efficiency of its individual financing instruments and spending programs. 21 This figure comprises preliminary estimates of possible privatization revenue from selling SOEs in the following sectors: power plants (US$ 1.8 b.), financial sector and banking (US$ 0.6 b.), and others including mining, infrastructure, utilities, and transport ($3.2 b.). 21 Issuing external government debt has similar effects to issuing domestic government debt if the domestic financial market participants (including the new AFPs) undo the increase in the country's net external debt by buying external assets for the same amount. This, however, is unlikely. Hence the risk premium paid on Colombia's external debt and the likelihood of a cut in the supply of external financing funds will increase with the level of net external indebtedness.  23 Long-Run Output and Welfare Effects of Colombia's Pension Reform if Financed by Fiscal Contraction How large is the long-run increase in output and welfare when the transition deficit is financed by raising taxes or cutting expenditure? An answer to this question is provided here by applying an overlapping-generations exogenous-growth model parameterized for the Colombian economy. The model simulations are performed for the pre-reform steady-state equilibrium under a PAYG system and the post- reform steady-state equilibrium with a FF pension scheme -- the latter relevant only for the very long run, that is 70 or more years into the future. Three comparative simulations are performed for Colombia, each of them comparing macroeconomic performance and consumer welfare levels of the initial PAYG and the final FF system. Simulation I is for homogeneous consumers which share a common subjective discount rate and pay a low mandatory pension contribution rate of 3.6%. The latter figure is close to the average contribution rate paid by Colombia's total labor force (comprising those paying and those evading payroll taxes) during 1994-2025 (see Annex 3). Simulation 2 distinguishes between consumers with low and high subjective discount rates -- the latter are strongly affected by credit constraints -- but both pay a low contribution rate. Credit constraints are reflected by a non-negativity constraint imposed on consumer assets. Finally, under simulation 3 the two groups of heterogenous consumers pay a high contribution rate. Model results are summarized in table 5.2 and life-cycle profiles for income, consumption, and non-pension assets are depicted in figures 5.1 - 5.3. The results quantify the qualitative predictions of pension reform theory for a tax-financed transition reform deficit, i.e., when the implicit PAYG debt is paid off by transition tax-paying generations. Simulation I for homogeneous consumers reflects that the initial PAYG debt is 62.9% of GDP when the pension contribution rate is 3.6%. The PAYG debt figure is close to the estimate for Colombia's implicit PAYG debt accumulated through 2025 (63.2% of GDP) Eliminating that debt by adopting a FF scheme changes the life profile of individual consumption and income as depicted in figure 5.1. Shifting resources from tax-paying to future cohorts raises long-term saving, capital, and consumption levels. Capital deepening (the capital/output ratio increases by 13 percentage points) leads to a fall in capital productivity and hence in the real interest rate, and to a higher wage rate. The lower interest rate is reflected by a less steep intertemporal life-cycle consumption schedule (Fig. 5. 1). The higher rate of return of the FF schemes -- that is, the difference between FF real interest rate and the PAYG 5 % steady-state GDP growth rate -- explains why the replacement ratio (pension/average wage during last 10 working years) increases by so much under the FF scheme. Output increases by a very modest 2.4% and future consumers enjoy a small welfare gain of 4.2% -- at the expense, obviously, of income and welfare losses suffered by the transition generations that have paid off the implicit PAYG debt. Simulation 2 differentiates between consumers with low and high discount rates. Consumers in the second category are credit-constrained for a significant fraction of their lives, forced by the mandatory pension schemes to save more than what they would voluntarily do. Because of these consumers, there is less capital in this economy, real interest rates are higher, and wages are lower. However, the shift of PAYG to FF -- which benefits long-term consumers because of the resource transfer -- has a stronger relative effect on all variables than under simulation I because credit-constrained consumers are forced to save a large fraction of their resource transfer. Welfare of unconstrained consumers increases by more than welfare of constrained people because the latter save more than what they would do in a voluntary saving scheme. The output levels is now 3.9% higher in the very long run.  24 Finally, simulation 3 reflects the case of a very high average contribution rate, that would be observed if Colombia's pension system reached a labor force coverage of 100%. In that sense these results reflect the upper bound of macro and welfare changes which could be expected from a tax- financed transition in this category of exogenous-growth models. The implicit PAYG debt at 175% of GDP almost triples that observed when average contribution rates are 3.6%. The life-cycle consumption patterns for low-discount-rate and high-discount-rate consumers is shown in figures 5.2 and 5.3, respectively. The aggregate steady-state results point now to large capital deepening, leading to significantly lower interest rates and higher wages. Output increases by 14.0% and welfare is raised by 13.2 - 17.8%. These simulation results point toward a number of important conclusions. Financing of Colombia's pension reform by raising taxes or cutting government current expenditure leads only to modest output and welfare gains that are all in the single-digit percentage range. However, if labor force coverage by the new pension system is significantly extended as a result of introducing FF -- say from the current 29.6% to 70% or more -- the output and welfare gains of tax financing could exceed 10%. The larger is the share of credit-constrained consumers, the more significant are the macroeconomic and welfare changes of tax financing. Finally, the larger is the pure FF component of pension contributions -- that is, the lower is the pure tax component -- the larger will be the incentive and macroeconomic effects of the pension reform. Long-Run Effects of Colombia's Pension Reform on Growth and Equity This section discusses the potential growth gains (through higher saving, a shift of labor toward formal sectors, and the development of capital markets) and the potential equity benefits that Colombia could reap from pension reform. Saving and growth The quantitative simulations of the preceding section show that financing the pension reform through fiscal contraction raises long-term saving and could induce a level effect on long-term output that at most would reach 14%. Beyond this effect it is also possible that overall saving is increased as a result of lower myopia if workers are made more aware of old-age savings by holding individual pension savings accounts. Further effects on output growth -- as opposed to effects limited to the level of output -- could only be reaped if higher saving translated into higher investment does not ultimately drive down the return on capital to the level of the discount rate. That would depend on the nature of Colombia's capital accumulation and growth process, i.e., on the endogeneity of long-term growth. Formal employment and growth Colombia's old pension system was characterized by low labor force coverage and low employment formality. Colombia's formal-sector employment shrunk from 47.5% in 1980 to 39.5% in 1992, a trend shared with the rest of Latin America excepting Chile (Table 5.3). Growing informality in Colombia and the region and declining informality in Chile is probably not unrelated to the fact that during 1980-92 Chile had a FF pension regime and closely related payroll contributions and benefits, while the opposite was true in Colombia and elsewhere.  