7| c .; LA .L- I 2 L-. LELQ L - L; LL I _ L r H~~~~~~~~~~~~~~~L., LO L, c L;~ L L_ L- LE I L XL- dLI L LES 2ii-6[Dealing DcIq with Public Risk in Private Ilnfrastructure Edited by Timothy Irwin Michael Klein Guillermo E. Perry Mateen Thobani |~~~~~~~~~~~~~~~~~~~~~~~~~~~jl ,l'l 1, jjj!1 ~~~~~~~~~~~~~~~~~~~~~~~~~~I_I . . ; .i ,. . _ ...... ..... _ _ . ,~~~~~~~lii WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES Viewpoints DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Edited by Timothy Irwin Michael Klein Guillermo E. Perry Mateen Thobani The World Bank Washington, D. C Copyright 1997 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 December 1997 This publication is part of the World Bank Latin American and Caribbean Studies series. Although these publications do not represent World Bank policy, they are intended to be thought-provoking and worthy of dis- cussion, and they are designed to open a dialogue to explore creative solutions to pressing problems. Comments on this paper are welcome and will be published on the LAC Home Page, which is part of the World Bank's site on the World Wide Web. Please send comments via e-mail to laffairs@worldbank.org or via post to LAC External Affairs, The World Bank, 1818 H Street, N.W, Washington, D.C. 20433, U.S.A. The findings, interpretations, and condusions 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, denominations, 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 encour- ages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. Timothy Irwin is an economist in the Private Participation in Infrastructure (PPI) group of the World Bank. Michael Klein is chief economist of Shell International in London and was formerly the manager of the World Bankls Private Participation in Infrastructure (PPI) group. Guillermo E. Perry is the chief economist of the Latin America and Caribbean (LAC) region of the World Bank and Director for LAC of the Poverty Reduction and Economic Management Network (PREM). Mateen Thobani is principal economist in the Poverty Reduction and Economic Management Unit (PREM) in the Latin America and Caribbean region of the World Bank. Library of Congress Cataloging-in-Publication Data Dealing with public risk in private infrastructure / edited by Timothy Irwin ... [et al.l. p. cm. - (World Bank Latin American and Caribbean studies) "Viewpoints." ISBN 0-8213-4030-1 1. Infrastructure (Economics) - Developing countries - Finance - Congresses. 2. Investments, Foreign - Developing countries - Congresses. 3. Risk management - Developing countries - Congresses. 1. Perry, Guillermo, 1945- . 11. Series. HC59.72.C3M37 1997 336.3'6'091724-dc2l 97-30555 CIP Contents Preface vii Contributors ix 1. Dealing with Public Risk in Private Infrastructure: An Overview 1 Timothy Irwin, Michael Klein, Guillermo E. Perry, and Mateen Thobani The Growth of Private Investment and Government Guarantees 3 Policies That Reduce Risk 6 Principles of Risk Allocation 8 Some Guidelines for the Allocation of Certain Infrastructure Risks 10 Measuring and Budgeting for Risk 13 Conclusion 1 7 Notes 1 8 References 1 8 2. Government Support to Private Infrastructure Projects in Emerging Markets 21 Mansoor Dailami and Michael Klein The Growth of Private Investment in Infrastructure 22 Why Infrastructure Is Different 23 Providing Financial Support to Attract Private Investors 28 Reforming Policy to Attract Investors 31 Conclusion 35 Notes 36 References 36 Annex 38 Comments on "Government Support to Private Infrastructure Projects in Emerging Markets" 43 3. Covering Political and Regulatory Risks: Issues and Options for Private Infrastructure Arrangements 45 Warrick Smith Characterizing and Evaluating Risks 46 Self-Help Strategies 54 Intergovernmental Commitments 60 iii DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Political Risk Insurance 64 Third-Party Guarantees 70 Strategies for Covering Particular Political and Regulatory Risks 74 Notes 77 References 78 Annex 81 Comments on "Covering Political and Regulatory Risks: Issues and Options for Private Infrastructure Arrangements" 86 4. Infrastructure Franchising and Government Guarantees 89 Eduardo Engel, Ronald Fischer, andAlexander Galetovic Government Financing of Private Infrastructure 92 Principles Governing the Design of Franchising Schemes 93 Fixed Term Contracts 98 Least Present Value of Revenues Auctions 99 Conclusion 102 Notes 103 References 104 Comments on "Infrastructure Franchising and Government Guarantees" 106 5. Managing Exchange Rate- and Interest Rate-Related Project Exposure: Are Guarantees Worth the Risk? 109 Ignacio Mas Types of Exchange Rate- and Interest Rate-Related Risks 110 Mechanisms through Which Government Shares Risk 115 Arguments in Support of Government Intermediation of Risks 118 Risk Allocation and Contractual Issues 121 Alternative Policy Options 123 Are Macro Guarantees Consistent with Liberalization? 125 When Are Macro Guarantees Appropriate? 126 Conclusion 126 Notes 127 References 127 Comments on "Are Guarantees Worth the Risk?" 129 6. The Management of Contingent Liabilities: A Risk Management Framework for National Governments 131 Christopher M. Lewis and Ashoka Mody An Integrated Enterprise Risk Management Framework 132 Using Enterprise Risk Management to Manage Government's Contingent Liabilities 135 Identifying and Quantifying the Risks 135 Budgeting for Expected Costs 142 Risk Preferences and Reserve Policy 144 Complementary Measures for Reducing Risk 148 Conclusion 151 Notes 152 References 153 Comments on " The Management of Contingent Liabilities" 154 iv CONTENTS Tables Table 1.1 Statement of contingent liabilities summary table 14 Table 2.1 Net long-term resource flows to developing countries 22 Table 2.2 Private cross-border financial flows to infrastructure 23 Table 2.3 Types of sovereign or supranational support for private infrastructure projects 26-27 Table 2.4 Types of government guarantees in private infrastructure projects 29 Table 2.5 Patterns of sovereign or supranational support for private infrastructure projects 30 Table 2.6 Credit ratings and signed project finance deals, 1996 32 Table A2.1 Signed project finance deals, by country, 1996 38 Table A2.2 Top ten emerging markets for project finance deals, 1996 38 Table A2.3 Top ten emerging markets, 1995-96 38 Table A2.4 Privatization transactions in selected emerging markets, 1991-95 39 Table A2.5 Sovereign credit ratings, country risk assessment, and sovereign defaults in selected emerging markets 40 Table A2.6 Capital market innovations, 1991-96 41-42 Table 3.1 New U.S. Overseas Project Insurance Corporation insurance in power and telecommunications, 1996 65 Table 3.2 Countries in which investors require insurance 66 Table 3.3 Investors' rankings of types of risk 66 Table 3.4 Indicative insurance premia 69 Table A3.1 Bilateral investment treaties concluded by selected Latin American and Caribbean countries, October 1996 81 Table A3.2 Political risk insurance schemes 82-85 Table 6.1 Expected government losses in Colombian infrastructure projects 141 Figlres Figure 1.1 Sources of net long-term resource flows to developing countries 3 Figure 1.2 Recipients of net long-term resource flows to developing countries 3 Figure 1.3 Private cross-border infrastructure finance 4 Figure 2.1 Public sector borrowing requirement 22 Figure 2.2 Cumulative private sector borrowing for infrastructure, 1985-95 24 Figure 2.3 Sectoral composition of infrastructure financing in developing countries 24 Figure 2.4 Net private flows to infrastructure, 1990-96 25 Figure 2.5 The privatization-nationalization cycle 25 Figure 2.6 Estimated cost of mispricing and technical inefficiency 35 Figure 5.1 The relationships between the different types of interest rate- and exchange rate-related risks 111 Figure 5.2 Determining whether selective macro guarantees are appropriate 127 Figure 6.1 Integrated enterprise risk management: Optimizing enterprise returns under uncertainty 133 Figure 6.2 Risk identification lattice 136 Figure 6.3 Sources of risk in the El Cortijo-El Vino toll road project 137 Figure 6.4 Sensitivity analysis for the Colombian toll road project 142 v Preface T his volume reports on the findings of a confer- Thobani, Senior Economist in the Latin American ence aimed at advising policymakers from and Caribbean Region Chief Economist's Office, were Latin America and the Caribbean (LAC) on the task managers for the project. Timothy Irwin, how to manage government exposure to private infra- Economist in Mr. Klein's Group, helped edit the structure projects. The conference was held in volume. Cartagena, Colombia in May 1997. It was conceived The Bank's Economic Development Institute pro- by Guillermo E. Perry, Chief Economist of the Bank's vided major funding for the conference and collaborat- Latin American and Caribbean Region, who also pro- ed in the design and delivery of the seminar. Additional vided overall guidance on the project. In his previous funding came from the Bank's Research Program, from capacity as Finance Minister for Colombia, Mr. Perry the Colombian Ministry of Finance, which funded the found that there was little practical guidance available local expenses for the conference, and from fees charged on whether governments should assume risks in private to nongovernment participants. Katharine Brewer infrastructure projects; on ways to reduce the risks; on played a key role, along with Mateen Thobani and L.K. how these risks should be allocated among taxpayers, Arora, in organizing the conference. Ms. Brewer also consumers, and investors; and on how the government assisted with all stages of preparation for the volume. R. should value and budget for any risks it assumes. David Gray read and commented on all the chapters. Michael Klein, then Manager of the Private Barbara Karni copyedited the volume. Cristina Palarca Participation in Infrastructure Group, and Mateen assisted with word processing. vii Contributors Roberto Abusada is an advisor to Peru's Ministry of ation, mineral extraction, transportation and civil Economy and Finance and the head of Econsult S.A., engineering. Recently he has been involved in a num- a private consulting firm. Dr. Abusada also teaches at ber of projects resulting from the UK Government's Peru's Catholic University, is a director of the Andean Private Finance Initiative. Mr. Alington is a graduate Development Corporation (CAF), chairman of the of the University of Cambridge. board of the Aeroperu airline, and the director and founder of the Peruvian Institute of Economy. William Chew is a managing director in charge of Between 1980 and 1983 he served as deputy minister project finance ratings at Standard & Poor's, where he of commerce, deputy minister of economy, chief advi- developed the Standard & Poor's project rating busi- sor to the Minister of Economy, and presidential advi- ness in 1991. He has been involved in rating all types sor and member of the Board of Directors of the of project financing, including independent power, Central Bank of Peru. Dr. Abusada has taught eco- energy, and infrastructure projects. Previously he nomics at the ESAN business school, the Catholic headed Standard & Poor's rating criteria in several and Pacific Universities in Peru, and Boston areas, including independent power, solid waste, University. He holds a Ph.D. in economics and com- pooled financing, short-term debt, and derivative pleted his studies at the Catholic University in Peru, securities. Mr. Chew is a member of Institutional as well as at Cornell and Harvard Universities. Investor's Infrastructure Finance Institute, the editor of The Journal of Project Finance, and is a former Nigel Alington has worked for the Bain Hogg Group member of the U.S. Environmental Protection in London, part of Aon Corporation, since 1968. Agency's Financial Advisory Board. He joined After specializing initially in construction insurance Standard & Poor's in 1979 after serving on the fiscal and surety bonding, he joined the projects depart- staff of the New Jersey State Legislature. Mr. Chew ment of The Credit Insurance Association Ltd. (a holds a B.A. from Trinity College, Connecticut, and Bain Hogg subsidiary) in 1975 where he specialized an M.A. from the University of Chicago. primarily in credit and political risk insurance and consultancy services for international projects and Mansoor Dailami is principal financial economist and investments. In recent years he has been responsible team leader in the World Bank's Infrastructure for coordinating Bain Hogg's services in all areas of Finance Group of the Economic Development insurance and guarantees related to major infrastruc- Institute's Regulatory Reform and Private Enterprise ture projects. He has been involved in concession and Division. In his eleven years with the Bank, he has BOT projects since 1988, in the fields of power gener- worked in the South Asia, Africa, Middle East, and ix DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Latin American regions; served as task manager on of Technology (1991) and a Ph.D. in statistics from major adjustment and financial sector loans to India; Stanford (1987) as well as an engineering degree from headed the Economic Unit of the Bank's resident mis- the Universidad de Chile. sion in India; and been involved in infrastructure pro- ject financing. Before coming to the World Bank, Mr. Jose Fernandez is a partner in the New York law firm Dailami worked at the United Nations Secretariat in of O'Melveny & Myers LLP, where he heads the Latin New York, the Massachusetts Institute of Technology, American practice group. Mr. Fernandez's work prin- and New York University. He holds a Ph.D. in eco- cipally involves financings, mergers and acquisitions, nomics from Harvard University. and securities offerings in Latin America and other regions of the developing world. In the last few years, Clemente del Valle is director general of public credit Mr. Fernandez has advised bidders in the privatization in the Colombian Ministry of Finance and Public of Compafnia An6nima de Telefonos de Venezuela Credit where he is in charge of the management of the (CANTV), and the governments of Uruguay and Republic of Colombia's public credit policy, and Zambia in the reform of their telecommunications responsible for the development and implementation of sectors. He headed the team representing the Republic the privatization policy of the Samper Administration. of Peru in the privatizations of CPT/Entel. In the Prior to his position in the Ministry of Finance, Mr. del finance area, Mr. Fernandez has represented many Valle was vice-president director of investment banking U.S. and foreign banks in loans and restructurings in for the Corporaci6n Financiera del Valle (1991-1994); Brazil, Bolivia, Chile, Colombia, Mexico, and director general of international trade and acting Venezuela dealing with the aircraft, telecommunica- viceminister at the Ministry of Economic Development tions, and energy industries. In the securities field, he (1989-1991); and deputy director of operations and is active in several infrastructure and project finance head of planning at the Ministry of Finance assignments around the world. Mr. Fernandez holds a (1986-1989). Mr. del Valle has taught economics at B.A. from Dartmouth College and a J.D. from the Universidad de Los Andes and is a member of the Columbia University Law School. board of directors of numerous organizations in Colombia, including the Instituto de Fomento Eduardo Fernandez-Arias is senior research economist Industrial (IFI), Financiera Energetica Nacional (FEN), in the Office of the Chief Economist at the Inter- ISA, and ISAGEN. He holds an M.A. in economics American Development Bank, and is presently on from both the London School of Economics and the leave from the World Bank's Debt and International Universidad de Los Andes. Finance Division. Dr. Fernandez-Arias has worked extensively and published in professional journals on Eduardo Engel is associate professor at the Center for issues related to international capital flows to develop- Applied Economics, Department of Industrial ing countries and the role of multilateral development Engineering, Universidad de Chile in Santiago. Dr. banks in that process. He holds a Ph.D. in economics Engel is also a faculty research fellow at the National and an M.A. in statistics from the University of Bureau of Economic Research (NBER) and was for- California at Berkeley. merly an assistant professor at Harvard University's Kennedy School. His main areas of expertise and Ronald Fischer is research professor of economics at interest are in macroeconomics (external shocks and the Center of Applied Economics, Department of stabilization, employment and productivity dynamics, Industrial Engineering, Universidad de Chile in investment equations), regulation (consumer protec- Santiago. From 1987-93 Dr. Fischer was assistant tion, infrastructure deregulation), and applied statis- professor of economics at the University of Virginia. tics (forecasting models, survey design and analysis, He has published articles in the American Economic and applied stochastic modeling). Dr. Engel holds a Review, International Economic Review, and Estudios Ph.D. in Economics from the Massachusetts Institute P4blicos among others. Dr. Fischer holds a B.S. in x CONTRIBUTORS civil engineering from the Universidad de Chile, and concentrating on Mexico and East Asia economies. an M.A. and a Ph.D. in economics from the Dr. Klein returned to the World Bank in 1993 to University of Pennsylvania (1987). establish the PPI Group, where he worked on issues of infrastructure financing and market structure on all Alexander Galetovic is assistant professor of econom- continents. He has published articles on aspects of ics at the Center of Applied Economics, Department economic policy in a number of developing countries, of Industrial Engineering, Universidad de Chile in concentrating most recently on the private provision Santiago. His current research interests include the and financing of infrastructure. Dr. Klein holds a economics of highway franchises, the industrial orga- Ph.D. in economics from the University of Bonn, nization of financial and high-technology markets, Germany. and the political economy of coups d'etat. Dr. Galetovic holds an M.A. degree from the Catholic Christopher Lewis is senior manager of the Tax University of Chile, and a Ph.D. in economics from Analysis and Economics Group of Ernst & Young Princeton University (1994). LLP. His special focus is on corporate risk manage- ment, insurance markets, small business loans, cata- Timothy Irwin is an economist in the Private strophic disaster risk, capital standards, government Participation in Infrastructure (PPI) group of the risk management systems, contingent claims analysis, World Bank. Since joining the Bank, he has worked and federal credit insurance programs. Prior to joining on public-management reform and the regulation of Ernst & Young, Mr. Lewis served as senior financial private infrastructure firms. Before that, he worked for economist in the Office of Federal Housing Enterprise the New Zealand Treasury on trade liberalization, Oversight (OFHEO) from 1995 to 1997-an inde- labor-market policy, and the implementation of pub- pendent office within the Department of Housing lic-management reforms. Mr. Irwin holds an M.PA. and Urban Development established to regulate the from Princeton University. Federal Loan Mortgage Corporation (Freddie Mac) and the Federal National Mortgage Association Alejandro Jadresic Marinovic is minister president of (Fannie Mae). From 1990 to 1995 Mr. Lewis served the National Energy Commission in Santiago, Chile. as senior economist in the Office of Management and Prior to that, he was in charge of the Industrial and Budget (OMB). He holds an M.A. in economics and Commercial Policy Department at the Ministry of a B.A. in international finance both from the Economics and served as president of the Preventive University of Connecticut. Antitrust Commission. Minister Jadresic is a professor of Industrial Organization at the Universidad de Chile Ignacio Mas is principal consultant with Analysys, and has published several articles on microeconomic Ltd., a telecoms strategy consultancy, based in policies, industrial organization, and economic regula- Cambridge, United Kingdom. Prior to joining tion. He holds a Ph.D. in economics from Harvard Analysys, Dr. Mas was a financial economist with the University. World Bank and the International Finance Corporation in Washington, D.C., where he worked on a variety of Michael Klein is chief economist of Shell capital market projects in emerging markets. He was International in London and was formerly the manag- also involved in trading functions in the Bank's own er of the World Bank's Private Participation in asset liability and portfolio management groups, and Infrastructure (PPI) group. He joined the World Bank was assistant to the managing directors. Dr. Mas holds in 1982 as an economist working in oil and gas opera- a Ph.D. in economics from Harvard University. tions, and subsequently on industrial, trade, and financial sector policies, and macroeconomic issues. In Colin Mayer is the Peter Moores Professor in 1991 he joined the Economics Department of the Management Studies at the University of Oxford in OECD to head its unit for non-OECD economies the United Kingdom. He has a professional fellowship xi DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE at Wadham College and is deputy director (Academic) books, including Public Finance, Stabilization and at the School of Management in Oxford. Professor Structural Reorm in Latin America (1994), and has Mayer is a leading international expert in corporate written numerous articles on macroeconomic, fiscal finance, accounting, and taxation. He has published policy, and energy policy issues. He holds a B.S. in several books and artides on these subjects, as well as engineering and an M.S. in economics both from the advised a number of companies in the public and pri- Universidad de Los Andes in Bogota, Colombia. vate sectors on the measurement of accounting rates of teturn, corporate strategy, the impact of regulation and David Roberts, now at NationsBank in New York, the cost of capital. Prior to his appointment at the was formerly senior vice president and managing School of Management, he was professor of economics director of the international division at Duff & Phelps and finance at the University of Warwick (1992-94), Credit Ratings Co. (DCR). At DCR Dr. Roberts Price Waterhouse Professor of Corporate Finance and headed the international credit and rating committee, Director of the Centre for Study of Financial and oversaw sovereign, international corporate, and Institutions at the City University Business School international structured finance ratings. Prior to join- (1987-92), and fellow in economics at St. Anne's ing DCR, Dr. Roberts was a senior officer at the College, University of Oxford (1980-86). He holds a Federal Reserve Bank of New York. He holds a Ph.D. D.Phil. in economics from the University of Oxford. in economics from Rice University. Ashoka Mody is presently on leave from the World Warrick Smith is senior private sector development Bank's Project Finance and Guarantees Group, where specialist in the World Bank's Private Participation in he is principal financial economist, advising on regula- Infrastructure Group. Since joining the Bank in 1993, tory and financial policy, and on project structuring. Mr. Smith has advised on regulatory and institutional He is the author of a number of papers and articles on reform issues associated with private involvement in these and related subjects. Prior to joining the World infrastructure throughout Latin America, Asia, Bank in 1987, he was with AT&T's Bell Laboratories. Eastern Europe, and the Middle East. Before joining Dr. Mody is teaching at the University of the Bank, Mr. Smith was secretary to Australia's Pennsylvania's Wharton School of Business for the National Competition Policy Review, an independent 1997-98 academic year. inquiry established by the Prime Minister to develop a new regulatory and institutional framework for com- Guillermo E. Perry is the chief economist of the Latin petition policy in Australia, with an emphasis on the America and Caribbean (LAC) region of the World introduction of competition into utility industries. He Bank and Director for LAC of the Poverty Reduction previously held posts with the Australian Attorney- and Economic Management Network (PREM). Prior General's Department, Department of Primary to joining the Bank, Mr. Perry served in various Industries and Energy, and the Department of Trade, capacities in the Colombian government: as minister and also spent a period in private legal practice. He is of finance and public credit from 1994-96, member trained in law, economics, and public administration, of the constitutional assembly (1991), senator of the with degrees from the Australian National University republic (1990), minister of mining and energy and Harvard University. (1986-88), and director of the general directorate of national taxes (1974-76). Mr. Perry was also director Mateen Thobani is principal economist in the of two of Colombia's leading economic think-tanks: Poverty Reduction and Economic Management Unit FEDESARROLLO from 1988-89 and the Center for (PREM) in the Latin America and Caribbean (LAC) Economic Development Studies (CEDE) from region of the World Bank. With the World Bank since 1972-74. He was a partner and served as general 1981, Dr. Thobani has most recently worked in the director of Majla, Millan & Perry Ltda., a consulting LAC Chief Economist's Office on improving the qual- firm, from 1976-85. Mr Perry is the author of several ity and relevance of the Bank's analytical and advisory xii CONrRIBUTORS services. Prior to that, he served as country economist journal artides and books in the areas of water mar- in several countries in Latin America and South Asia, kets, education, urban transport, agricultural pricing, and in the environment and agriculture operations and trade policy. Dr. Thobani holds a Ph.D. in eco- division for Peru and Brazil. He has published several nomics from Yale University. XIii 1 Dealing with Public Risk in Private Infrastructure: An Overview Timothy Irwin, Michael Klein, Guillermo E. Perry, and Mateen Thobani ABSTRACT The current wave of infrastructure privatization is extensive guarantees. With those policies in place largely a positive development. The transfer of risk to developing country governments should be able to private operators should lead to the development of restrict their risk-bearing to certain political and regu- new infrastructure, improvements in the operation of latory risks over which they have direct control. existing infrastructure, and a reduction in budgetary When governments do provide guarantees they subsidies. Yet it also raises problems for governments. should attempt to measure the costs of these guaran- Infrastructure privatization in the developing world tees and improve the way they treat them in their has frequently been accompanied by extensive residual accounts and budgets. Measurement and budgeting risk-bearing by governments, which threatens to viti- are critical to improving decisions about whether to ate its efficiency benefits and confront future govern- provide guarantees, to improving project selection and ments with large financial liabilities. To solve these contract design, and to protecting governments from problems governments need to institute policies that unknowingly entering into commitments that might make investment attractive even in the absence of jeopardize future budgets. 1 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE P rivate investment in infrastructure has increased To protect themselves from these risks private enormously over the past decade. As recently as investors often ask the host government to provide 1988 foreign private investment in infrastructure extensive guarantees against risks such as those of amounted to $100 million; in 1996 the figure was $20 nonpayment by purchasers, cost overruns, or low billion. The increase is welcomed for several reasons: demand. In other words, they ask the government to * Private firms typically have stronger incentives enter into some form of arrangement that results in than government enterprises to build and run the government's-and not their-net wealth varying infrastructure businesses effectively and at low cost with the risky outcome. Anxious to encourage invest- and-if prices reflect costs and the firm's profits ment, governments often consent. depend on consumer demand-to choose good Poorly designed guarantees threaten, however, to projects and avoid "white elephants." undermine the benefits of privatization. First, they can * Privatization encourages and facilitates the charg- blunt the private investors' incentives to choose only ing of cost-covering tariffs, thus addressing the good projects and to run them efficiendy. If the govern- problem of underpricing that has afflicted many ment bears the risk of the project's failing, the private publicly provided infrastructure services. investor is willing to invest in projects that are likely to * Greater efficiency and cost-covering prices together fail; having invested in a project, the private investor has allow investments and services to be provided that little interest in maximizing its chance of success. might not otherwise have been possible, while Second, guarantees may impose excessive costs on the simultaneously improving the government's fiscal host country's taxpayers or consumers and expose them position by making available the same quantity and to too much risk. Since guarantees rarely show up in the quality of service with smaller budgetary subsidies. government's accounts or budgets, governments may Thus there are both microeconomic and macroeco- not know what their exposure is. Moreover, a severe nomic benefits of private investment in infrastructure. recession or economic crisis could trigger many guaran- But such investments subject investors to major risks, tees simultaneously; many of the government's contin- since the investments are often large and their costs can gent liabilities might thus become actual and current all be recouped only over long periods of time. Two special at once. The problem may not be immediate, but as features of infrastructure create additional risks. First, government's infrastructure-related risk exposure grows the investments are largely sunk; the assets cannot be the chances of trouble arising will also increase. used elsewhere except at great cost. Second, infrastruc- Governments just embarking on the transition to a ture projects often provide services that are considered more market-oriented economy may find they face essential and are provided by monopolists. As a result political constraints that prevent them from introduc- services are highly politicized. As many of the chapters ing all the policies that would permit privatization in this volume emphasize, this combination of factors without large-scale risk-bearing. Until they can raise makes investors especially vulnerable to opportunistic prices to cost-covering levels, for example-or raise govemment actions. Before the investment is made the taxes to provide explicit subsidies-they may have to government has every reason to promise to treat the bear certain risks. Compared with the alternative of investor fairly-to allow cost-covering tariffs and to continued public ownership, privatization with signifi- avoid changing regulations in a way that would adverse- cant risk-bearing may be desirable. Yet, as this volume ly affect the investor. Once the investment is made, argues, governments that introduce good policies can however, the government has an incentive to renege on attract private investment without themselves bearing its promises, since it can satisfy political demands to commercial or macroeconomic risks. When they do reduce prices or otherwise appropriate the investor's assume risk, they need to identify it and, where feasi- profits without causing the investor to pack up and ble, measure and budget for its expected cost. leave. Because of these characteristics private investors' The rest of this chapter introduces the papers that returns are uncertain and are more sensitive than in follow. It first describes how private investment in most industries to the host government's behavior. infrastructure (construed broadly to include telecom- 2 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW munications, power, gas, water, and transport) has receipts rose only from $63 billion to $85 billion over grown and what sorts of risks governments have the same period. retained in private projects. It then considers how Private cross-border finance for infrastructure pro- governments can reform policy to make infrastructure jects has exploded as well, rising from $0.1 billion in projects more attractive to private investors and thus 1988 to $20.3 billion in 1996. By now, more than reduce or eliminate the amount of risk governments 100 governments have involved the private sector in have to assume. Finally, it shows how governments infrastructure provision. should decide whether or not to bear risks and how Despite this tremendous growth private investment they should measure, budget for, and account for the still accounts for only about 15 percent of total invest- costs and risks they do assume. ment in infrastructure. Given the continuing disen- chantment with the performance of government-provid- ed infrastructure in much of the developing world and The Growth of Private Investment and the continuing desire of most governments to reduce Government Guarantees cash expenditures, the demand for private finance will likely continue to grow. On the supply side, investors The Growth of Private Investment also have an interest in private infrastructure, because of the opportunities it affords in terms of returns and As Dailami and Klein note in chapter 2, long-term diversification (see World Bank 1997c). Thus although flows of private capital (for all sectors, not just infra- a continuation of the trend toward private investment is structure) have grown rapidly in recent years. From not inevitable-large-scale macroeconomic problems or 1990 to 1996 the net flow of private funds rose from a new wave of expropriation and nationalization could $44 to $244 billion a year. During the same period, both deter investors-it appears likely. public flows fell-from $56 to $41 billion. The recipients, as well as the providers, of capital , , . , . ~~~~~The Growth and Nature of are now predominantly private. With growing privati- zation the annual flow of resources received by the private sector in developing countries rose from $38 Investors in private infrastructure have usually not billion in 1990 to $200 billion in 1996. Public been willing to bear the risks of these projects alone, FIGURE 1.1 FIGURE 1.2 Sources of net long-term resource flows Recipients of net long-term resource flows to developing countries to developing countries Billions of dollars Billions of dollars 300 300 250 250 200 200 150 150 100 100 rPrivate | R rSivate | 50 so 1990 1996 0 1990 1996 Source: World Bank 1997a. Source: World Bank 1997a. 3 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE FIGURE 1.3 that the private owner receives at least a minimum Private cross-border infrastructure finance level of revenue when demand is lower than expected, Billions of dollars thus shifting some of the risk of variation in demand 20 to the government. In the El Cortijo-El Vino toll road project in Colombia, for example, discussed in 15 chapter 6, the government undertook to reimburse the concessionaire if traffic was less than 90 percent of D the specified level, agreeing to pay the concessionaire 10 an amount equal to the toll times the difference as 1)cbt between 90 percent of the estimated number of vehi- cles and the actual number of vehicles. S5 Governments bear similar risks in other sectors. The Colombian government, for instance, provided a minimum-revenue guarantee when it awarded a 1988 1990 1992 1994 1996 build-operate-transfer concession for a new runway at Source: World Bank 1997a. Bogota's El Dorado airport in 1995 Juan 1996). And many governments, through their utilities, have agreed and have demanded that government assume some of to pay independent power producers a fixed amount the risk (on the definition of "risk," see box 1.1). each year that is independent of the actual level of Although the magnitude of the risk borne by govern- power subsequently demanded from them. These vol- ments is not known, anecdotal evidence from many ume or revenue risks are the focus of chapter 4. projects suggests that government risk-bearing in pri- vately financed projects has grown commensurately Payment risk. An agreement by a state-owned utili- with private infrastructure. Governments share vari- ty to pay an independent power producer irrespective ous types of risk, including demand risk, payment of demand protects the investor from the risk of risk, exchange and interest rate risk, and political and fMling demand for power or of new and cheaper gen- regulatory risk. They also bear implicit risks. erators coming on stream in the future. But it does not protect the investor from the risk of the utility Demand risk. In privatizing toll roads the host defaulting on its obligations. To protect themselves government has often committed itself to ensuring against this risk, investors usually ask the government, Box 1.1 Defining "risk" In finance theory and the applied fields of securities analy- according to plan. Thus, for example, investors estimate sis and portfolio management, "risk" is often used to refer the returns they will earn on the assumption that the to the volatility of returns around an average or expected government does not expropriate their investment but return. In this sense, risk is equivalent to the statistical note a risk of expropriation. An increase in expropria- concept of variance, and a project's risk can increase with- tion risk in this sense does not just increase the volatility out any change in the expected (or mean) return on the of returns, it reduces the expected return. Even risk-neu- project. Investors who were risk neutral (in the sense used tral investors would prefer to avoid these "risks." in economics and finance) would be indifferent to risk in Diversification cannot eliminate this risk; it can only this sense, and risk of this sort can be effectively eliminat- spread the loss among many people. ed by diversification if it is not systematic. In this chapter the term "risk" is used in the sense of In project finance, on the other hand, "risk" fre- variance-or volatility around a statistically expected quently refers to the ways in which actual results may be outcome. Expropriation risk, for example, is thus the worse than planned. Here the benchmark is not the volatility in returns around an expected return attribut- expected return of the project but the (generally higher) able to uncertainty over whether the government will return that investors would receive if everything went expropriate. 