25 Colombia's new pension system offers the potential benefit of reducing the incentives for payroll tax evasion and hence contributing to larger shares of more productive formal-sector employment and production. However, two features of Colombia's reform make it unlikely that such gains could be reaped as a result of pension reform alone. First, pension contribution rates have been increased substantially, from 0-6.5 % prevalent before the reform to 13.5-14.5% in 1996 and thereafter (Table 5.4). This is troublesome for myopic contributors, those that because of their high subjective discount rates do not value much the pension benefits they will receive in a distant future even though these benefits are closely linked -- through market returns -- to their current contributions. For this group of contributors, the pension reform induces a further disincentive to work in formal sectors, encouraging a larger shift to informality. Hence it is only for less myopic people that the reduction in the pure-tax component of payroll taxes offers a positive incentive to go formal. Second, pension contributions are a relatively minor component of overall payroll contributions for seven social programs that, after the pension and health reforms, range between 39.83 % and 46.33 % (Table 5.4). These are very high levels, not only in absolute terms but, most important, because of their high pure-tax components. Under certain assumptions,22 the pure tax component of payroll contributions could be as high as 13 to 16% for non-myopic contributors -- a powerful incentive for informal employment. This points to the inference that in order to raise the share of formal-sector employment and production, payroll contributions other than pension should also be closely linked to expected benefits at the individual level. Therefore the distributive components of all programs should be financed by general taxation. Only then would the potential contribution of pension reform to formal-sector employment and hence higher average labor productivity growth be realized. Capital market development and growth Colombia's reform combines a partial privatization of pension services with an explicit focus of the government on its redistributive and regulatory functions. Colombia's administrative costs ranged between 5% and 15% of contributions under its old pension system (Ayala 1994). The limited international experience suggests that some administrative costs could fall with privatization while others -- fundamentally marketing expenses -- increase when pension services are provided by a competitive industry of private providers among which contributors can choose.' Net rates of return of pension 22 If 13.5% of pension contributions, 6.0 to 8.0% of health contributions, 5.33% of unemployment insurance contributions, 1.0 to 2.5 % of accident insurance, and 1.0% of training contributions are closely related to expected benefits at the individual levels but the remaining contributions are unrelated to benefits, the pure tax component of total payroll contributions is 13 to 16%. 23 Cost comparisons between different pension systems is difficult because of differences in quality across pension regimes. (For a discussion and cross-country cost comparisons see Valdes-Prieto 1994 and World Bank 1994b). Privatized competitive pension systems like the Chilean scheme show significant marketing costs that obviously are absent under a monopolistic state-run scheme. However, the quality of collection, investment, insurance, and benefit payment services is higher when provided by a competitive industry of private firms.  26 savings, portfolio diversification, and the overall quality of pension services should increase significantly with privatization. The government's specialization on redistribution and regulation/supervision of the pension service industry should also contribute to efficiency gains reflected in better pension services at lower cost. Beyond the provision of pension services, Colombia's FF private pension system endowed with broad portfolio choices by AFPs should spur the development of capital markets in general and the market for long-term securities in particular. If Chile's experience offers any indication, it is that capital- market development takes off with a privatized FF pension system. This involves both a large growth of domestic equity and bond markets and portfolio diversification by investing part of pension savings abroad. As a positive externality of this growth in capital markets, Colombia's legal and regulatory framework for capital markets will require further improvements and the country's integration into world financial markets will increase. Colombia's financial intermediation will improve and its capital cost could fall -- both factors contributing to additional efficiency and growth gains. Equity and poverty Colombia's old pension system -- as most state-run PAYG schemes' -- was inequitable and ineffective in providing support to the old-age poor. Intra-generational distribution of income (from lifetime rich to lifetime poor) through the pension system was regressive in Colombia for two reasons. First, most of Colombia's poor engage in informal activities that pay lower wages than those prevalent in formal sectors (World Bank 1994a), in part because of the existence of a formal-sector PAYG system. Second, income redistribution within the pension system was highly regressive, allowing for large differences in contributions and benefits among different pension plans that benefitted those groups that were more successful in pushing for higher net benefits -- typically not the poorest. Evidence in support of the regressivity of Colombia's old pension system is provided in Table 5.5. All pensioners received positive net transfers from the government (i.e., from younger generations) that, as a share of benefits, were similar across all income groups. However, the absolute value of the transfer received by high-income earners was 7.4 times the value received by minimum-wage earners. By eliminating the pure tax component of pension contributions (at least for non-myopic contributors), the pension reform offers the possibility of attracting low-income groups. A more important instrument, however, is the direct redistribution toward low-income contributors (through partial subsidies of their contributions by the Fondo Nacional de Solidaridad Pensional) and low-capital workers at retirement (through the Minimum Pension State Guarantee). These explicit redistributive programs targeted to the poor, complemented by the elimination of perverse redistribution pervasive under the old system, makes the new system much more equitable and effective in addressing old-age poverty. What can be said about intergenerational distribution? During the reform transition the massive transfer to current pensioners is maintained. Older contributors keep their generous benefits but pay 24 The empirical evidence available for OECD countries shows little if any redistribution from the lifetime rich to the lifetime poor. Indirect evidence for developing countries suggests that intra- generational distribution in conventional public pension schemes is strongly regressive. (For examples see World Bank 1994b, pp. 131-138).  27 larger contributions, and younger contributors see their benefits cut and are required to contribute more. Further intergenerational redistribution will mostly depend on how the transition deficit is financed. As discussed above, deficit financing through fiscal contraction implies income redistribution from transition generations toward future cohorts. Other forms of financing -- by issuing debt, using privatization and oil revenue, or cutting public investment -- leave intergenerational distribution largely unaffected. 