4 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW which is more creditworthy than the utility it owns, to discussed above, but the state government bears sev- guarantee the utility's payments (on the definition of eral risks that are tied to actions taken, or influ- guarantee," see box 1.2). enced, by the government. If, for example, the gov- ernment subsequently bans toll roads or takes Exchange and interest rate risk. Governments have actions that deliberately reduce the profitability of sometimes borne the risks associated with adverse the private investor, the government will compensate fluctuations in exchange and interest rates. The the investor. The government also bears risk associat- Spanish government, for example, had many private ed with possible court findings that aboriginal land toll roads built during the 1960s and early 1970s rights have been violated and with strikes on the and bore exchange rate risk on foreign loans that construction site that are undertaken as a protest financed the roads. G6mez-Ibafiez and Meyer (1993) against the state government rather than as part of describe the guarantees given and their rationale as site-specific disputes. follows: Under the government of Pakistan's policy frame- work for private power generation, to take a second The Spanish government had required the early conces- example, the government agreed to "cover certain sions to finance a large part of their costs from foreign political and governmental force majeure risks, pro- debt in order to ease Spain's balance-of-payments prob- vide protection against changes in certain taxes/duties, lems and to avoid drawing away domestic savings from and ensure foreign exchange convertibility for the pro- other projects. The 1972 law [on toll road concessions] jects." (International Finance Corporation 1996, set standards that at least 45 percent of construction costs p. 49). be financed from foreign loans, at least 10 percent from equity, and no more than 45 percent from domestic Implicit risk-bearing. Government risk-bearing loans. The early Spanish [highway] companies had trou- need not be made explicit in contracts or laws. ble raising funds from foreign capital markets, however, Sometimes everyone concerned expects that the gov- and in return the government agreed to guarantee some ernment will in fact bail out a company if it would of these loans and to protect the companies from otherwise fail. The case of private Mexican toll roads exchange rate fluctuations. The 1972 law specified that may provide such an example. They were partly the government would guarantee up to 75 percent of the financed by commercial banks, which were owned at foreign loans; moreover, all foreign loans would be denominated in pesetas with the government assuming Box 1.2 the full exchange rate risk (p. 126). Defining "guarantee" That is, if the peseta depreciated relative to the foreign As Smith points out in chapter 3, the term "guaran- tee" is used in different ways. In Smith's strict use of currencies in which the loans were made, the conces- the term, one party can guarantee another party s sionaire's loan repayments would remain the same but behavior, but it cannot guarantee its own. Thus, for the Spanish government would make an additional example, the government can guarantee the payment performance of a legally separate business it owns, but payment to ensure that the foreign lenders received no it cannot guarantee its own permission to convert cur- less foreign currency. In the end the Spanish taxpayer rency. Only a third party, such as a multilateral devel- about $2.7 billion as a result of the guarantees opment bank, can do that. spent about 32.7 bllhon as a result of the guarantees The term "guarantee' is also used more broadly to (see chapter 2). mean simply a commitment to bear a risk. When a gov- ernment gives an exchange rate guarantee, in this sense, it is just agreeing to assume exchange rate risk. In this Political and regulatory risk. Governments often chapter the term is used in the broad sense to mean the bear certain political and regulatory risks, even when assumption of a risk. As with "risk," it is important to they bear none of the risks mentioned above. In the be clear about how "guarantee" is being used in a given context, since the implications of giving a guarantee Melbourne City Link, a private toll road in dependonwhichtypeofguaranteeitis. Australia, private-sector parties bear most of the risks 5 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE the time by the government and which, it has been investment despite its policy of not bearing even reg- argued, exercised less care than they should have in ulatory risks except where they relate specifically to a assessing the credit risks they were assuming. project (United Kingdom 1995). When developing Although the Mexican government did not explicitly countries have introduced good policies and main- agree to bear the credit risks taken on by the banks, it tained them for a few years, they have also been suc- did in the end bail them out when, partly as a result cessful in attracting private infrastructure capital of the poor financial performance of the private toll without guarantees. In Chile private firms have roads, they got into trouble. Some argue that the bail- invested in telecommunications, power, and gas with- out had been expected by the banks and that this out government guarantees (see the commentary by expectation had an effect similar to that of an explicit Jadresic following chapter 2). In Colombia investors government guarantee. have "gradually dropped requirements for guarantees of the performance of government purchasers" (Klein Otherforms of risk-bearing. Governments bear risk 1996a). In Argentina the complete restructuring and in other, less obvious ways as well. They may lend privatization of the power industry permitted the directly to projects and bear repayment and perhaps government to attract private investment without interest rate risk. They may become part owners of a having to assume major risks or issue guarantees project and thereby bear a proportion of the overall (Klein 1996a). risk of the project. Moreover, as Mas points out in I Policy reform in four areas-macroeconomic poli- chapter 5, governments own a share of many firms, in cy, regulatory policy, information disclosure, and capi- an economic if not a legal sense, through the corpo- tal market liberalization-can help attract private rate tax system: if profits are high, the government investment that does not depend on government gets more corporate income tax; if they are low, it gets guarantees. (For more on these issues see also Asian less. Development Bank 1997a and 1997b). Policies That Reduce Risk Pursuing Stable Macroeconomic Policy As Mas notes in chapter 5, a stable macroeconomic Governments issue guarantees in order to make pro- environment does a great deal to reduce risks for pri- jects attractive to investors, often using risk-bearing as vate investors. The government can make a large con- a way of compensating for shortcomings in the gov- tribution to creating and maintaining a stable envi- ernment's present, and expected future, policies. But ronment by maintaining stable prices and a balanced as the authors in this volume show, the assumption of budget. It can also take actions to increase the likeli- risk by government creates its own problems-not for hood that it will continue to act prudently in the investors but for the government and its citizens, who future. For example, it can issue inflation-indexed are subject to usually hidden costs and unknown risks. local currency debt to reduce the temptation to reduce The first, and most desirable, thing governments can the real level of the debt by inflating. When good do to make projects more attractive is therefore to put macroeconomic policies are in place, the likelihood of in place good policies that reduce risks and raise large changes in the exchange rate and interest rates expected returns. Governments that have established are reduced, though not eliminated, and the pressures good policies and have persuaded investors that they on governments to prevent convertibility and transfer- will maintain them can attract private investment ability are lessened. Demand becomes easier to fore- without extensive risk-bearing. cast as well. Private investment in infrastructure without Over time governments in developing countries much government risk-bearing is common in many such as Chile and Colombia have developed reputa- OECD countries. In the United Kingdom, for exam- tions for pursuing reasonably stable macroeconomic ple, the government attracts large amounts of private policies and have received investment-grade credit rat- 6 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW ings. The governments of certain rich countries, such * Signing international treaties (that bind the gov- as those of Singapore and Switzerland, have even ernment to permit convertibility and transferabili- stronger reputations. ty, for example) * Agreeing to be bound by international arbitration. The design of concession contracts-themselves a Designn God R P a form of regulation-can also reduce risks when com- Appropriate Contracts petition is not feasible, as Engel, Fischer, and In regulatory policy, as in macroeconomic policy, one Galetovic show in chapter 4. Changes in contract of the government's major challenges is to develop a design resulting from the measurement of risk under reputation for treating firms reasonably-a challenge alternative project designs can reduce project risks, as that is easy to state but hard to meet. As Smith shows Lewis and Mody show using the example of a toll in chapter 3, however, there are several steps that gov- road in Colombia (chapter 6). ernments can take to reduce regulatory risks before they have had a chance to build reputation. Perhaps the most important measure is to expose o the infrastructure service to competition whenever The government can also reduce risk by publicly dis- possible. Competition encourages better performance closing relevant information. In the macroeconomic by firms and enables the government to let private domain Mas notes that the government can improve firms make decisions about which investments to private investors' ability to forecast the future, and undertake. Moreover it reduces the political pressure thus reduce the perceived riskiness of projects, by on governments to intervene in markets. When firms making relevant information publicly available. have monopolies consumers will look to the govern- Indeed, one of the actions taken by the Mexican gov- ment to keep prices down, and the government will emient in the wake of the 1994 crisis was to publish come under pressure to keep prices below costs. When quarterly macroeconomic updates that are available a firm operates in a competitive market, little or no on the Internet.t The government can publish fre- economic regulation is necessary and consumers will quent and regular accounts showing the development look to the firm's competitors to keep prices down. As of its financial position. It can compare outcomes Mas notes, the government will also be better able to with forecasts and regularly update the forecasts. resist the pressure to bail out firms that have failed in Those statements can, moreover, include the sorts of competitive markets-that is to avoid giving implicit information on guarantee exposure discussed below. guarantees of commercial risk. The monetary authorities can explain publicly their In some instances monopolies may be unavoid- model of the macroeconomy and routinely discuss able. In such instances governments can still reduce their policy intentions. The statistics department can risk, mainly by establishing laws and regulations that collect and quickly publish data on macroeconomic protect property rights and by enforcing them in a fair outcomes. The regulatory authorities can clearly and and consistent manner. Specific economywide options comprehensively explain the regulatory framework cited by Smith include the following: before private investors have to commit themselves. * Establishing expert regulatory agencies that have some independence from the rest of the govern- Liberalizing Capital Markets ment and are thus partially insulated from popular pressure to keep prices below costs The government can help others manage risk at lower * Reforming the constitution to impose limits on cost by liberalizing financial markets. In the local mar- the power of the executive to act arbitrarily ket it can remove barriers to entry by new firms and * Strengthening the independence and quality of remove restrictions on the services that firms can offer. the judicial system, so that it can act as a restraint It can also give local citizens and firms full access to on the executive international capital markets and the diversification 7 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE and hedging instruments they provide-while ensur- transactions costs of any allocation must also be con- ing that banking supervision, exchange rate, and other sidered. An allocation assigning each of a project's policies are consistent with the liberalization. many risks according to each party's control over the The most direct advantage of capital market liber- outcome and its costs of risk bearing may require alization is that it permits improvements in the alloca- detailed analyses, tough negotiations, complex legal tion of risks. Free local capital markets allow risks to contracts, expensive monitoring arrangements, and be redistributed within the country to those people possibly the high costs of settling disputes in court. that can bear them at least cost. The removal of barri- The optimal allocation of risk takes these costs into ers to international financial markets permits further account. diversification and redistribution. Permitting local cit- izens to invest in foreign countries and in foreign cur- rencies, for example, lets them diversify their portfolio internationally and thus reduce their exposure to the Some risky outcomes are more easily controlled by health of the local economy. As Mas notes, removing private firms, others by the government, as illustrated capital controls also gives the government quick feed- by the examples below. back on its performance, since bad policies can quick- Consider the demand risk in a telecommunications ly cause increases in interest rates. concession-that is, the fact that demand may be high- er or lower than the best forecast. Since the firm can increase demand by keeping quality high-preventing Principles of Risk Allocation faults, fixing them quickly when they do occur, improv- ing sound quality, introducing new services, and so How should governments decide whether to bear risks on-quality may be higher if the firm bears demand in a private infrastructure project? If they do decide to risk (that is, if the firms' profits vary with demand).2 bear risk, which risks should they rake on? Other things being equal, then, demand risk in a Infrastructure project risk can be allocated, at a telecommunications concession should be allocated to broad level, to government, firms, or consumers. the firm. (Note that the same line of reasoning may not These agents can in turn redistribute risks to others. give the same results in all sectors; see the discussion of Firms, for instance, choose how to allocate risks demand risks in toll roads below.) among lenders, shareholders, and insurers. Risks allo- To take a second example the government alone cated to the governments are ultimately borne by the controls whether local currency can be converted into country's taxpayers. foreign currency (convertibility risk) and whether for- Two critical factors determine whether an agent eign currency can be transferred out of the country should bear risk: the degree to which the agent can (transferability risk). Convertibility and transferability influence or control the outcome that is risky and the risks can therefore be reduced by allocating them to agent's ability to bear the risk (that is, its cost of risk- the government. Note, however, that governments do bearing). Other things equal, risks should be allocated not have the same control over the exchange rate to agents who can best control the risky outcome and itself, so there is no comparably strong argument for to agents who can bear the risk at the lowest cost their bearing exchange rate risk. (because they are the least risk-averse, because they can most easily insure or hedge against the risk, or Governments responsiveness to financial incentives. because they can spread the risk among many people). Allocating risks to the government will improve out- These two factors often push in different direc- comes only if the government responds to financial tions-the group or organization that has most con- incentives. Since government decisionmakers often do trol over the risky outcome may not be in the best not act in the interests of their citizens, governments position to bear the risk. In this case the various costs are generally less responsive than firms to financial and benefits must be weighed against each other. The incentives. Allocating government-controlled risks to 8 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW the government may thus do less good than allocating whom invest only a small fraction of their wealth in firm-controlled risks to firms. When the financial con- the company. As a result, the private sector can proba- sequences of the government's risk-bearing do not bly bear risk as cheaply as governments.3 show up in its budgets or accounts, government may be less responsive still. If, for instance, the likelihood Customers versus investors and locals versus foreign- of the government's permitting convertibility and ers. Risk-bearing costs should also inform decisions transferability is unaffected by any obligation to pay about the allocation of risks between customers and compensation in case of malfeasance, there is no value investors and between locals and foreigners. Consider, in allocating these risks to government. A similar for example, a case in which consumers bear demand point applies to the cost of risk-bearing discussed risk, something that happens under revenue-cap price below: since political decisionmakers face weak incen- regulation. Under a revenue cap, prices are set in order tives to lower costs (and have poor information about to give the firm a certain level of revenue but no more. the government's exposure to risk) governments may If demand rises, the regulated maximum price falls; if be less adept than private investors at taking advan- demand declines, the regulated maximum price tage of opportunities to reduce risk, through diversifi- increases. This form of regulation may create high cation or hedging for example. risk-bearing costs, since the value of local consumers' assets, and the income they get from the assets, may The Costs of Bearing Risk be closely correlated with the demand for the regulat- ed service. Policies that require the government to Governments versus private firms. In general, the bear demand risk are also likely to have similar conse- public and private sectors appear to have similar costs quences, since the government will suffer the conse- of risk-bearing. Because governments can spread risks quences of low demand just when its tax revenues among all their taxpayers, the governments of large have fallen and expenditure on nondiscretionary wel- countries with broad tax bases may have a relatively low fare spending has risen.4 cost of bearing risks-especially the risks entailed by Foreign investors, on the other hand, may be well small projects. Such governments may have a lower cost placed to assume the risk, since they probably hold a of risk-bearing than some private firms. Consider again portfolio of assets whose value is little correlated with the telecommunications concession mentioned above, local business conditions. Notice that what ultimately and suppose the concession was owned by a single risk- matters is not the portfolio of projects owned by the averse entrepreneur. The cost of bearing the demand foreign companies that are involved in the project; risk would probably be higher for a single risk-averse their business may be concentrated in one sector in entrepreneur than for the government and its taxpayers, just a few countries. Rather it is the portfolios of the and that cost would have to be passed on to consumers. ultimate investors-the individual shareholders or, in In this simplified example the choice of allocating the the case of public foreign investment, the taxpayers- risk to the government or the entrepreneur would that determine the costs of risk bearing. depend on a comparison between the incentive benefits associated with allocating demand risk to the entrepre- The Tradeoffbetween Risk Allocation Criteria neur and the lower cost of risk-bearing associated with allocating it to the government. As mentioned above, the benefits of allocating risks to This example is simplistic, however. Governments those who can best control the risky outcome must do not always have a lower cost of risk-bearing than sometimes be weighed against the benefits of allocat- private investors, and the relative costs of risk-bearing ing them to those who can bear them at least at cost. do not in general justify allocating risks to govern- Spreading a risk among a large number of sharehold- ments. The private sector is at least as capable as gov- ers or taxpayers may lower the costs of risk-bearing, ernments of spreading risks, and large companies can but allocating a risk to a small number of agents who have thousands, even millions of shareholders, most of have control over the risk may help ensure the success 9 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE of the project. Rather than diversifying risks com- evolve over time; the government may not be able to pletely, then, it usually pays to give managers and implement the ideal policies immediately even if it strategic investors significant stakes in the project. faces no political constraints. Such a government has a Private investors can better make this tradeoff than choice between privatization with significant govern- can governments. First, they can more easily choose ment risk-bearing and continued public ownership. between the benefits of spreading risks evenly among Faced with that choice it may justifiably prefer the many shareholders and the incentive sharpening second-best option of privatization without the full achieved by giving certain investors and managers transfer of commercial risks to private investors. large stakes in the project. Governments pass on risks Governments that have just embarked on a transi- to all their taxpayers, which tends to lead to an out- tion will also suffer from a lack of reputation, even if come at the risk-reducing extreme of the spectrum. they succeed in implementing good policies. Even if Private investors could choose such an extreme, but their laws and regulations are as good on paper as those they may prefer to sacrifice some diversification for of, say, the United States, investors will be considerably the sake of maintaining strong incentives to perform more wary about investing there. Governments in well. Second, private investors have stronger incen- countries in transition may have to provide explicit tives to choose the optimal combination of risk- undertakings-to allow convertibility or to compensate spreading and incentive-sharpening (Klein 1 996b). in case of expropriation, for example-where industrial country governments do not. In countries such as the Policy Transitions United States, investors may think it unnecessary to Yoltcv Transitions seek certain explicit project-specific guarantees, either Compared against the criteria for risk allocation dis- because the risks are negligible or because they are con- cussed here, many governments in the developing fident that they will be protected by the legal system world appear to bear too much risk when they priva- and the courts in case of problems. In countries that tize infrastructure. Their policies may nevertheless have reformed their policies only recently, investors represent improvements over those of the past. Under may want the government to assume these risks explic- traditional public ownership the government bears all itly in a contract. the commercial risk: privatization almost always trans- Investors may also doubt that the newly reformed fers some risk to the private sector. When the govern- government will maintain its good policies. If the gov- ment guarantees a private toll road sponsor 90 percent ernment does intend to do so, an argument can be of expected toll revenue, for example, it bears less risk made in favor of allocating some project risk to the that it would if it owned the road and bore 100 per- government (in addition to those political and regula- cent of the risk. Privatizing the road and providing tory risks that are directly under its control). Caution guarantees to the concessionaire may thus be better is warranted, however, since governments may mis- than having the government build and operate it judge the future course of their own, or their succes- without private participation-or not building the sors', policies. The best response may be to try to con- road at all. vince investors that they are too pessimistic by provid- Governments in developing countries moving ing them with all the information on which the gov- toward more market-oriented policies may find them- ernment bases its more optimistic assessment. selves unable to introduce all the reforms that would be required for privatization without government guarantees of commercial risks. They may, for exam- Some Guidelines for the Allocation of Certain ple, find it impossible politically to raise prices to Infrastructure Risks cover the risk-adjusted cost of capital or to raise taxes to pay direct instead of contingent subsidies to The principles of risk allocation outlined here can be investors. Moreover, the government's understanding applied to the risks that governments are often asked of what constitutes the optimal policy framework will to bear in infrastructure privatizations. In this sec- 10 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW tion, we consider the classes of risk that are the sub- On the one hand, these risks are clearly under the ject of three of the following chapters: political and government's control. On the other, it is sometimes regulatory risks (chapter 3), demand and cost risks desirable for the government to change laws in ways (chapter 4), and exchange and interest rate risks that adversely affect investment projects. It may be (chapter 5). beneficial to increase taxes to fund new and socially valuable public expenditure, for example, or to impose regulations to mitigate newly recognized envi- ronmental problems. In many cases, such as that of In chapter 3 Smith distinguishes three types of risks: new environmental regulation, the government can * Traditional political risks, which include the risks bear the risk and still change policy-it just needs to of expropriation, political violence, currency compensate firms for the policy change. In other inconvertibility, and currency nontransferability cases, however, compensation cannot be reconciled * Regulatory risks, which relate to the application with flexibility. If governments had to compensate and enforcement of regulatory rules (expressed in everyone for imposing higher taxes, for example, they laws, regulations, or contracts) at either the econo- could never increase their (net) revenue. mywide or the project-specific level The same issue arises when considering project- * Quasi-commercial risks, which pertain to the risks specific regulations, such as rules setting the maxi- of contractual nonperformance by government mum prices the infrastructure investor can charge for agencies in their capacity as suppliers to or pur- services. Contractual commitments by the govern- chasers from the private infrastructure project. ment to apply specified price-control rules shift risk to the government, but they make it harder to adapt the Traditionalpolitical risks. Smith accepts that tra- rules to changing circumstances; general principles ditional political risks should be borne by the gov- that require considerable discretion to implement cre- ernment. The risks of expropriation, currency ate flexibility but fail to shift much risk to the govern- inconvertibility, and currency nontransferability are ment. Smith leaves open the question of exactly which directly under the control of the government, and regulatory risks governments should bear, arguing for there is good reason to encourage the government a case-by-case approach and noting that countries not to create losses associated with any of these with better reputations for treating investors reason- three risks. The main issue is how the government ably can adopt more flexible rules. can credibly commit itself to bearing the risk-that is, to commit itself not to create the conditions that Quasi-commercial risks. Quasi-commercial risks would lead to the loss or to fully compensate arise when an investor contracts with public suppliers investors if it does. or purchasers that may renege on contractual commit- The case for government bearing the risks of polit- ments, often as a result of political pressure. When the ical violence is more subtle, since the government has public agency is a state-owned company with a legal less than complete control over it. Smith notes that identity different from the government's, it may well international law generally requires governments to be less creditworthy than the government itself, and exercise only due care and does not hold them strictly private investors will want the government to bear the liable for losses. Governments are thus required to risks of subsidiary's nonperformance. The degree to compensate investors only if they fail to take reason- which nonperformance by the government agency is a able steps to prevent the violence. political risk depends on the agency's degree of auton- omy from the government; it is government involve- Regulatory risks. Regulatory risks pose trickier ment in the operation of the agency that makes the questions. Should the government commit itself not risk quasi-commercial rather than an ordinary com- to change the laws and regulations affecting the mercial contracting risk. If the agency has little auton- investment project or to compensate in case it does? omy, government guarantees may be desirable. But 11 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE increasing the agency's autonomy by privatizing it is, long as certain minimum standards are met traffic on Smith notes, a preferable solution. some roads or bridges may vary little with increases in quality; if so, there is less to be gained by allocating Demand and Construction Cost Risks demand risk to the road owner. Engel, Fisher, and Galetovic observe that by In chapter 4 Engel, Fischer, and Galetovic discuss two changing the way they regulate the infrastructure ser- risks governments are often asked to bear, especially in vice governments can reduce the demand risk faced by toll road, bridge, or tunnel projects: the risks that concessionaires and thereby reduce the concession- profits will be higher or lower than their expected aires' demand for guarantees. The most common value as a result of variability in demand or construc- form of regulation of infrastructure such as roads, tion costs around their expected values. Both risks tunnels, bridges-and many other services, as well- tend to be critical. Construction costs can be very involves auctioning the right to operate the service for high, and must be incurred upfront. Demand over the a fixed period of time. The authors note that there is useful life of a toll road is inherently difficult to pre- an alternative to a fixed term, which is to allow the dict, especially when there is no historical data to term of the operating concession to vary with facilitate forecasts. In chapter 6 Lewis and Mody note demand. If demand is higher than expected, the con- that demand and construction risks were the two cession will be shorter; if demand is lower, the conces- most important for the government in the El sion will be longer. The method, which has been used Cortijo-EI Vino toll road in Columbia. in the United Kingdom for bridges, reduces the vari- Although the pressure for demand and construc- ance of the investors' profits: compared with the fixed- tion-cost guarantees may be strong, the rationale for term concession, profits are lower when demand is them in terms of the framework set out above is weak. strong but higher when demand is weak. The concessionaire usually has considerably more Engel, Fisher, and Galetovic propose an ingenious control than the government over construction costs, auction that differs from that used in the United even if it cannot control them completely. Moreover, Kingdom. Under their scheme the concession is if the concessionaire bears the construction risk, the awarded to the bidder seeking the lowest present value incentives to avoid white elephants are stronger. of revenue, calculated with a discount rate specified by Accordingly Engel, Fischer, and Galetovic conclude the government in advance. The concession ends that construction risk should not be borne by the gov- when the concessionaire's revenue reaches the present ernment. value it had sought. The concessionaire still bears Demand risks are more problematic. The govern- some demand risk-if demand is too low, revenue ment can influence some of the factors that affect may never reach the target value-but it bears much demand. The quality of the government's policies will less. Moreover, the investor still has an incentive to affect average incomes and therefore demand, for select only those projects that are likely to be finan- example, as will its decisions about whether to build cially attractive without government subsidies. other roads. Some roads would compete with the pri- vate toll road, thus lowering demand, while others would act as feeder roads, increasing demand for the private road. But the government is only one of many Exchange and interest rate risks are sometimes among influences, and government guarantees of demand the most important risks facing private infrastructure create incentive problems. If investors are shielded investors. If large infrastructure investments are fund- from demand risk, they have less reason to screen pro- ed by floating-rate loans or a series of short-term jects carefully with a view to investing only in those in fixed-rate loans, the projects' profits will be highly which expected demand is sufficient to justify the pro- sensitive to changes in the interest rates. Projects often ject. At the same time, however, the toll road operator also involve considerable foreign financing. If project may have little control over the demand risk either. As revenues are in local currency but the investors want 12 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW to earn foreign-currency profits, foreign investors will be blamed on the government, which allowed the suffer if the local currency depreciates. Ensuring that exchange rate to fall, or the firm, which left itself the right parties bear interest rate and exchange rate exposed by borrowing in foreign currencies. risk is thus