6. CONCLUSIONS AND FUTURE POLICY CHALLENGES This study offers a number of conclusions on Colombia's pension reform and its prospects. In addition to summarizing the main findings, this section closes the paper by identifying five challenges faced by the government to ensure that the potential efficiency and equity gains of pension reform are realized. Conclusions Colombia's old pension system was beset by all the ills affecting conventional state-managed pay- as-you-go (PAYG) schemes all over the world. It was characterized by uneven contributions and benefits across different pension sub-systems, large transfers to the first generations of pensioners, weak relations between worker contributions and expected pension benefits, low coverage of Colombia's labor force, inefficient public management, very poor funding of pension liabilities, and rising fiscal transfers to cover pension deficits. Moreover, such systems impose serious efficiency and equity costs. The large transfers to the first generations of pensioners reduce long-term saving and growth. The weak links between payroll contributions and PAYG benefits embedded in a PAYG system with implicit redistribution and large benefit uncertainty contributes to tax evasion and lower growth by shifting employment and production to less productive informal markets. Both the lack of funding and public-sector management precludes the positive contribution to capital-market development and growth made by privately-managed fully- funded (FF) schemes that invest pension savings in long-term financial instruments. State-managed pension funds with non-transparent cost and benefit structures and strongly subject to political and special- interest pressures show high administrative costs relative to their low-quality services reflected in low steady-state benefits to pensioners. Income redistribution in such systems is often perverse and regressive, seldom reaching the old-age poor. However, as a result of Colombia's early stage of demographic and systemic maturation most of these deficiencies and costs were still relatively small in Colombia. The 1993 pension reform precluded an explosion of efficiency costs and systemic inequities which Colombia would have suffered in coming years as a result of demographic and systemic maturity. The pension reform, initiated in June 1994, corrects many of the flaws of the preceding system by combining a reduction in net benefits paid to future pensioners and a gradual substitution and partial privatization of the existing PAYG scheme by a dual PAYG-FF system. Two alternative mandatory pension sub-systems, chosen on an either/or base by contributors, are put into place. The first is a state-managed defined-benefit partially-funded PAYG scheme run by the Instituto de Seguros Sociales (ISS). The second is a privately-managed defined- contribution FF system comprised by private providers of contribution collection and investment services (the Administradoras de Fondos Pensionales or AFPs) and by private providers of pension services and annuity payments (insurance companies).  28 Income redistribution is made transparent by starting explicit programs in support of pension benefits for the poor. Government guarantees are extended to pension savings managed by AFPs whose management and investments are regulated and supervised by a specialized government agency. The institutional and financial complexity of the pension reform transition is large. Colombia's transition requires putting in place various temporary and permanent pension institutions and transfer programs, as well as effective regulatory and supervisory institutions. Explicit repayment of PAYG pension rights accrued to current workers requires a lengthy process of establishing their individual histories of past affiliations. Colombia's reform -- like other pension reforms in Latin America -- combines two very different changes: a reduction in net benefits paid to future pensioners by the existing PAYG system, and a gradual substitution of the existing PAYG scheme by a dual PAYG-FF system. The first reform component reduces pension system deficits and the implicit PAYG debt of the government incurred vis-a- vis current and future pensioners, improving the financial position of public pension institutions. The second reform component entails a regime change that makes explicit the currently implicit PAYG debt as current PAYG pensioners are paid off and past PAYG pension liabilities accrued to current workers are paid to those shifting to a FF scheme. During this extended horizon the making explicit of PAYG debt is reflected by reform transition deficits incurred by the government. Long-run simulations show that the first component of the pension reform -- comprising higher contribution rates for all and higher retirement ages for younger cohorts -- is financially very effective. It reduces the capitalized value of the pension deficits which would have been incurred during 1994-2025 by the equivalent of 37.6% of GDP. To understand the second reform component it is useful to distinguish between pension rights of current (pre-reform) pensioners and those accrued to current workers in lieu of their past affiliation. Continuing government transfers are made available to pay for pensions of current pensioners (as long as the last current pensioner survives) while the shrinking contribution base implies a resource shortfall termed operational deficit. In addition, the government honors pension rights of current workers by issuing pension recognition bonds to currently active workers. Recognition bonds will be issued by the pension institutions were workers were affiliated before shifting to AFPs or [SS. Recognition bonds are debt instruments that make explicit defined-benefit pension rights accrued to workers in their old pension funds (proportional to the number of years of past affiliation) and that mature at the date of retirement, giving rise to recognition bond deficits. Under the base-case simulation, Colombia's pension reform transition deficits are projected to start at 0.88% of GDP in 1994, continue growing to peak at 2.61 % of GDP in 2013, and then decline gradually to 1.95% of GDP in 2025 and converging to zero by, say, 2060. The capitalized value of pension deficits for the total 1994-2060 transition period is large: 83.6% of GDP. Of this figure, the largest part (69.2% of GDP) corresponds to the capitalized value of operational deficits. The remainder (14.4% of GDP) is recognition bond debt. Colombia's pension transition debt of 83.6% of GDP can also be decomposed according to the two components of the reform. The equivalent of 24.4% of GDP would have been observed if only the first component of Colombia's reform had been enacted. The remaining 59.2% of GDP is due to the substitution of the initial PAYG system by a mixed PAYG-FF scheme, i.e., this figure corresponds to the total implicit PAYG debt made explicit by the pension reform. Adding to this figure the remaining  29 steady-state implicit PAYG debt due to the survival of ISS with a core number of affiliates in 2060 and beyond -- estimated at only 4.5% of GDP under the base-case simulation -- implies a total implicit PAYG debt of 63.7% of GDP. Finally, the total sum of Colombia's explicit pension reform and remaining implicit PAYG debt at 88. I% of GDP reflects what would have been the total explicit debt if the PAYG system (ISS) had been phased out entirely by the pension reform, as done in Chile, for instance. How does Colombia's projected net pension debt (net of initial pension assets) compare to other countries? Its net pension debt at 86.5% of GDP is significantly lower than Chile's comparable 126% of GDP. It is also lower than net implicit pension liabilities in most G-7 countries, where GDP ratios range between 44% of GDP (U.S.) and 251% of GDP (Canada). The lower debt ratios in Colombia should not come as a surprise as the other countries are characterized by larger coverage, more mature PAYG schemes, and older populations. The implication for Colombia is that although its current pension system is relatively small (only 29.6% of the labor force contributes to the pension system), the country has incurred in sizable pension liabilities, implying a long period of substantial reform deficits. This raises Colombia's need to define the ways by which it will finance its transition deficits. Colombia -- as any pension reformer -- faces two fundamental choices in deciding how to finance its pension reform deficit. The first alternative is paying off the implicit PAYG debt by a combination of issuing domestic debt, using future net income from oil discoveries, using revenue from SOE privatization, and cutting government capital expenditure. The second option is reducing its non- pension fiscal deficit by raising taxes or reducing current expenditure. The first option is the "purest" form of pension reform. It leaves the government's net total wealth or asset position (comprising both explicit and implicit PAYG liabilities) unchanged, even