important to the success of the project. In Second, in flexible exchange rate regimes, chapter 5 Mas addresses the question of whether it exchange rate guarantees may have undesirable as well might be appropriate for the government to accede to as desirable incentive effects on the government, since requests to take on these risks. they discourage governments from allowing their cur- The key argument in support of government bear- rencies to depreciate in the wake of a terms of trade ing these risks stems from an incentive effect. Private shock, for instance. In this case, government exchange investors have almost no control over the exchange rate guarantees would not necessarily encourage (or rate or prevailing interest rates, both of which are signal) good macroeconomic policy. The problem is affected by government actions. Macroeconomic poli- that interest rate and exchange rate guarantees fail to cies that lead to price stability and balanced budgets isolate that which is under government control will tend, for example, to reduce the volatility of the (macroeconomic policy) from that which is not (the exchange rate and the probability of a large deprecia- terms of trade, for example). tion. If governments bear interest and exchange rate Mas also notes problems related to the cost of risk- risks they have a financial incentive to adopt macro- bearing, noting that governments and the taxpayers economic policies that tend to prevent depreciation or who back them may already be exposed to the risks interest rate increases. associated with interest rate and exchange rate shocks. Because of this incentive effect the bearing of inter- An adverse terms of trade shock, for example, might est or exchange rate risks by the government may have a lead to both a depreciation and a decline in local useful signaling effect. Since the risk-bearing will be incomes, forcing the government to compensate expensive for governments that plan to adopt impru- investors just when its tax base had shrunk. Foreign dent macroeconomic policies, governments that choose investors would not face this problem and, contrary to to issue guarantees can be assumed to be more likely to a common recommendation, may be in the best posi- act reasonably (at least if the political decisionmakers tion to bear the risk. are concerned about the government's fiscal position). Risk-bearing by the government may thus signal good intentions at the same time as it provides the govern- Measuring and Budgeting for Risk ment with an incentive to carry them out. Mas argues cogently, however, that guarantees of Whichever risks a government does take on, it needs to these macroeconomic outcomes are unlikely to have net consider how it can measure them and incorporate them benefits, except perhaps during the early stages of policy in its accounts and budgets. Without good measurement reforms. The incentive and signaling benefits of and the incorporation of those measurements in exchange rate and interest rate guarantees are likely to be accounting and budgeting, governments will have diffi- limited and could create significant costs, for several rea- culty making good decisions about whether to assume sons. First, it is difficult to separate the effects on project risks. Moreover, they may court financial disaster. profitability of exchange or interest rates and business Instances in which public and private institutions decisions. Exchange rate depreciations, for example, will have lost money partly because management was not have a direct, easily measurable effect on interest pay- carefully monitoring the institutions' exposure to risk ments on foreign loans measured in terms of local-cur- are not hard to find. Many of the private savings and rency project revenues. But they may also affect the cost loans institutions in the United States first ran into of other inputs and demand for the service. These trouble because they had made long-term loans at fixed effects cannot easily be measured. Nor can responsibility interest rates that they had funded with short-term bor- for losses resulting from, say, exchange rate depreciations rowings. When tight monetary policy caused market always be easily assigned. Losses from depreciation could interest rates to rise steeply in the early 1980s, the sav- 13 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE ings and loans found their costs had jumped relative to example, presents this information in its statement of their revenues. The differences in the average maturity contingent liabilities, which includes table 1.1.5 of their assets and liabilities subjected them, that is, to interest rate risk. Better reporting of risk exposures Ca might have encouraged the institutions to take a less risky approach. At the same time better monitoring and A listing of guarantees and associated maximum possi- reporting by the U.S. government might have protected ble losses is helpful but has limitations. In particular, taxpayers from the losses they subsequently incurred it provides no information on the likelihood of losses. when the government bailed out many of the savings It reveals maximum possible losses, that is, but does and loans. The Latin American debt crisis of the early not indicate what losses the government should 1980s and the Mexican crisis of 1994 might also have expect. Where possible, then, it is useful to quantify been less severe had the governments involved had not only the maximum possible loss but the likeli- access to, and published, data on their financial posi- hood of losses and, therefore, the expected loss. tions and their vulnerability to interest rate, exchange Sometimes it is simple to estimate expected losses. rate, and other shocks. If the government guarantees a $1 million payment by In the area of infrastructure, too, prudence suggests one of its state-owned enterprises and there is a 10 that the government attempt to measure and control its percent chance of the enterprise defaulting (and a 90 exposure. At the simplest level this would require that it percent chance of full payment), the expected cost to know what guarantees it has issued and how much it the government of the guarantee is $100,000. In more might lose if the guarantees were called. A government realistic cases, the calculation of the expected cost is should also estimate what its expected losses are-that more difficult. There may be more than two relevant is, what it will most probably lose-and what the prob- possibilities, and the estimation of the probabilities abilities are of various greater losses. Where feasible may be extremely difficult. these estimates should be incorporated in the govern- Nevertheless, as Lewis and Mody show, the calcula- ment's accounts and its budgets, so that decisionmakers tion of expected losses is sometimes feasible using rela- and those who monitor them (taxpayers, voters, the tively straightforward techniques. The most tractable press, government debt holders) can more easily assess cases will be those in which the government has issued the financial effects of different decisions. Ultimately, a a large number of similar guarantees for many years government should have a consolidated picture of its and has recorded information on defaults. In these overall exposure to risks, taking into account the corre- cases the expected cost of the guarantees can be esti- lations between different risks. Once it has such a pic- mated in the same way as, say, car insurance premiums ture it can consider taking a more active role in manag- are calculated. The reforms the U.S. government ing that exposure through actions such as hedging. enacted with the Federal Credit Reform Act of 1990, Lewis and Mody's chapter provides a sophisticated which Lewis and Mody discuss, are informative. overview of the practice of risk management in gov- ernment. Some basic information is presented here for TABLE 1-1 Statement of contingent liabilities summary table readers less familiar with the steps governments can (millions of New Zealand dollars) take to measure and manage risks. Quantifiable As at As at contingent liabilities 31 May 1996 31 May 1997 Identifting and Listing Guarantees Guarantees and indemnities 493 536 Uncalled capital 1,752 2,248 The first and simplest step that governments can take Legal proceedings and disputes 1,317 971 to improve the monitoring and management of risks Other contingent liabilities 1,249 1,177 is to compile and publish a consolidated list of their Total quantifiable contingent liabilities and the maximum amounts they contingent liabilities 4,811 4,932 stand to lose. The New Zealand government, for Source. Government of New Zealand. 14 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW The cost of some unique guarantees can also be ment's accounts and budgets. In the accounts the pre- estimated simply. Full credit guarantees in which the sent value of the expected future costs of guarantees government guarantees the repayment of a loan to issued in the current year can be recorded as an another party can, for example, be valued by looking expense in that year. In the budget the legislature can at the difference between the interest rate charged on authorize, alongside the cash expenditures, a certain guaranteed and nonguaranteed loans. If a firm is pay- value of guarantees to be issued, where the value is not ing 15 percent interest on its nonguaranteed debt and the maximum potential loss but the expected cost. 10 percent on loans guaranteed by the government, (The budget might also limit the maximum possible the annual value of the guarantee is 5 percent of the exposure.) If guarantees are accounted for and budget- amount borrowed. (For more on this type of guaran- ed in this way, the government will be even less likely tee valuation see Mody and Patro 1995 as well as to prefer them to less costly cash subsidies. This is chapter 6 in this volume.) likely to improve the allocation of project risk as well The techniques developed in the past twenty-five as project selection and contract design. Budgeting for years to value financial derivatives (such as options, expected losses will also create "fiscal space" in the futures, and swaps) can also be used to value guaran- event that past guarantees are called. tees and contingent liabilities, including more compli- Most governments' budgets and accounts are cash cated ones. Extending a credit guarantee, for example, based. The budget authorizes the government to incur is equivalent to the government's selling a put option certain cash expenditures; the accounts show how to the lender, which gives the lender the right to put much cash the government has received and spent. the loan to the government. The valuation of other Noncash items, such as the depreciation of assets dur- types of guarantees is much more difficult, requiring ing the year and revenues earned but not received in the skills of financial specialists, and the feasibility of cash (such as taxes owed but not yet paid), do not timely, reliable, and cost-effective valuation has not appear in the budget or the accounts. Such govern- yet been widely tested. But the possibilities are not ments do not report their balance sheets or net worth. merely theoretical: as Lewis and Mody show, guaran- While it is possible to note guarantees and other tees have already been valued using option-pricing noncash items in what are essentially cash-based bud- techniques in both Colombia and the United States. gets and accounts, fully incorporating them requires a Valuing the government's guarantees and other switch away from cash-based systems. With standard contingent liabilities-and not simply noting maxi- accrual accounts and budgets, most noncash expendi- mum exposure-has important advantages. By calcu- tures show up in the government's budget and in its lating the expected cost of the government's guaran- operating statement, and the government has no fiscal tees, the government and its observers can more easily incentive to prefer these noncash expenditures to cash compare guarantees with cash subsidies. When guar- expenditure. But although standard accrual account- antees are not valued a government may prefer to pro- ing discloses guarantees (as in the table from the New vide a guarantee instead of a subsidy, even when the Zealand government's accounts shown above), it guarantee is more costly than the subsidy, because the records them as expenses only if the loss is considered costs of the guarantee are hidden and may be borne probable and can be quantified (Afterman 1997). by a future administration. When guarantees are val- From an economic point of view, the distinction ued, decisions are more likely to be made on the basis between probable and improbable losses is not always of real rather than apparent costs and benefits. useful; a 10 percent chance of losing $1 million is worse than a 90 percent chance of losing $100,000. Incorporating Expected Losses in Accounts More useful is an estimation of the present value of Incorporatsng Losses in Accounts the expected loss arising from the contingent liability. Present-value accounting, as described by Lewis and If expected losses can be reliably calculated, the next Mody in chapter 6, therefore attempts systematically step is to incorporate the estimates in the govern- to report the expected present value of contracts 15 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE entered into. Under a system of present-value budget- first organizations to report value at risk, the princi- ing a government has no fiscal incentive to issue guar- ples behind such reporting apply to all organizations. antees instead of giving subsidies of equivalent value, Governments should be thinking about how they can because both show up as expenditures affecting the apply the principles to their own operations and what deficit and both require appropriation by the useful estimates they might be able to publish.6 legislature. Although most governments still work with cash- Taking a Governmentvide Approach to based budgets and accounts, several have either RikM moved, or are in the process of moving, toward the accrual system and a greater emphasis on present val- Expected losses can be measured individually and ues. Iceland and New Zealand have implemented then aggregated; the government's total expected loss integrated accrual accounts and budgets, while from issuing several guarantees is simply the sum of Australia, Canada, Sweden, the United Kingdom, and the expected losses associated with each guarantee. the United States have taken important steps in that Risks, however, cannot normally be estimated individ- direction. Because of its benefits accrual accounting ually and then summed: the total risk depends also on with more widespread reporting of present values may the relationships between the individual risks. in the future become the norm in government. Because what matters for a government is not the risk relating to any one guarantee but the riskiness of Measuring Risk As Well As Expected Losses its portfolio of assets and liabilities, value-at-risk reporting is likely to be most useful when done for the Estimating, reporting, and budgeting for expected government as a whole. Measuring and managing the losses is important, but expected costs do not tell the risks associated with infrastructure privatization there- government and those that monitor it everything they fore requires assessment of the riskiness of all of the need to know. Just as guarantees with the same maxi- government's operations. Exchange rate guarantees, mum exposure differ significantly if the expected loss for instance, are likely to be more risky if the govern- differs, so too may guarantees with the same expected ment also has net debt denominated in foreign cur- but different maximum losses. The whole range of rencies, since a depreciation of the local currency may possible outcomes-that is, risk in the sense of volatil- simultaneously increase debt service payments and ity-matters. In the terminology of Lewis and Mody, trigger payments under the guarantee.7 "unexpected" as well as expected losses need to be Portfoliowide risk measurement is desirable for considered. another reason as well: only by considering its total Governments should therefore develop systems for portfolio of risks can a government determine which summarizing and reporting the major risks-as well as risks are most important. How important are risks the expected costs-they face. In some cases it is con- associated with infrastructure privatization relative to venient to describe the overall risk with a single num- the risks associated with government debt, pensions, ber. The approach that has been adopted by many the banking system, and debt owned by city and banks is to report the largest loss that can be incurred provincial governments? What proportion of the gov- with a probability greater than, say, 1 percent or 5 ernment's total risk measurement and risk manage- percent. The value, known as value at risk, is in ment effort should be devoted to infrastructure? Are essence an application of statistical theory to the the infrastructure guarantees small enough relative to description of assets and liabilities. For example, a the government's total assets and liabilities, and suffi- bank may report that its daily value at risk at the 1- ciently uncorrelated with them, that the government percent level is $10 million, meaning that there is can reasonably consider only their expected costs and only a 1-percent chance, under normal market condi- ignore the risks they involve? These are critical ques- tions, that it will lose more than $10 million in the tions that can be answered only after assessment of the next day (Jorion 1997). Although banks have been the government's overall portfolio. 16 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW The creation of consolidated accrual accounts is a Moreover, because of transaction costs or government step toward portfoliowide risk monitoring. Combined restrictions, they may have limited opportunities to with other information on likely future revenues and hedge against government risk. In that case govern- expenditures, the balance sheet can provide an indica- mental risk management may be desirable. The aim of tion of the government's vulnerability to exchange such management would be to achieve a level of risk rate, interest rate, and other shocks. It can also indi- the government finds acceptable, given its citizens' cate the relative importance of monitoring and man- risk preferences, at the lowest possible cost. Options aging infrastructure guarantees on the one hand and for reducing risk include the following:8 debt on the other. * Diversifying the government's financial assets and Sophisticated portfoliowide value-at-risk analysis liabilities (by placing ceilings on the possible guar- would require more, however, than construction of a antee exposure to any one sector, for example) consolidated government balance sheet. It would * Selling assets and paying down debt, thus reduc- require numerical estimates of the volatility of the val- ing the leverage of the portfolio ues of the assets and liabilities on the balance sheet * Setting aside reserves to cover unexpected losses, as and the correlations between them and a considera- well as expected losses tion of the "near" assets and liabilities that do not * Hedging with derivatives, such as options, futures, appear on the balance sheet. The power to tax may be and swaps the most significant determinant of the government's * Helping taxpayers diversify their financial assets financial position; estimates of the value of this "near" and liabilities (by removing capital controls that asset and its exposure to various risks would need to discourage investment in foreign stocks and be estimated. Likewise, future spending programs may bonds, for example) so as to reduce the riskiness of not constitute liabilities in an accounting sense when the government's tax base. there is no legal obligation to incur the expenditure, but in practice the government's financial health will be sensitive to the variability of expenditure associated Conclusion with changes in economic variables. Value at risk is thus not something that most governments will be Whether infrastructure privatization will realize its able to report any time soon. potential depends on how governments allocate the risks facing the privatized business. Government can Risk Management increase the benefits of privatization by assuming risks it can control itself (convertibility risk, for example), When, and only when, a government has information but it should normally avoid bearing other risks. That on the risks to which its total portfolio is exposed, it is way, investors face strong incentives to select projects in a good position to consider managing its portfolio well and to run those that they do select efficiently. In to reduce those risks. Whether it should, in fact, act to many infrastructure privatizations, however, govern- reduce risk-and if so to what extent-is perhaps an ments have assumed risks that would be better borne open question, analogous to the question of whether by investors, both because the investors have been firms should attempt to reduce the variance of share- understandably wary of taking on the considerable holders' returns or simply maximize those returns (see risks involved and because governments have been chapter 5). That citizens, like shareholders, are usually able to offer guarantees without incurring any imme- risk averse is not in doubt; the question is whether the diate cash costs. A government can thus take two steps government should manage risk on their behalf or to improve the environment for risk allocation. It can simply publicize its risk exposure and permit taxpayers reduce the extent of the risks investors face, by pursu- to diversify and hedge their portfolios in ways that ing stable macroeconomic policies, disclosing infor- give them the risk exposure they want. In practice, cit- mation, implementing good laws and regulations, and izens often lack sophistication in considering risk. liberalizing financial markets. And it can improve the 17 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE way it measures, budgets, and accounts for the guar- face, however, rather than the risks faced by the governments antees it does give, so that the costs and risks are clear of the countries, the issue addressed by Lewis and Mody. at the time the guarantees are issued-not only when the government must subsequently pay up. References Notes Afterman, Allan B. 1997. GAAP Practice Manual Vol. 1. Warren, Gorham, and Lamont. 1. See http://www.shcp.gob.mx/english/. Asian Development Bank. 1997a. "Facilitating Private 2. Notice that in this case outcomes are better even if Sector Participation in Infrastructure Development the firm's actions to increase quality increase expected Voluntary Principles and Collaborative Initiatives." demand, and therefore expected profits, without reducing Background paper for the Fourth APEC Finance the volatility of demand. Strictly speaking, what matters is Ministers Meeting, April 4-6. control over the risky outcome, not control of the risk per se. . 1997b. "Promoting Financial and Capital Market 3. Note that the relatively low borrowing costs of gov- Development: Voluntary Principles and Collaborative ernments do not by themselves imply a lower cost of risk- Initiatives." Background paper for the Fourth APEC bearing, since they reflect an unremunerated credit guaran- Finance Ministers Meeting, April 4-6. tee given by taxpayers (Klein 1996b) Claessens, Stijn. 1992. "How Can Developing Countries 4. In principle, both local consumers and the govern- Hedge Their Bets?" Finance and Development, 29: ment might reduce their exposure to the fortunes of the local 13-15. economy in various ways (buying foreign assets, selling short . 1993. Risk Management in Developing Countries. local assets). In practice, governments typically do not do so World Bank Technical Paper 235. World Bank, and sometimes prevent their citizens from doing so by impos- Washington D.C. ing capital controls. Even in wealthy countries with highly Claessens, Stijn, and Y Qian. 1991. "Risk Management in developed and open capital markets, however, local investors Sub-Saharan Africa." In I. Hussain and J. Underwood, seem not to take full advantage of the opportunities for inter- eds., African External Finance in the 1990s. Washington national diversification (see French and Poterba 1991). D.C.: World Bank. 5. Available at http:l/www.treasury.govt.nz. French, Kenneth R., and James M. Poterba. 1991. "Investor 6. For more on the analysis of value at risk, see Diversification and International Equity Markets." chapter 5, Jorion (1 997); 'J.P. Morgan's "RiskMetrics" docu- American Economic Review 81(2): 222-26. mientation, which is available at http://www.jpmorgan.com; G6mez-Ibafiez, Jose A., and John R. Meyer. 1993. Going and the various reports available at http:/Jwww. Private: The International Experience with Transport contingencyanalysis.com. Privatization. Washington, D.C: Brookings Institution. 7. In the extreme case of a guarantee of a risk that is International Finance Corporation. 1996. Financing Private negatively correlated with the value of the government's Infrastructure: Lessons of Experience. Washington, D.C. portfolio, assessment of risk in isolation would lead the gov- Jorion, Philippe. 1997. Value at Risk: The New Benchmark ernment to think the guarantee created risL, when in fact it for Controlling Market Risk. Chicago: Irwin. reduced it. That is, because the value of this guarantee tends Juan, Ellis. 1996. "Privatizing Airports-Options and Case to fall when the value of the rest of the government's portfo- Studies." Public Policy for the Private Sector: Special lio rises, and vice versa, the total volatility of the govern- Edition, Infrastructure. Washington D.C.: World Bank. ment's wealth would be gieater without the guarantee. Klein, Michael. 1996a. "Managing Guarantee Programs in 8. A useful short summary of asset-liability manage- Support of Infrastructure Investments." World Bank, ment can be found in Claessens (1992). Claessens (1993) Private Sector Development Department, Washington, and Masuoka (1990) provide longer summaries. Claessens D.C. and Qian (1991) apply the techniques to African countries. . 1996b. "Risk, Taxpayers, and Role of Government All of these papers address the risks that countries as a whole in Project Finance." Policy Research Working Paper 18 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE: AN OVERVIEW 1688. World Bank, Private Sector Development United Kingdom. 1995. Private Opportunityf Public Benefit: Department, Washington, D.C. Progressing the Private Finance Initiative. London. Masuoka, Toshiya. 1990. "Asset and Liability Management in World Bank. 1 997a. Global Development Finance 1997. Vol. the Developing Countries: Modern Financial Techniques: 1. Washington, D.C.: World Bank. A Primer." Policy, Research, and External Affairs Working . 1 997b. World Development Report 1997: The State in Paper WPS 454. World Bank, Washington, D.C. a Changing World. New York: Oxford University Press. Mody, Ashoka, and Dilip Patro. 1995. Methods of Loan . 1997c. Private Capital Flows to Developing Guarantee Valuation and Accounting. CFS Discussion Countries: The Road to Financial Integration. New York: Paper Series, 116. Washington, D.C.: World Bank. Oxford University Press. 19 I 2 Government Support to Private Infrastructure Projects in Emerging Markets Mansoor Dailami and Michael Klein ABSTRACT Driven by fiscal austerity and disenchantment with nomic policy matters because it affects the credibility the performance of state-provided infrastructure ser- of a price regime and trust in the convertibility of the vices, more than 100 governments have turned to the currency, which is essential for foreign investors. private sector to build, operate, finance, or own infra- Privatization of assets without government guaran- structure in sectors such as power, gas, transport, tees or other forms of financial support is possible, telecommunications, and water. Private capital flows even when governments are politically unable to raise to developing countries are increasing rapidly, and prices, since investors can achieve the returns they some 15 percent of infrastructure investment is now demand by discounting the value of the assets they are funded by private capital in emerging markets. purchasing. For new investments (greenfield projects), Relative to needs, however, private investment in however, this is not possible. If prices have been set too infrastructure is progressing slowly. The reasons why low and the government is not willing to raise them, it are clear: governments are reluctant to raise consumer must provide the investor with some form of financial prices to cost-covering levels, while investors, mindful support, such as guarantees. Guarantees and other of historical experience, fear that governments may forms of subsidy can facilitate worthwhile projects that renege on promises to maintain adequate prices over would otherwise not go ahead. But government guar- the long haul. Investors, therefore, ask for government antees shift costs from consumers to taxpayers, who support in the form of grants, preferential tax treat- subsidize the price consumers pay. Much of that sub- ment, debt or equity contributions, or guarantees. All sidy is hidden, since the government does not record of these forms of support are subsidies. They differ in the guarantee in its fiscal accounts. Moreover, taxpay- the way in which they allocate risks between private ers provide unremunerated credit insurance to the gov- investors and government. When private parties ernment, since the government borrows based on its assume risks that they can manage better than the ability to tax citizens in case the project fails, rather public sector, efficiency gains will be largest. than on the strength of the project itself. If citizens are When governments put in place good policies-in to reap the full benefits of private participation in particular, cost-covering prices and credible commit- infrastructure, governments must correct their policies. ments to stick to them-investors are willing to invest In particular, prices need to be raised to cost-covering without special government support. Good macroeco- levels and private investors need to assume risk. 21 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE The Growth of Private Investment FIGURE 2.1 in Infrastructure Public sector borrowing requirement Percent of GDP Following the debt crisis of the early 1980s develop- 0 ing countries significantly restricted public borrowing. The combined public sector borrowing requirement of all developing economies shrank from 6 percent of 2 GDP in 1982 to 1 percent in 1993 (figure 2.1). While public funding has been reduced, infra- 3/ structure investment requirements remain high. In 1994 the World Bank estimated them at $200 billion -4 a year for developing countries. Since then other World Bank studies have increased these estimates. In -5 East Asia and Latin America alone average annual investment requirements through 2005 have been -6 estimated at $150 and $60 billion, respectively. 1980 1982 1984 1986 1988 1990 1992 . a a ] X X ~~Source: World Bank 1997a. Investment requirements tend to be dominated by the transport sector, followed by energy, telecommunica- have turned to private solutions for financing and pro- tions, and water. Required investments often reflect viding telecommunications, energy, transport, and excess demand for services. That is, consumers would water services (World Bank 1994). The trendsetters be willing to pay more for services, but prices are set were Chile, the United Kingdom, and New Zealand. at levels that are too low to attract suppliers. Deregulation of many sectors-including telecommu- (Telecommunications may be an exception, as con- nications, airlines, independent power generation, sumer prices exceed cost-covering levels in several natural gas production and transmission, and freight countries, albeit sometimes because excise taxes are traffic by road and rail-began even earlier in the high.) United States in the late 1970s. During the 1990s the Driven by fiscal constraints and growing disen- dual trend toward private involvement in infrastruc- chantment with the performance of state-provided ture and deregulation has caught on in almost all infrastructure services, more and more governments countries. TABLE 2.1 Net long-term resource flows to developing countries 1990 1996 In billions As share In billions As share of dollars of total of dollars of total Total flows 100.6 100 284.6 100 Sources Official development finance 56.3 56 40.8 14 Private flows 44.4 44 243.8 86 Recipients Public sector 62.8 62 84.8 30 Private sector 37.8 38 199.7 70 Foreign direct investment (24.5) (24) (109.5) (38) Portfolio equity flows (3.2) (3) (45.7) (16) Nonguaranteed debt (10.1) (10) (44.5) (16) Bond (0.1) (0.1) (20.8) (7) Source: World Bank 1997a. 22 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS Private markets are responding with vigor.' From are invested from local sources in private infrastruc- 1990 to 1996 total net resource flows to developing ture projects, so that total private investment may cur- countries rose from $101 to $285 billion a year (table rently account for about 15 percent of a total estimat- 2.1). Private flows rose from $44 billion to $244 bil- ed investment requirement of $200 billion a year. lion, while official development finance dropped from Almost half of all private cross-border infrastruc- $56 to $41 billion. Cross-border flows dominate ture finance appears to have been invested in East Asia, infrastructure finance, even in countries with very and more than a third was invested in Latin America high national saving rates, partly because of the bene- (table 2.2 and figure 2.2). Power projects have attract- fits investors gain from diversification but partly ed the highest share of investment, accounting for because of the underdevelopment of local capital mar- more than 40 percent of the total, followed by kets in these countries. telecommunications and transport (figure 2.3). Increasingly, private capital has funded private Between 1990 and 1994 private infrastructure projects and firms rather than public expenditures. finance to developing countries grew at an annual Between 1990 and 1996 public sector borrowing average rate of 67 percent, reflecting the low base from private sources rose from $63 billion to only $85 from which it started. Since 1994 growth has averaged billion, barely offsetting the drop in official develop- 14 percent a year, well below the 19 percent growth ment finance. In contrast, private capital (debt and rate of total private capital flows to developing coun- equity) to private recipients rose from $38 billion to tries (figure 2.4). (See also annex tables A2.1-A2.4.) $200 billion. Total infrastructure financing raised by developing countries rose from less than $1 billion in 1988 to Why Infrastructure Is Different more than $27 billion in 1996. Finance for private infrastructure rose from virtually nothing in 1988 to To understand why private financing of infrastructure more than $20 billion in 1996 (table 2.2). Although has not kept pace with overall financial flows to pri- the data on infrastructure capital flows are not strictly vate entities it is necessary to recognize how infra- comparable with the data on capital flows, cross-bor- structure differs from other industries. der private infrastructure finance appears to account First, infrastructure services are often considered for about 10 percent of all private-to-private cross- essential by consumers, and they are frequently pro- border capital flows. About half of cross-border flows vided by monopolists. Together these factors increase TABLE 2.2 Private cross-border financial flows to infrastructure (billions of U.S. dollars) 1988 1989 1990 1991 1992 1993 1994 1995 1996 Total 0.1 0.9 2.0 3.5 5.8 12.3 15.7 15.6 20.3 Loans 0.1 0.8 1.4 0.1 1.5 6.3 6.0 11.1 7.7 Bonds 0 0.2 0.5 0.7 1.1 3.9 5.8 3.3 7.2 Equity 0 0 0.1 2.6 3.1 2.1 3.9 1.3 5.4 Latin America and the Caribbean 0 0.2 0.3 3.1 3.6 4.7 6.6 2.1 7.8 Loans 0 0 0 0.02 0.2 0.3 1.6 0.7 0.7 Bonds 0 0.2 0.3 0.6 1.0 3.3 3.7 1.4 4.4 Equity 0 0 0.1 2.5 2.4 1.1 1.3 0 2.8 East Asia and Pacific 0.1 0.8 1.5 0.4 2.0 5.7 6.8 8.8 9.3 Loans 0.1 0.8 1.3 0.0s 1.2 4.6 3.4 6.1 4.9 Bonds 0 0 0.3 0.2 0.2 0.3 2.1 1.7 2.4 Equity 0 0 0.02 0.1 0.6 0.8 1.3 1.0 2.0 Source: World Bank 1997a. 23 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE FIGURE 2.2 Cumulative private sector borrowing for infrastructure, 1985-95 Billions of dollars 80 0 *Private 60 40 20 Sub-Saharan East Asia South Asia Europe and Latin America Middle East Africa and the Pacific Central Asia and the and North Africa Caribbean Source: Euromoney Loanware and Bondware; World Bank staff estimates. FIGURE 2.3 political sensitivity to the prices charged. Pressure Sectoral composition of infrastructure financing from consumers to keep prices low makes it politically in developing countries difficult for governments to maintain prices that cover 1986-90 costs. Indeed, the World Bank (1994) estimated that Telecommunications user fees fell far short of costs in gas, electricity, and Other water. i 9frastructure - Second, infrastructure projects typically require large sunk investments that take ten to thirty years to recoup. Over such long periods of time investors are Power exposed to serious risks, in particular the risk that public authorities will not honor their agreements on tariff policy and payments to investors (Klein and Roger 1994). Once investors are committed to pro- Transpo jects-and can pull out only by taking a huge loss- 33% governments may be tempted to lower prices or not raise them as agreed. Investors thus risk being the 1991-95 victims of what has been called the "obsolescent communications bargain." 