though the official government deficit records an increase by the amount of the pension deficit. The second option, however, avoids a higher official government deficit although it implies an increase of net total (explicit and implicit) government wealth. The latter option implies an income transfer from current and future transition generations (those hurt by higher taxes or lower government expenditure) toward long- term future generations freed from PAYG transfers (which would have been paid if the old PAYG system were maintained) or higher taxes (which would have been required to finance the shortfall of returns as a result of lower non-pension net assets under the first option). As any other contractionary fiscal policy, the resource transfer toward future generations embedded in the second option raises long-term saving, output, and welfare levels, at the expense of income and welfare losses of transition generations. Simulation results for pension reforms financed by fiscal contraction in representative economies show that saving, output and welfare effects accrue only in the very long term -- say 80 years after the reform start. In addition, output level gains are not very high, typically in the 3 to 10% range. Higher long-term growth is only reaped when the higher saving (and investment) does not peter out in the long- term -- which occurs only when the return on capital is not ultimately driven down to the discount rate. Simulation results for a pension deficit financed by raising taxes or cutting government current expenditure in Colombia also point toward modest output and welfare level gains, that are in the single- digit percentage range. However, if labor force coverage by the new pension system is significantly extended as a result of introducing FF -- say from the current 29.6% to 70% or more -- the output and welfare gains of tax financing could exceed 10% to a maximum of 14% if coverage reaches 100%. The larger is the share of credit-constrained consumers, the more significant are the macroeconomic and welfare changes of tax financing. And the larger is the pure FF component of pension contributions -  30 - that is, the lower is the distributional component -- the larger will be the incentive and macroeconomic effects of Colombia's pension reform. In addition to considering the long-term effects of a fiscal contraction, two other criteria should help in deciding among the two ways of financing transition deficits: the value of public-sector assets available for pension reform financing and the possible financial effects of issuing explicit debt substituted for implicit PAYG debt. While not all resources from oil discoveries and privatization could be used for paying for pension deficits (due to budgetary and political pre-commitments and restrictions), it is useful to compare the upper bounds of potentially available resources with reform financing needs. Adding the total estimated present value of net income accruing to the government from the oil discoveries (21.2% of GDP) to an estimate of the present value of privatization revenue (10.0% of GDP), yields a figure of 31.2% of GDP, that falls substantially below total long-term pension financing requirements (83.6% of GDP). Still, 31.2% of GDP is equivalent to the capitalized value of total pension deficits during the first 18 years of reform (1994-201 1). What effect would be felt in domestic financial markets -- particularly on interest rates -- if the remainder of the pension debt (52.4% of GDP) were entirely financed by issuing domestic debt, hence raising the outstanding domestic central government debt from 6.0% in 1993 to 58.4% of GDP by, say, 2060? It is important to keep in mind that issuing explicit domestic debt for paying off implicit government debt does not have first-round macroeconomic and financial effects if financial-market participants see through this debt swap, i.e., do not suffer from fiscal illusion. If financial market participants, including the new AFPs, realize that the government's net wealth position is unaffected by a debt-financed pension reform that simply increases both the supply of government debt and the demand for government paper (through the investment of AFP resources), interest rates on government debt (and hence domestic interest rates at large) should not change. This seems to have happened in the aftermath of the Chilean reform throughout the 1980s when explicit government debt grew rapidly while PAYG pension debt was repaid, but real interest rates did not change much. However, some degree of fiscal illusion should not be ruled out a priori, justifying a careful initial use of debt financing in Colombia. Pension reform -- substitution of a FF privately-managed pension system for a PAYG state- managed system -- can have significant positive effects on growth and welfare beyond those accruing from higher long-term saving. These gains are reaped through efficiency gains in the two markets most affected by pension reform: labor and capital markets. Colombia's new pension system offers the potential benefit of reducing the incentives for payroll tax evasion and hence contributing to larger shares of more productive formal-sector employment and production. However, two features of Colombia's reform make it unlikely that such gains could be reaped as a result of pension reform alone. First, pension contribution rates have been increased substantially. This is troublesome for myopic contributors, for which higher contributions are equivalent to higher taxes. It is only for less myopic people that the reduction in the pure-tax component of payroll taxes offers a positive incentive to go formal. Second, pension contributions are a relatively minor component of overall payroll contributions that range between 39.83 % and 46.33 % after the pension and health reforms. These are very high levels, not only in absolute terms but, most important, because they support some programs whose benefits are only weakly or not at all related to worker contributions, implying a large pure tax component of payroll contributions.  31 Colombia's reform combines at least partial privatization of pension services with an explicit focus of the government on its redistributive and regulatory functions. The limited international experience suggests that some administrative costs could fall with privatization while others -- fundamentally marketing expenses -- increase wnen pension services are provided by a competitive industry of private providers among which contributors can choose. More systematic international experience suggests that net rates of return of pension savings, portfolio diversification, and the overall quality of pension services increase significantly with privatization. The government's specialization on redistribution and regulation/supervision of the pension service industry should also contribute to efficiency gains reflected in better pension services at lower cost. Beyond the provision of pension services, Colombia's FF private pension system endowed with broad portfolio choices will spur the development of capital markets in general and the market for long- term securities in particular. If Chile's experience offers any indication, it is that capital-market development takes off with a privatized FF pension system. This involves both a large growth of domestic equity and bond markets and portfolio diversification by investing part of pension savings abroad. As a positive externality of this growth in capital markets, Colombia's legal and regulatory framework for capital markets will require further improvements and the country's integration into world financial markets will increase. Colombia's financial intermediation will improve and its capital cost could fall -- both factors contributing to significant efficiency and growth gains. Colombia's old pension system -- as most state-run PAYG schemes -- was inequitable and ineffective in providing support to the old-age poor. Intra-generational distribution of income through the pension system did not conform to objectives of income redistribution or poverty alleviation. It allowed for large differences in contributions and benefits among different pension plans, benefiting those groups that were more successful in pushing for higher net benefits -- typically not the poorest. By