30% j | | _ These factors help explain the familiar privatiza- tion-nationalization cycle that has been observed repeatedly (figure 2.5). Private entrepreneurs may ini- Power tially develop infrastructure-building the first electric- l44% | ity networks, for example.2 As these networks expand Othe l _toward territories operated by other entrepreneurs, infras~tructue l-companies merge with or acquire their neighbors, creat- 13% ing larger, consolidated firms. These new firms are per- Transport ceived as possessing significant monopoly power, and 13% the services they provide-once considered luxuries- Source: World Bank 1994. are now considered essential, creating pressure for 24 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS FIGURE 2.4 FIGURE 2.5 Net private flows to infrastructure, 1990-96 The privatization-nationalization cycle Billions of dollars 200Inta to0 private pro s s uts, provision Total net flows en 150 to privatepto prvlemtmae neurs in2r e 1990 1991 1992 1993 1994 1995 1996 subsidyicus Consfidatio oe Wrld BankPublic ofcatfee alrease of firms takeovrrand ervice 100/ e ct / monopoly regulation.Regulation,inturn,reduce Des o prices0and profitability,fwhich discourages maintenance Source: Gomezneand Meyeriency f 1 an ae Total net flows qa to private infrastructuc project 4 sap the fnnsotest-ndfmolgnte vlPing countries are financed with significaublic nte goenett usdiei.Tevr vilblt fsb m untsioffoeigs aitl tpical finaningmi 1990 1991 1992 1993 1994 1995 1996 i s) and 6 p d i \ / 6 ~~Withdrawal Sorcen World Batrk 1997a. p s comeri of capital b i tinadteccen backup f and services a monopoly regulation. Regulation, in turn, reduces prices and profitability, which discourages maintenance Source d Gomez-Ibanez and Meyer 1993. and new investment. In the face of declining qualiny and a slowdown in the industry's growth, the govern- The heavy foreign fi rcix tng of infrastructure cre- ment nationalizes the firm. Low prices and inefficiency ates additional risks. Most infrastructure projects in sap the finances of the state-owned firm, obliging the developing countries are financed with significant government to subsidizte tTo very availabiliy of sub- amounts of foreign capital. A s ypical financing mix sidies, however, encourages more inefficiency. consists of 2040 percent equity (provided by project Eventually, concerns about fiscal subsidies and ineffi- promoters) and 60th 0 percent debt, in the form of ciency create pressure for price increases and privatiza- syndicated commercial bank loans, bond issues, tion-and the cycle begins again. bridge and backup facilities, and multilateral and Bnecause of the problem of sunk costs, and the his- export credit agency loans and guarantees. Exposur. torical experience of the "obsolescent bargain," to currency risk, which is a relatively minor concern investors are ypically unwilling to make investments for foreign investors in export-oriented manufacturing without adequate, frequently complex, contractual industries, is a critical feature of infrastructure project protection (Dasgupta and Sengupta 1993; Edlin and investmcnt. Project revenues are often generated in Reichelstein 1996). The negotiation of such contracts local currencies, while servicing of foreign debt and is time consuming and costly, however, and even the equity involves payment in foreign currency. best contracts cannot fully protect investors against the Fluctuations in the exchange rate of the domestic cur- efforts of a cletermined government. Enforceabilit, of rency, as well as capital controls limiting currency these contracts is essential, but it is difficult to achieve. convertibility and transferability, create risk for for- Investors are continually faced with the possibility of eign investors and financiers. changing contractual agreements or failure by the gov- While prospects for currency convertibility and ernment to implement tariff adjustments because of transferability have improved in many developing political considerations. Even if arbitration and settle- countries with the liberalization of their capital ment of disputes in a third country are agreed on in accounts and the surge in foreign capital inflows, the advance-such as in the case of the Enron-Dahbol scope for exchange rate hedging and risk management power project in India-such procedures can be time through the use of forward markets or derivatives consuming and can add to the cost of the project. remains limited. With the exception of Brazil, 25 DEALING WITH PUBLiC RISK IN PRIVATE INFRASTRUCTURE Malaysia, Mexico, and Thailand, where currency swap would be able to convert into hard currency its peso and forward markets have grown in the past two revenues from gas deliveries to state-owned Gas del years, foreign exchange markets in developing coun- Estado. In 1982 Argentina's foreign exchange reserves tries suffer from a lack of instruments and liquidity. were low because of the conflict with the United The case of the Argentine private natural gas Kingdom, and the government would have had trou- transport company, COGASCO, illustrates several of ble honoring its convertibility guarantee. Gas del these problems. COGASCO started operating in Estado then reviewed the contract with COGASCO 1981, with a guarantee from the central bank that it and claimed breach of contract, complaining that TABLE 2.3 Types of sovereign or supranational support for private infrastructure projects Multilateral banks Multilateral banks and export and export credit Government Informal credit agency agency debt guarantees agreementsa guarantees Country and Honduras: India: Mexico: Peru: project Electricidad de Cortes Dabhol 695-MW Mexico City Aguaytia 145-MW gas- S. De R.L. de C.V power plant; Toluca Toll Road fired power plant (Elcosa I) 60-MW oil combined cycle; fired power plant; imported liquefied 15-years PPA natural gas (LPG)/oil distillate; 20-year PPA with Maharashtra State Electricity Board; tariff 2.4 rupees ($.126) per KWh Project cost $70 million $922 million $313 million $235 million Date financial 1994 1995 February 1992 October 1996 closure Example by IFC: 12-year counter- Concession OPIC: mechanism $10.5 million senior guarantee from the guarantees $60 m political risk debt (LIBOR + 375 government of India traffic volumes by guarantee bps, 12-year maturity) for tariff-payments vehicle category, by the Maharashtra if traffic volumes FMO: State Electricity fell short of (Dutch)$10 milion senior Board; and term- amounts specified debt (LIBOR + 375 bps, ination guarantee contract. 12-year maturity) (capped at $300 Concessionaire million) entitled to request IFC B: an extension of the $10 million loan, concession term to 8-year maturity permit recovery of its investments. IFC: $3.5 million subordinated debt FMO: (Dutch) $1.0 million subordinated debt a. Informal agreements include comfort letters, side agreements, nonbinding tariff increases, and other similar agreements. 26 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS COGASCO had found a more efficient way to run a parent company went bankrupt and foreign invest- liquid petroleum gas extraction plant than foreseen in ment in the gas sector ground to a halt. the contract. The dispute meant that COGASCO was Because of this kind of risk, investors require high not paid, mooting the issue of currency convertibility. ex ante rates of return. In many cases real rates of Because the investor's costs were sunk it had little return on equity exceed 20 percent (see annex table leverage with the government and the government A2.5). This often results in prices that are higher than was unable to renege on its commitment. The dispute they were before privatization, when the real cost of lasted until the late 1 980s, when COGASCO and its capital was not taken into account. Government Government debt Multilateral Preferential equity (senior and equity Government tax participation subordinated) participation grants treatment Malaysia: Pakistan: Philippines: Brazil: Chile: Kuala Lumpur Rousch 412-MW Pagbilao 735-MW Linha Amerala 450-MW Empresa Sepang Airport power plant CCPP power plant, coal (10-year, 15 km, Electrica Pangue residual fuel oil; fired, 25-year PPA six-lane road) 30-year PPA with with National Power Water and Power Corporation Development Authority $3,924 million $507 million $933 million $174 million $465 million 1993 1996 1993 June 1996 1993 $390 million in $40 million standby IFC: $60 million $112 million grant $10 million in equity provided loan by National ADB: $40 million from the Rio de deferred tax duties by the government Development Finance CDC: $35 million Janeiro municipal of Malaysia Corp. (NDFC); $140 government million subordinated debt channeled to the Pakistan Fund from the World Bank ($70 million) and JEXIM ($70 million) 27 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Providing Financial Support to Attract Private mechanisms and render arbitral awards enforceable. Investors In some cases counterparties may lack the cash flow with which to pay investors. Investors thus often To render projects attractive to investors despite these seek additional assurances that any compensation due risks, governments have to raise user fees or provide them under the terms of their contract will actually be special financial support to projects. Whichever route paid. For example, the central governments may be they choose, they need to provide credible assurances asked to provide assurances that a publicly owned elec- to investors that sensible binding obligations (the tric utility will honor its contracts with the private gen- "rules of the game") will be honored. erating plants from which it buys power. Investors may Governments use an array of mechanisms to pro- also seek guarantees that their local currency earnings vide financial support to private infrastructure pro- will be convertible and transferable out of the country. jects (table 2.3).3 Some of these mechanisms, includ- In sum, infrastructure investors require special ing preferential tax treatment, grants, and equity or assurances that money due to them will be paid when subordinated debt contributions for which govern- due, in the currency they require. In this sense, all ments do not expect commercial returns, directly forms of government support ultimately amount to enhance project cash flow. In contrast, guarantees are cash flow support to a project and have a significant targeted at particular risks, such as the risk that a fiscal impact. state-owned party will renege on an obligation. The government's obligations to provide support Support through Government Guarantees can be defined in laws, decrees, statutes, licenses, con- cessions, contracts, or other legally binding docu- Governments often provide financial support by ments. Most countries have also signed some of the means of guarantees (box 2.1 and table 2.4). Central more than 1,200 bilateral investment treaties that governments often guarantee the performance of sub- define investor rights. sovereign entities, including public enterprises and Investors and their counterparties normally agree on provincial or municipal governments.4 suitable methods for dispute resolution. If local courts Through central government guarantees, project are not credible, the parties can agree to international risks, such as the ability of a public utility to pay its arbitration. Most countries have agreed to international private suppliers, can be transformed into countries conventions, which establish appropriate arbitration risk. Countries can reduce their exposure by replacing Box 2.1 Government guarantees in OECD countries Governments throughout the world provide guarantees years several industrial countries have suffered large losses to private investors in a variety of activities. Prominent under some of their guarantee programs, including deposit among such guarantees are deposit insurance for bank insurance and export credits. During the 1980s OECD depositors and pension or social security insurance. export credit agencies incurred losses equivalent to about Guarantees for housing, agriculture, students, exports, 20 percent of new business, while collecting premiums of and public corporations dominate the picture in OECD only 3 percent. Most of the export credit losses were on countries; little is known about the make-up of guarantee medium- and longer-term credit. This experience prompt- exposure in developing countries. Even in OECD coun- ed a change in guarantee management procedures. The tries information on guarantee exposure is sketchy. Data United States has instituted more transparent accounting suggest that total guarantee exposure may amount to principles for its guarantee operations under the 1991 15-20 percent of GDP, or more than a quarter of gross Credit Reform Act. The experience of export guarantee debt. This does not, of course, capture implicit guaran- schemes is relevant for governments considering guaran- tees, under which government may feel obliged to bail teeing long-term infrastructure investment, as risks are out failing firms or banks or help uninsured citizens in similar (medium- to long-term country risk), although the need (in the wake of natural disasters, for example). risk in infrastructure investment may be higher because of Guarantee programs can provide valuable support for the risk of regulatory failure or creeping expropriation for private economic activity. But they can be costly: in recent firms with immobile investments, such as power plants. 28 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS TABLE 2.4 Types of government guarantees in private infrastructure projects Type of guarantee Projects Contractual obligations of government entities * Guarantee of off-take in power projects Birecik Hydro Power Plant, Turkey Electricidad de Cores, Hungary Paguthan & Dabhol Power Plants, India Mt. Aop Geothermal Plant, Philippines . Guarantee of fuel supply in power projects Termopaipa Power Plant, Colombia Lal Pir Power, Pakistan Policy/political risk * Guarantee of currency convertibility Lal Pir Power, Pakistan and transferability * Guarantee in case of changes of law or regulatory Rousch Power, Pakistan regime lzmit Su Water Treatment Plant and Pipeline, Turkey Financial market disruption/fluctuations * Guarantee of interest rate North-South Expressway, Malaysia * Guarantee of exchange rate North-South Expressway, Malaysia • Debt guarantee Toll roads, Mexico Termopaipa Power Plant, Colombia Market risk * Guarantee of tariff rate/sales risk guarantee Don Muang Tollway, Thailand Western Harbour Tunnel, Hong Kong Buga-Tulua Highway, Colombia Toll roads, Mexico . Revenue guarantee South access to Concepci6n, Chile MS Motorway, Hungary full credit guarantees with more narrowly defined it is useful to calculate the subsidy implicit in each guarantees such as power purchase agreements. Such form of support. These "subsidy equivalents" help unbundling of risks presumes that the parties can be determine, for example, whether it is cheaper for the trusted to honor their commitments; if they cannot be government to provide a guarantee or some other trusted, investors will prefer full guarantees. This form of support. (For more on the role of guarantees helps explain why countries with low credit ratings in infrastructure finance see Dailami 1997.) rely heavily on full financing by export credit agencies The fact that government guarantees can be val- or multilaterals, whereas countries with higher credit ued and may be expensive to government does not ratings offer guarantees for specific risks (see table imply that governments should charge investors for 2.5). Support by multilaterals and export credit agen- the guarantees. When government guarantees merely cies appears to substitute for an international contract substitute for low prices, charging the full cost of the enforcement mechanism. guarantee would defeat the purpose of the guarantee. When the guarantor can manage or bear the risk better than the investor, however, the value to the guaranteed party is higher than the cost to guarantor, Guarantees provide (contingent) cash flow support to and the investor may be willing to pay part or all of projects and are, in many respects, similar to loans or the cost for a guarantee. Some commercial risks are grants. To be able to compare all forms of assistance, insured by private insurance companies for this rea- 29 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE TABLE 2.5 Patterns of sovereign or supranational support for private infrastructure projects Type of support Number Pattern Multilateral banks and export 37 Greater incidence of debt assistance by multilateral banks and export credit agency debt credit agencies in non-investment grade emerging markets (27). Government guarantees 28 Nearly three times as many government guarantees in non- investment-grade countries (24) than in investment-grade countries (9). Informal agreementsa 28 Although 9 agreements were issued in Mexico, use of informal agreements is more common in investment grade countries (11). Multilateral banks and export 26 Slightly more examples among noninvestment-grade emerging markets credit agency guarantees (15) than in investment-grade countries (11). Government equity participation 18 Greater incidence of government equity participation in investrnent- grade countries (11). Government debt (senior 14 Equal split among noninvestment and investment-grade countries. and subordinated) Multilateral equity participation 13 Much greater incidence of equity shareholding by multilateral banks and export credit agencies in noninvestment-grade emerging markets (11). Government grants 12 Greater incidence of government participation through grants in non-investment grade countries (8). Preferential tax treatrnent 2 Limited use of preferential tax treatment in investment grade countries. a. Informal agreements include comfort letters, side agreements, nonbinding tariff increases, and other similar agreements. Note: Financing packages of 78 projects (39 power, 26 transport, 7 water/waste, 4 telecommunications, and 2 gas) were disaggregated and then tabulated by type of mechanism and source of funds. All 78 projects have direct participation by the private sector through the provision of debt, equity, or both. son. Governments, however, should not be insuring countries, however, the risk of sovereign default is real, commercial risks, even on a fee basis. and its implications must be considered in structuring To the extent that private insurers are willing to government support to private infrastructure compa- provide cover for political risk, they need to charge for nies. The key task is to evaluate infrastructure projects the value of a guarantee. Governments, however, financially within the country risk environment pre- would be extracting rents from good policy by charg- vailing in developing countries (see Dailami and ing for such guarantees: charging for political risk Leipziger 1997). guarantees would be akin to demanding protection When there is a risk of default, one or more credi- money. Governments should instead ensure that the tors or investors may lose all or part of their invest- benefits to investors of such guarantees are passed on ment. By obtaining government guarantees an to consumers-by awarding projects competitively, investor or creditor obtains a position near the front for example. of the queue for repayment and secures access to sources of compensation not related to the project, Complications Arising from the Risk generally taxation. By obtaining a supporting guaran- tee from an institution such as the World Bank, a pri- of Sovereign Default vate investor can buy a place right at the front of the Sometimes the government's power of taxation queue, benefiting from the preferred creditor status of enables it to honor any obligations it has entered into the World Bank. It is not clear, however, whether such to provide support to a private infrastructure project. guarantees simply improve some investors' positions Official export credit and mortgage insurance schemes relative to others' or whether it contributes to a better in the United States are examples. In some developing overall outcome (see Dooley 1997). 