eliminating the pure tax component of pension contributions (at least for non-myopic contributors), the pension reform offers the possibility of attracting low-income groups. A more important instrument, however, is the redistribution toward low-income contributors and low-capital workers at retirement. These explicit redistributive programs targeted to the poor, complemented by the elimination of perverse redistribution pervasive under the old system, makes Colombia's new system much fairer and more effective in addressing old-age poverty. Policy challenges From the preceding findings we identify five policy challenges faced by the government of Colombia to ensure that the growth and poverty-alleviation benefits of the pension reform come to fruition. First, the complex institutional changes required by the pension reform should be executed effectively and promptly. This includes determining the solvency of national and regional pension cajas and funds, phasing out insolvent institutions, starting new institutions (the national and regional pension payment programs and the distributive transfer programs), performing calculation and payments of recognition bonds, supporting smooth changes in contributor affiliation from old public funds to the AFPs (and ISS), and ensuring effective regulation and supervision of both ISS and AFPs by the Intendency for Pension Funds. Second, the role played by ISS as an important provider of pension services in Colombia should be clarified and, in the long-term, revised. First, there is a risk that the efficiency and equity costs of  32 a state-managed PAYG system will be preserved in proportion to the size of ISS. Second, considerable financial and economic uncertainties arise from the right of contributors to shift affiliation back and forth between ISS and AFPs. Addressing these problems could be done in two stages. The first step involves maximizing funding of ISS liabilities by ensuring that new affiliates bring their actuarially-fair recognition bonds with them, improving the quality of portfolio investments of ISS reserves, and maximizing financial and political autonomy of ISS from the central government. For the longer term a reform of ISS ensuring full funding of its pension liabilities and partial or complete privatization should be given serious consideration. Third, diversification of investment portfolios of pension funds towards private and international instruments should be gradually but continuously permitted to ensure reaching higher risk-return frontiers for all contributors. Larger international portfolio diversification would allow Colombia's workers to hedge against country-specific shocks and risks. Fourth, the government should decide on how it will finance its pension reform deficit. While this decision could be revised over the years as part of overall fiscal policy decisions, it would be beneficial to identify certain forms of financing from the outset. The use of revenue from oil discoveries and SOE privatization and issuance of explicit government debt would be the first choice if the government wants to de-link its pension reform from a contractionary fiscal policy. Issuing debt could be an effective choice as long as financial markets understand that this involves a simple swap of explicit government debt for implicit PAYG debt and hence do not react by requiring higher risk premia for holding goveri.ment paper. Finally, and most important, pension reform has to be complemented by other reforms of social security and labor market programs that are currently financed by payroll contributions. In order to reduce evasion of payroll taxes and raise the share of formal-sector employment and production, the high pure-tax components of today's very large payroll contributions should be eliminated. This requires identifying alternative sources of general taxation to finance these programs. Once these changes are implemented, Colombia will start reaping the full efficiency and growth gains derived from larger formal- sector employment and production.  33 REFERENCES Aparicio, M. and W. Easterly (1995): Crecimiento Econ6mico: Teorfa. Instituciones y Experiencia Internacional. Banco Mundial - Banco de la Repdblica, Colombia. Arrau, P. (1990). "Social Security Reform: The Capital Accumulation and Intergenerational Distribution Effect," PRE Working Paper # 512, The World Bank, Washington, D.C., December. -------- (1991). "La Reforma Previsional Chilena y su Financiamiento Durante la Transici6n," Colecci6n Estudios CIEPLAN 32 (June): 5-44. -------- (1992). "El Nuevo R6gimen Previsional Chileno." in Fundaci6n Friedrich Ebert de Colombia (FESCOL): op. cit. Arrau, P. and K. Schmidt-Hebbel (1993). "Macroeconomic and Intergenerational Welfare Effects of a Transition from Pay-As-You-Go to Fully-Funded Pension Systems", manuscript, The World Bank, Washington, D.C., June. Arrau, P. and K. Schmidt-Hebbel (1994). "Pension Systems and Reforms: Country Experiences and Research Issues", Revista de Analisis Econ6mico 9 (1): 3-20. Arrau, P., S. Vald6s-Prieto and K. Schmidt-Hebbel (1993). "Privately Managed Pension Systems: Design Issues and the Chilean Experience", manuscript, The World Bank, Washington, D.C., April. Auerbach, A. and L. J. Kotlikoff (1987). Dynamic Fiscal Policy, Cambridge University Press. Ayala Oramas, U. (1994). "La Reforma Pensional Colombiana," presented at the Seminar on Privatization. Regulation and Social Security Reform: Colombian and International Experience, Santaf6 de Bogota, Colombia. May 12. Bosworth, B.P., R. Dornbusch and R, Laban (editors) (1994). The Chilean Economy: Policy Lessons and Challenges. Brookings, Washington, D.C. Cardenas, M.E. (1992). "Seguridad social y regimen pensional: Balance de argumentos para una reforma," in Fundaci6n Friedrich Ebert de Colombia (FESCOL): op.. cit. Cifuentes, R.S. and S. Vald6s-Prieto (1994). "Transitions from PAYG to Funding in the Case of Credit Constraints", paper presented at the Conference on Pensions: Funding. Privatization and Macroeconomic Policy, Catholic University of Chile, January. Contralorfa General de la Repdblica de Colombia (1992). "La Reforma de la Seguridad Social en Colombia: Una Aventura Econ6mica", Serie Estudios Ocasionales No. 3, Santaft de Bogota. Corsetti, G. and K. Schmidt-Hebbel (1995). "Pension Reform and Growth", in S. Valdes-Prieto (ed.): Pensions: Privatization. Fundinz. and Macroeconomic Policy, Cambridge University Press, forthcoming.  34 Diamond, P. and S. Valdes-Prieto (1994). "Social Security Reforms", in Bosworth et al.: op. cit. Fern-indez Riva, J. (1992). "La Regresividad del Sistema Pensional", Carta Financiera (October): 29-39, Bogota. Fundaci6n Friedrich Ebert de Colombia (FESCOL) (1992). Regfmenes Pensionales. Santaf6 de Bogota, Colombia. Garcfa Garcfa, J. (1995). "El Crecimineto Econ6mico Colombiano," in M. Aparicio and W. Easterly (eds.): op.cit. Helmsdorff, L. (1994). "Simulaciones de Efectos Financieros de la Reforma de Pensiones en Colombia", manuscript, Ministerio de Hacienda de Colombia - Banco Mundial, May, Bogota. Iglesias, A. and R. Acuna (1991). Sistema de Pensiones en America Latina. Chile: Experiencia con un R6gimen de Capitalizaci6n 1981-1991. Regional Project on Financial Policies ECLAC-UNDP, Santiago, Chile. Instituto de Seguros Sociales (1994). Estadfsticas 1993: Resumen Ejecutivo. Santaf6 de Bogota, February. Llano, J.R. (1992). "La reforma pensional propuesta por el Gobierno," Debates de Coyuntura Econ6mica, #26, FESCOL. L6pez, C. (1992). "Elementos para un debate sobre la reforma a la seguridad social en Colombia", in Fundaci6n Friedrich Ebert de Colombia (FESCOL): op.cit. L6pez, H. (1992). "Ciclo de vida, seguridad social y atenci6n a la tercera edad en Colombia", in Fundaci6n Friedrich Ebert de Colombia (FESCOL): op.cit. Lora, E., H. Zuleta and L. Helmsdorff (1993). "Viabilidad Econ6mica y Financiera de un Sistema Privado de Pensiones", Coyuntura Econ6mica 22 (1), Fedesarrollo. Marshall, J. and K. Schmidt-Hebbel (1994). "Fiscal Adjustment and Successful Performance in Chile", in W. Easterly, J. Rodrfguez, and K. Schmidt-Hebbel (editors): Public Sector Deficits and Macroeconomic Performance, Oxford University Press, forthcoming. Mc Greevey, W. (1990). "Social Security in Latin America: Issues and Options for the World Bank", World Bank Discussion Pager 110, Washington, D.C. Mesa-Lago, C. (1978). Social Security in Latin America: Pressure Groups. Stratification and Inequality. Pittsburgh University Press. -------- (1993). "Pension Reform in Latin America: Importance and Evaluation of Privatization Approaches", paper presented at the Seminar on the Economic and Social Impact of Privatization in Latin America, Institute of the Americas, La Jolla, California, January. Ministerio de Trabajo y Seguridad Social (1992). Exposici6n de Motivos del Proyecto de Lay Por el cual  35 se crea el Sistema de Ahorro Pensional y se dictan otras disposiciones sobre seguridad social, Santafe de Bogota, D.C., September. Ministerio de Trabajo y Seguridad Social (1993a). Sistema General de Pensiones: Anexo Estadfstico y de Proyecciones. Santafe de Bogota, D.C., July. Ministerio de Trabajo y Seguridad Social (1993b). Ley de Seguridad Social (Ley 100 de 1993). Santafe de BogotA, D.C., December. Ministerio de Trabajo y Seguridad Social (1993c). La Previsi6n Social para los Empleados del Sector Pdblico. SantafA de Bogota, D.C., December. Montenegro, A. (1995): "El Crecimiento Econ6mico Colombiano", in M. Aparicio and W. Easterly (eds.): op.cit.. Ocampo, J.A. (1992). "La Propuesta Gubernamental de Reforma al Regimen Pensional: Analisis y Alternativas", Debates de Coyuntura Econ6mica 26: 28-48. Posada, C.E. and A. Gaviria (1995): "El Crecimiento Econ6mico y la Distribuci6n del Ingreso: El Caso Colombiano Posterior al 1950", in M. Aparicio and W. Easterly (eds.): op.cit.. Ramfrez, H. (1992). "La Reforma Pensional Propuesta por el Gobierno", Debates de Coyuntura Econ6mica No. 26: 7-12. Schmidt-Hebbel, K. (1994). "Pension Reform Transitions from State Pay-As-You-Go to Privately- Managed Fully-Funded Systems", presented at the International Seminar on Privatization in Colombia, Santaf6 de Bogota, May 11-12. Sub-Direcci6n de Desarrollo Social (1994). "Cuadro Resumen de Costos de las Pensiones en 1993", manuscript, Bogota. Uribe, J.D. (1995): "Inflaci6n y Crecimiento Econ6mico en Colombia: 1951-1992", in M. Aparicio and W. Easterly (eds.): oprciL. Uthoff, A. (1994). "Pension Systems Reform in Latin America: What is Difficult in a Transition to an Individual Capitalization Scheme?", Revista de Analisis Econ6mico 9 (1): 211-35. Valdes-Prieto, S. (1994). "Administrative Charges in Pensions: Chile, the U.S., Malaysia, and Zambia", Policy Research Working Paper Series, The World Bank. Valdes-Prieto, S. and R. Cifuentes (1993). "Credit Constraints and Pensions", manuscrip, Catholic University of Chile, December. Van den Noord, P. and Richard Herd (1993). "Pension Liabilities in the Seven Major Economies," OECD, Working Papers No. 142. World Bank (1987). Colombia: Social Security Review, manuscript, Washington, D.C.  36 World Bank (1994a). Colombia: Colombia Poverty Assessment Report, manuscript, Washington, D.C. World Bank (1994b). Avertinz The Old Age Crisis: Policies to Protect the Old and Promote Growth. Oxford University Press. Zuleta, H. (1992). "El Regimen Pensional de Colombia: La Necesidad de un Cambio Radical", in Fundaci6n Friedrich Ebert de Colombia (FESCOL): op.cit.  37 ANNEX I Model Assumptions for Simulating the Fiscal Effects of Colombia's Pension Reform The following model assumptions are shared by all simulations: (1) The general mortality tables (by ages and gender) are based on 1980-1989 ISS experience. (2) Age-wage profiles (by gender) are based on 1980-1989 ISS experience. (3) The simulations are carried out for the three pension system components: ISS, SPP, and AFPs. (4) The simulation horizon is 1994-2025. (5) All financial variables are in constant-price (1994) pesos. (6) The labor force grows at 2% per year. (7) The calculation of recognition bonds (reflecting past and future contributions of workers affiliated to ISS or SPP when affiliating with AFPs and paid at worker retirement dates) is based on legal entitlement and an annual real interest rate of 4%. (8) In all reform simulations, 80% of new labor market entrants affiliate with AFPs and 20% with the ISS. (9) Initial (1994) reserves of ISS (Source: ISS) imply an interest income of 0.15% of GDP in 1994. Initial reserves of SPP are assumed to be zero. Subsequent stocks of outstanding reserves (or debts) are determined by net revenue/losses of each institution in each period. The following assumptions are used for different sub-sets of simulations: (10) Pension system contribution rates under the no-reform simulation are 6.5% for ISS and zero for SPP. For the partial-reform and actual-reform simulations, contribution rates are 8% in 1994, 9% in 1995. 10% in 1996 and thereafter for ISS, SPP and the AFPs. (11) Pension system coverage (i.e., the share of economically-active population affiliated to the three segments) is assumed to be constant through time at the current (1993) estimate of 29.6% in the first (no-reform) simulation. Introducing a fully-funded pension component reduces the pure tax component of pension contributions and hence provides to formerly unaffiliated workers an incentive to join AFPs. Therefore a gradual increase of pension system coverage can be anticipated as a result of the pension reform; hence total coverage (by all three segments) of economically-active population is assumed to grow gradually from 29.6% in 1994 to attain 46.6% in 2025. In order to isolate the effects of coverage extension from all other (more direct) effects of pension reform, three counter-factual scenarios are considered: two non-reform scenarios (with constant and increasing coverage) and a partial reform scenario (with increasing coverage). Under the increasing-coverage no-reform and increasing-coverage partial-reform scenarios, AFPs do not exist and hence higher coverage implies rising ISS affiliation -- obviously an unrealistic assumption that helps only to isolate the impact of increased coverage. SPP affiliates grow at 0. 1 % per year in all no-reform and partial-reform simulations. (12) The speed of shift of affiliation from public pension funds to AFPs varies between a base case, a low speed, and a high speed case as follows (annual percentage of stock of affiliates):  38 Women younger than 35 Women 35 or above and men younger than 40 and men 40 or above Base case From ISS 20% 5% From SPP 2% 0 Low speed case From ISS 12% 2% From SPP 0% 0 High speed case From ISS 60% 15% From SPP 30% 5% In all reform simulations, new labor market affiliates to ISS shift affiliation (at a rate specified below) only after three years of affiliation to the ISS. There are no shifts from ISS to AFPs of people 5 years or less from retirement. (13) The relevant annual macroeconomic variables vary between three sets of alternatives: Real Interest Rate Real GDP Growth Rate Real Wage Growth Rate (r) (g) (w) Base Case 5% 5% 3% Low Growth Rates 5% 3.5% 1.5% High Interest 8% 6.5% 4.5% and Growth Rates All assumptions satisfy the steady-state growth condition that the difference between real GDP growth and real wage growth (the Harrod-neutral rate of technical progress) is equal to labor force growth.  39 ANNEX 2 Macroeconomic and Welfare Effects of Pension Reform: How Do They Arise? A growing analytical literature analyzes the macroeconomic effects of pension systems and reforms that substitute FF for PAYG pension schemes. (See Arrau and Schmidt-Hebbel 1993, 1994 and Corsetti and Schmidt-Hebbel 1995 for surveys of this literature). What can be inferred from these results for real-world pension system reforms? First, it matters how the transition deficit is financed. The straightforward form is by issuing new government debt, i.e., to swap the old implicit PAYG debt for new explicit government debt. By avoiding an inter-generational transfer paid by tax-paying cohorts, debt financing has only a limited effect on national saving, capital stock and the inter-generational distribution of welfare, arising only from possible second-order efficiency changes caused by the reform. On the other hand, tax-financing of the deficit is equivalent to combining the pension reform with a contractioriary fiscal policy. A completely tax-financed transition reverts the initial PAYG transfer from workers to pensioners -- associated to the start of the initial PAYG scheme -- by fully paying-off the implicit PAYG debt. This hurts tax-paying transition generations and benefits non-taxed post-transition generations. Tax-financing of the transition - - as any restrictive fiscal policy which pays off government debt through taxes and hence shifts resources from current to future generations -- encourages higher saving and capital formation, therefore raising future per capita income and wage levels. These first-order effects on saving and capital formation are added to potential second-order effects of the pension reform due to net efficiency changes. Second, a PAYG-FF reform can have formal labor-market and efficiency conseguences. Real- world PAYG payroll taxation is distortionary and so is general taxation (falling on income, consumption or any other tax base). Hence a shift from PAYG to FF reduces labor market distortions at the cost of inducing other tax-related distortions which are either permanent (if the transition deficit is debt-financed) or temporary (if it is tax-financed). If general taxation is at the margin less distortionary than payro!l taxation, a pension reform brings positive net efficiency gains and hence raises the economy's Pareto efficiency. Since the pure-tax component of PAYG borne by formal sectors (where payroll taxes are paid) disappears (at least for non-myopic workers) in a FF scheme, a pension reform encourages shifting employment and production from informal to formal markets sectors. Third, the effect of changes in national saving on investment and output depend on the economyvs financial openness. In an economy completely integrated into world financial markets, a reform-induced increase in national saving (due to tax-financing of the transition deficit) raises foreign asset holdings without changing the capital stock and output levels. Most countries are at best financially semi-open and hence a rise in national saving raises to a much larger extent domestic capital and output levels than the holdings of foreign assets. In an economy large enough to affect international interest rates, a tax- financed PAYG-FF reform reduces international interest rates, leading to a terms-of-trade change whose sign depends on the country's net creditor or debtor position. Fourth, real-world caring for parents and children weakens the potential effects of substituting PAYG by FF. However, to the extent that voluntary inter-generational transfers are not generalized throughout the population or that people do not offset one-to-one the PAYG transfers to the old by higher bequests or lower support of the old, a pension reform still has some of the saving effects when it is tax- financed.  40 Fifth, mandatory pension schemes raise saving of all those who are required to save more under the pension scheme than what they would voluntarily save. This raises the saving effects of a tax- financed transition. Sixth, myopia limits the ability of relating adequately current pension contributions to old-age consumption. The larger is the share of myopic people unable to distinguish between PAYG and FF contributions, the smaller are the effects of a pension reform on labor markets and overall Pareto efficiency. Awareness of future consumption needs could be raised -- i.e. myopia could be reduced - - by a tully-funded scheme with individual pension accounts, causing people to save (including mandatory pension saving) a larger proportion of income than before. Seventh, the larger is the group of credit-constrained people and the more stringent those constraints are, the larger are the saving and output effects of a tax-financed reform. Eighth, a high old-age dependency ratio -- reflecting an old population structure and/or a low retirement age -- requires a high contribution rate. Therefore the higher is old-age dependency, the larger are the effects of a pension reform on both saving (when the transition deficit is tax-financed) and on labor market and overall Pareto efficiency. Finally, a fully-funded scheme creates a supply of long-term savings, promoting the development of long-term capital markets. This potential externality, which hinges on the existence of capital market imperfections, improves financial intermediation and therefore raises capital productivity.  41 ANNEX 3 An Overlapping-Generations Model for Simulating Steady-State Macroeconomic and Welfare Effects of a Pension Reform The model is a one-sector, two-factor (exogenous labor and endogenous capital) optimal- consumption life-cycle specification for the stationary equilibrium of 55 overlapping cohorts for a closed economy. It allows for credit constraints and heterogenous consumers differentiated by subjective discount rates, following Cifuentes and Valdes-Prieto (1994), an extension of the homogeneous-consumer model by Arrau (1990, 1991), which in turn is based on Auerbach and Kotlikoff's model for the U.S. economy. The simulations performed here were performed on the Cifuentes-Valdes-Prieto (1994) program for this type of exercises. The following are the main assumptions on model parameter values for the Colombian economy. Assumptions shared by all simulations: Labor force growth rate: 2% Labor productivity or average real wage growth: 3% Intertemporal elasticity of consumption substitution: 1.0 Non-pension domestic public debt/GDP: 15% Government consumption/GDP: 10.7% Taxes are income-based Cobb-Douglas production function with Harrod-neutral technical progress and labor share: 65% Capital depreciation rate: 3.5% Active (working) life: 40 years; passive (retirement) life: 15 years Social discount rate (for welfare comparison): 6.5%. Assumptions specific to Simulation I (homogeneous consumers) and 2-3 (heterogeneous consumers): Simulation 1 Simulations 2-3 Consumers: Share of consumers with low discount rate 100% 50% Subjective discount rate of constrained (unconstrained) consumers 3% 3% (10%) Assumptions specific to Simulations 1-2 (with low contribution rate) and 3 (with high contribution rate): Simulations 1-2 Simulation 3 Pension system contribution rate: 3.6% 10.0% The first figure is the product of 36% -- a figure in-between the 1994 rate of pension system coverage (29.6%) and the projected 2025 rate of coverage (46.6%) -- and the new pension system contribution rate in Colombia (10%). That is, in order to take into account the small share of Colombia's pension system coverage in a model where all workers contribute, we assume for simulations 1-2 that each worker contributes a fraction equal to the share of total pension contributions in the total (unobserved) wage bill. However, for simulation 3 we assume that eventually pension system coverage will reach 100%.  TABLE 2.1 COLOMBIA: PENSION SYSTEM INSTITUTIONS BEFORE AND AFTER TME REFORM Pro-Reform (December CAJANAL National Pension Reeional Pension Special Government ISS Financial-Sector and 1993) Funds and Cajas (55) Funds and Caias (991) Employees and SOE Employer-Based Government sponsored, Pension Funds and Cajas Gov. sponsored, Retirement Plans mandatory affiliation, Government sponsored, Sponsored by mandatory affiliation, UF, DB; 0.170 m. mandatory affiliation, departmental and Government and SOE- PF, DB; 3.425 m. Mandatory and voluntary central-government UF-PF, DB; 0.340 m. municipal govs., sponsored, mandatory or private sector affiliation, main and contributors specialized gov. and SOE mandatory affiliation, voluntary affiliation, contributors complementary retire- contributors UF-PF, DB; 0.291 m. main or complementary ment saving, PF-FF, regional gov. retirement savjng, PF- DB-DC; 0.100 m. contsibutors FF, DB; 0.300 m. public privatc and public sector sector contributors contributors Reform Transition Fo ndo Nacional Slinimum Pension Fondo Nacional Fondos Solvent National, ISS AFPs Financial-Sector and | 1994-? de Solidaridad State Guarantee de Pensiones Territoriales Resional, Special Employer-Based Pensional Publicas de Pensiones Government, and Gov. sponsored, Privately sponsored, Retirement Plans Gov. sponsorod Publicas SOE Pension Funds mandatory saving, mandatory saving, Gov, sponsored transfer program, Gov. sponsored and Caias voluntary affiliation, voluntary affiliation, Voluntary transfer program, subsidy to retire- transfer program, Reg. gov. PF, DB; private and FF, DC; private affiliation, subsidy to ment capital to DB pension sponsored transfer Gov., reg. gov. and public-sector and public-sector complementary contributions by poor private and payments programs, DB SOE sponsored, contributors contributors retirement saving; poor and special public sector to pre-reform pension payments to mandatory saving, PF-FF, DB-DC;_| groups of private- contributors/ central gov. pre-reform reg. voluntary affiliation, private and publc sector contributors pensioners pensioners gov. pensioners FF, DB; gov., reg. sector contributors gov. and SOE contributors Reform Steady Fondo Nacional Minimum Pension 3 Public-Sector Pension ISS AFPs Financial Sector State de Solidaridad State Guarantee Funds and Employer- (2060) Pensional Based Retirement IGov or SOE sponsored, mandatory Plans saving, voluntary affiliation, PF- |_____________ _ | |FF, DB; teachers, oil workers, armed forces Notes: Arrows indicate possibic reaffiliations by contributors. UF-PF-FF; unfunded - partially funded -- fully funded; DB-DC: defined benefits - defined contributions; SOE: state-owned enterprises.  43 TABLE 2.2 COLOMBIA: POPULATION, EMPLOYMENT AND PENSION SYSTEM COVERAGE, 1990-1993 (thousands unless indicated otherwise) 1990 1991 1992 1993 1 . POPULATION (1) 32,300 32,841 33,391 33,951 1.1 Young (0-19) 14,871 15,006 1.2 Active (20-59) 16,412 16,783 1.3 Old (60+) 2,109 2,162 2. LABOR FORCE (2) 13,692 13,949 13,917 14,275 3. EMPLOYMENT (2) 12,598 12,834 12,945 13,300 4. MANDATORY PENSION SYSTEM (*) Contributors 4,225.0 Pensioners 532.0 Pensioners/Contributors 12.6% Contributors/Labor Force 29.6% Contributors/Employment 31.8% Pensioners/Old Population 24.6% 4.1 SPP (3) Contributors 800.0 Pensioners 266.7 4.1.1 Cajanal Contributors Pensioners (4) 91.9 4.1.2 National Pension Fund and Cajas (55 in 1993) Contributors Pensioners (4) 105.9 4.1.3 Regional Pension Funds and Cajas (5) (991 in 1990) Contributors 291.1 Pensioners 48.7 68.9 4.2 ISS (6) Contributors 2,724.3 2,876.9 3,167.1 3,425.0 Pensioners 207.7 221.9 244.3 265.3 Note: (*) Excludes the three public pension funds exempted from the pension reform. Sources: (1) DANE. (2) Instituto de Seguros Sociales. (3) Ministry of Finance estimates. (4) Sub-direccion de Desarrollo Social: Cuadro-Resumen de Costos de les Pensiones en 1993, 1994. (5) For 1990: 1989-90 National Pension Census, summarized in Ministerio de Trabajo y Seguridad Social (1993c). For 1993: calculated residually. (6) ISS (1994).  44 TABLE 2.3 COLOMBIA, LATIN AMERICA AND THE OECD: PUBLIC PENSION SCHEME COVERAGE (Percent) Contributors/ Pensioners/ Pensioners/ Labor Force Persons over 60 Contributors Colombia (1993) 29.6 24.6 12.6 Latin America and 38.3 30.8 21.0 the Caribbean (around 1990) OECD (around 1990) 93.9 84.1 39.2 Sources: Colombia (1993): Table 2.2. Colombia ISS (1993) and other regions: World Bank (1994b), Table A.4.  45 TABLE 2.4 COLOMBIA AND THE WORLD: PUBLIC PENSION COVERAGE AND SPENDING (Percent) Contributors/ Pension Pension Spending/ Labor Force Spending/GDP Government Expenditures Colombia (1993) 29.6 2.3 11.1 Country Averages by Income Groups (for years between 1985 and 1992) Low 10.2 (19) 0.7 (32) 3.9 (18) Lower-Middle 27.9 (16) 2.9 (29) 10.1 (16) Upper-Middle 50.7 (7) 6.7 (20) 23.8 (13) High 95.8 (12) 8.2 (23) 23.1 (20) Note: Country averages are unweighted averages. Number in parenthesis indicates number of countries for which data are available. Sources: Colombia: Table 2.2 and 2.6. Country averages: World Bank (1994b), Table 4. 1.  46 TABLE 2.5 COLOMBIA: CONTRIBUTION RATES AND BENEFITS OF OLD PENSION SYSTEM ISS SPP Pension System Pension System for for Private Sector Public Sector Workers Workers 1. CONTRIBUTION RATES (2/3 paid by employer, 6.5% Generally Zero 1/3 paid by employee) 2. RETIREMENT BENEFITS 2.1 Retirement Ages 55 for women, 60 for Generally, 50 for women, 55 for men men (minimum 10 (minimum 20 years of contributions); years of many special regimes contributions) 2.2 Defined Benefit Minimum [45% +3% Generally, 75% of average nominal wage Formula (t-10), 90%] of the before retirement; many special regimes average nominal wage during the two years ________________________ preceding retirement 2.3 Minimum Pension 100% of minimum wage 2.4 Average ISS Pension 125% of minimum wage 2.5 Pension Dispersion 100%-6,000% of minimum wage Sources: Ayala (1994) and Ministerio de Trabajo y Seguridad Social.  TABLE 2.6 COLOMBIA: PENSION SYSTEM CONTRIBUTIONS, BENEFITS, SURPLUSES, AND RESERVES, 1992-1994 (Col. $ billion, unless defined otherwise) 1992 1993 1994 (4) Contrbs Benefits Surplus Reserves Contrbs | Bnefits | Surplus | Reserves Contrbs | Benefits | Surplus Reserv A. Sistema Pensional Puiblico n.a. 592.2 n.a. n.a. % of GDP 1.42 0 1.33 -1.33 n.s. Cajanal (I) n.a. 175.7 n.a. 0 59 National Pension Funds and Cajas (I) n.a. 284.8 n.a. 30.0 991 Regional Pension (2) Funds and Cajas n.a. 131.7 n.a. n.a. B. ISS (3) 248.6 255.0 -6.4 516.4 406.5 349.6 56.9 650.2 % of GDP 0.75 0.77 0 1.6 0.97 0.84 0.1 1.6 1.06 0.97 0.09 n.a. Sources: (I) For 1993: Sub-Direcci6n de Desarrollo Social: "Cuadro-Resunien de Costos de las Pensiones en 1993." 1994. (2) Estinmted proportional to Cajanal pension benefits as the product of average Cajanal pension and the number of Cajanal pensioners. (3) 1992-1993: Instituto de Seguros Sociales (1993)). 1994: the author's counter-factual simulations for no-reform case. (4) For 1994: the author's counter-factual simulations for no-reform case. Note: Current-price GDP lcvels are S33,064 for 1992 and $41,724 for 1993. n.a. not available.  48 TABLE 2.7 COLOMBIA: CONTRIBUTION RATES AND BENEFITS OF NEW PENSION SYSTEM ISS AFPS Government-Managed Defined - Privately-Managed Defined- Benefit Pension System Contribution Pension System 1. CONTRIBUTION 13.5% 13.5 % * RATES (11.5% in 1994, 12.5% in 1995) (11.5% in 1994, 12.5% in 1995) (75% employer, plus I % of (wage base exceeding plus I % surtax of (wage base 25% employee) 4 minimum wages) exceeding 4 minimum wages) 2. RETIREMENT BENEFITS 2.1 Retirement Ages 55 and 60 for women and men Minimum age of [60 years for older than 35 and 40 years, women and 62 years for men, age respectively, or for whom at which pension savings finance a contribution periods exceed 15 pension annuity larger than 110% years in 1994 of minimum wage] 57 and 62 years for younger women and men, respectively (min. 1,000 weeks of l _____________________ contributions) 2.2 Pension Benefits Old-system DB formula for Based on returns on contributions women and men older than 35 and 40 years. DB formula for younger: {min [65% + 2% for each 50 weeks of contribution between 1,000 and 1,200 weeks + 3 % for each 50 weeks of contribution between 1,200 and 1,400 weeks, 90%] of average nomimal wage last 2 years (last I yr.) before retirement of private (public) sector workers} * Sum of 10% contribution for retirement savings and 3.5% for fund administration expenses and invalidity and survivor insurance premia. Source: Ministerio de Trabajo y Seguridad Social (1993b).  49 TABLE 2.8 COLOMBIA AND THE WORLD: FEATURES OF MANDATORY PENSION SYSTEMS Normal Retirement Age Covered Years Payroll Tax Required for Rate for Women Men Full Pension Pensions (%) Colombia (New System): 1994-1995, All 11.5-13.5 1996- , All 13.5-14.5 1994-2013, ISS 55 60 10-20 2014- , ISS 57 62 20 1994- , AFPs* 60 62 20 Argentina 55 60 15 11.0 Chile 60 65 20 13.0 Peru 55 60 15 13.3 Latin America and the Caribbean 58.7 60.8 13.9 10.5 Sub-Saharan Africa 56.0 56.2 13.8 9.1 Asia 55.6 56.5 10.4 13.0 Middle-East and North Africa 57.8 60.4 13.1 10.6 Eastern Europe and the Former Soviet Union 55.3 60.3 25.0 25.5 OECD 62.9 64.4 18.3 16.3 * For those qualifying for minimum pension. Note: Regional averages are weighted averages. Sources: Colombia: Ministerio de Trabajo y Seguridad Social. Other countries and regions: World Bank (1994b), Table A.7.  50 TABLE 3.1 FISCAL EFFECTS OF COLOMBIA'S PENSION REFORM: MAIN FEATURES OF EIGHT ACTUARIAL SIMULATIONS Simulation 1: Constant Coverage. No Reform * Constant pension system coverage (affiliation to mandatory pension system at 1994 level of economically-active population) * No pension reform * Base-case economic variables (annual real interest rate r = 5 %, annual real GDP growth rate g = 5%, annual real wage growth rate w = 3%) Simulation 2: Increasing Coverage. No Reform * Increasing pension system coverage (affiliation to mandatory pension system grows from 29.6% of economically-active population in 1994 to 46.6 % in 2025) * No pension reform * Base-case economic variables Simulation 3: Increasingz Coveragze. Partial Reform * Increasing pension system coverage