30 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS The key issue is whether and how the structure of OECD countries. Germany, for example, actually val- government liabilities may affect the outcome of gov- ues certain guarantees the same way as debt with the ernment liability renegotiations. Even if renegotiation same maximum exposure. of government liabilities over extended periods of Beyond making claims more similar to each other, time preserves the net present value of creditor or can a commitment mechanism be chosen to facilitate investor claims, there may be real economic losses, speedy claims resolution? The COGASCO example, since assets funded by investors may not be used as mentioned earlier, illustrates that project-basedc rene- efficiently as they would otherwise have been during gotiation can last as long as sovereign debt settlement, the often acrimonious work-out process. For example, with deleterious consequences for investment in a par- a water concession may not be maintained as well ticular sector. It may therefore be useful to involve during a dispute as otherwise. multilateral creditors, because their interests and Different creditors or investors hold different actions may be most closely aligned and they may types of claims. They thus have varying interests to thus help advance resolution most speedily. negotiate. Some "tough" investors may hold up rene- It is thus by no means clear that finely tuned risk gotiation, thus imposing real losses (due to the less allocation is always the right approach. Blunter instru- efficient use of assets during the renegotiation), for ments, such as straight sovereign debt, may at times which the tough investor does not pay. When a gov- be preferable. The argument for seeking participation ernment issues guarantees to an infrastructure investor by multilaterals may have little to do with the nature it tends to create yet another type of claim. In particu- of the risk management or product they provide and lar, the guarantee may be issued to an investor who more with the role they are likely to play in debt rene- has some physical control over the assets. This gives gotiation. the guarantee holder bargaining power that differs from that of a holder of sovereign debt, for example. To some extent that may be justified for the same rea- Reforming Policy to Attract Investors son that trade credit gets treated preferentially during debt renegotiations so as not to disrupt basic econom- Although guarantees can provide some comfort to ic activity with adverse consequences for all. investors, a country's interests are better served by To achieve a solid and reasonably speedy settle- thorough-going policy reform. The best way of ment in order to minimize economic disruption attracting private investment is by establishing stable resulting from inefficient asset use, a mechanism macroeconomic policies, adequate tariff regimes, a needs to be in place that allows creditors and investors track record of honoring commitments, and reason- to resolve their differences quickly. This is achieved able economic policymaking. In many OECD coun- more easily if the claims held by different investors are tries and other industrial economies, such as similar and the government has the flexibility to come Singapore, investors may not require guarantees or up with various ways of settling its obligations. other government support, and they may be willing to When a country properly accounts for its contin- accept "change of law" risk, which may affect tax rates gent liabilities and reserves for them fiscally, they or other project cost or revenue parameters. appear more like normal debt. In fact, it may be In many emerging markets, however-including preferable for the government to support projects by relatively advanced economies such as Chile- providing debt finance rather than guarantees. If so, it investors may not find the right policies in place, or could be argued that, to provide governments with they may doubt the government's ability to sustain the right incentives to do so, exposure under govern- such policies over long periods of time. Governments ment guarantees should be valued like debt and not still have a variety of options for reducing the need for be reduced by adjusting for probability of default. In a special project support. sense such an ultra conservative policy is equivalent to Projects are subject to country- and project-specif- debt management policies in various advanced ic risks. Risks related to a country's overall health tend 31 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE TABLE 2.6 Credit ratings and signed project finance deals, 1996 Value of Value of deals deals per capita as a percentage Country Rating (S/population) Country Rating of GDP Qatar BBB 8,564 Hong Kong A 13.5 Hong Kong A 3,229 Indonesia BBB 7.1 Australia AA 705 Thailand A 5.7 Greece BBB- 282 Chile A- 4.9 Chile A- 234 Pakistan B+ 4.5 United Kingdom AAA 227 Malaysia A+ 4.2 Saudi Arabia NR 214 Australia AA 3.7 United States AAA 185 Greece BBB- 3.2 Malaysia A+ 178 Saudi Arabia NR 3.1 Thailand A 159 Turkey B 2.4 Canada AA+ 151 India BB+ 2.1 Argentina BB 99 Argentina BB 1.2 Italy AA 78 China BBB 1.2 Germany AAA 76 United Kingdom AAA 1.2 Indonesia BBB 73 Brazil BB- 0.8 Turkey B 63 Canada AA+ 0.8 Brazil BB- 37 United States AAA 0.7 Pakistan B+ 21 Italy AA 0.4 India BB+ 7 Germany AAA 0.3 China BBB 7 Qatar BBB n.a. Note. Population and GDP data are for 1995. Source: Euromoney; World Bank 1997b; World Bank staff estimates. to be of prime importance. Risks such as currency and problems that can remain when projects go ahead, interest-rate risks reflect macroeconomic volatility and with various forms of government support, in the the risk that the government will not honor its obliga- absence of serious policy problems. tions (country risk proper). The Mexican toll road program generated several That governments with stable macroeconomic poli- billion dollars of nonperforming assets in the domes- cies can attract private infrastructure investors more eas- tic banking system. No explicit guarantees had been ily is reflected in the sovereign debt ratings given by issued to creditors, but local banks expected the gov- various rating agencies and services (see annex table emient to bail them out once the toll roads ran into A2.5). As country ratings improve, governments are financial difficulties. The government was forced to able to attract more and more project finance (table come to the banks' aid at the worst possible time- 2.6) (although project finance accounts for only a small during the currency crisis of 1994/95. percentage of GDP in the most creditworthy countries, The failure of private toll roads has caused prob- where corporate finance is used to finance deals).5 lems in other countries as well. In Thailand the Bangkok expressway required government rescue after Problems with Financial Support the authorities declined to raise tolls in line with earli- without Poicy Reform er agreements. In Spain the government was obliged to pay out $2.7 billion when exchange rate guarantees The jury is still out on the consequences of govern- were called during the 1970s and 1980s. ment guarantees and other forms of financial support: Other types of projects have also been affected. although they may have increased the volume of Malaysia's power company, TENAGA, contracted with investment, they may not have solved the underlying private generators (backed by a government guarantee) problems. Several examples illustrate the types of to supply more power, but consumer tariffs were left 32 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS unchanged. As a result TENAGA was not able to carry because taxpayers stand behind them, providing unre- the full cost of private generation forward and was munerated credit insurance. If taxpayers were remu- squeezed financially, forcing it to neglect maintenance nerated for their exposure, the ostensible advantage of and investment. Power cuts throughout the country government finance would presumably disappear. If followed-exactly the outcome the new generation not, governments should finance everything, includ- capacity was intended to prevent. ing large corporations-a return to GOSPLAN, In Mexico a water concession in Aguascalientes which appears nonsensical (Klein 1996). was concluded in 1993. To guard against currency Government support to private projects compen- risk, variable-rate debt financing was obtained in the sates private investors for the risks they are unwilling local markets. Water prices were thus not indexed to to bear given the prices they receive. Investors may be exchange rate movements but (partially) to changes in attracted to infrastructure projects without guarantees interest rates on domestic debt and inflation. if the expected returns are high enough (that is, when Following the foreign currency devaluation in rates charged to consumers are high enough).6 In that 1994/95 inflation and domestic interest rates rose, sense the search for guarantees or other forms of gov- which should have caused large nominal tariff increas- ernment support is a search for suckers who can be es. A political decision was made, however, not to made to pay what others are not willing to pay. raise tariffs as foreseen in the concession contract. Guarantees themselves do not appear to affect the cost Instead the government took on the financing of new of capital, which is determined by the risks of the pro- investment that the concessionaire was supposed to ject, not the financing structure. As recent review of have made. the effect of World Bank partial credit guarantees These cases have some key features in common. (Huizinga 1997) suggests, the existence of guarantees First, problems were resolved by negotiation, as they did not reduce nonguaranteed interest rates, and the usually are in cases of government-related risks. In duration of nonguaranteed debt remained relatively contrast, disputes over technical or commercial risks short. are often resolved in court. Second, the government generally ended up bearing a substantial part of the Privatization ofExistingAssets costs-costs that could have been avoided if the gov- ernment had allowed consumer prices to cover full Recent transactions have shown that even countries project costs. with subinvestment grade ratings can attract sizable These examples reveal how the basic forces that private investment without special government guar- drive infrastructure privatization assert themselves. antees if sound sector policies are made credible. Private investors do not-and should not-pay for Privatizing existing assets reduces the role of govern- projects; they can only finance them. Either consumers ment and with it fears of noncommercial interference. or taxpayers have to pay for projects in the end. If the In Argentina, Bolivia, and Peru, for example, where government cannot raise money from taxpayers, con- certain sectors, such as electricity, were privatized, pri- sumer prices must be adequate. Therefore, when priva- vate investment has been made without government tization is motivated by fiscal constraints, user fees guarantees. must be raised to cost-covering levels. Projects that Privatization also allows investors to earn high rates cannot be funded by user fees should not, in the of return without raising consumer tariffs, since absence of important positive externalities, -be built. investors discount the sale value of assets to the point at Government support could lower overall project which existing tariffs generate the required rate of cost only if the government had a lower cost of capital return, rather than by raising tariffs, as they would have than private parties. Although government borrowing to do in greenfield projects. In fact, tariffs can actually costs are often ostensibly lower than private borrow- fall after privatizations, as they did in the Buenos Aires ing costs, governments borrow at lower rates not water concession, in which the assets of the system were because they tend to operate lower risk projects but given to the private investor free of charge.7 33 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Privatization has also attracted more equity infrastructure projects. In China, an investment-grade investors than have new investment projects. Since country, investors have been willing to accept guaran- equity markets are easier to develop than long-term tees from provincial governments in place of the debt markets in most developing countries, privatiza- national government. In India, a subinvestment-grade tions have been able to rely more on local currency country, the verdict is still out, but it appears that pro- financing than have greenfield investment projects. jects going ahead require heavy backing from state- The typical new investment project requires about owned financial institutions. two-thirds foreign finance, whereas the typical privati- Colombia, an investment-grade country, has been zation has attracted two-thirds of its finance from local able to move away from sovereign guarantees in pro- markets (International Finance Corporation 1996). jects in which ECOPETROL, the state-owned oil Many privatizations have occurred in subinvest- company, is backing payment obligations (Centragas ment grade countries (that is, in countries with credit and Transgas). Several Colombian entities have ratings of less than BBB-), including Argentina, recently issued investment-grade paper (for the El Bolivia, and Peru. Privatization has allowed these Dorado airport expansion and the city of Bogota). countries to attract investment despite their unstable Petropower, a Chilean co-generation project, was able macroeconomic environments, allowing them to to issue bonds in the U.S. capital markets without the make the most of existing assets rather than to add help of the government or supranational agencies. new investments. Although Argentina is not an investment-grade coun- try, Transportadora de Gas del Norte in Argentina was Greenfield Projects able to issue investment-grade paper with the help of International Finance Corporation (IFC) participa- Government guarantees and financial support are tion (other innovative capital market issues are more difficult to avoid for new investments, for which described in annex table A2.6). prices must be raised. Well-structured project finance for greenfield projects may allow governments to Rethinking the Problem ofFuture Investment avoid guarantees or other forms of support, however. Requirements Under project finance investors look to cash flow gen- erated by the project to amortize debt and to pay The "financing gap" may in fact be a "policy gap"'- interest payments and dividends.8 Project finance can what is needed is not so much the mobilization of help investors structure a project so that different risks new financial resources on a vast scale but a thorough- can be separated and allocated to the parties most going reform of policy. Raising consumer prices to willing to bear them. An example is the Mamonal cost-covering levels would generate some $123 billion power project in Colombia, where a foreign power a year, allowing infrastructure companies to fund generator sells electricity directly to private firms at most of the $200 billion a year needed for infrastruc- cost-covering prices. This project structure has ture from internal cash generation, leaving only $77 allowed the project company to set high user fees and billion to be funded in the financial markets (World rely on payment discipline by creditworthy corporate Bank 1994). In addition, private participation could customers rather than on government guarantees. create efficiency gains of $55 billion a year, reducing Several countries are trying to reduce reliance on financing requirements to $22 billion (figure 2.6). sovereign support for new infrastructure projects. Moreover, the increase in tariffs to consumers should Most of the countries that have been successful in reduce demand and therefore investment require- doing so have had investment-grade ratings. Indonesia ments. To be politically able to raise consumer prices attracted investors by issuing comfort letters on for- and to obtain the benefits of greater efficiency, govern- eign exchange convertibility in its PAITON power ments should proceed with privatization. If they project. China and India have declared that they are choose to go this route, however, the long-run financ- unwilling to issue sovereign guarantees for private ing problems will be minimal-financing require- 34 GOVERNMENT SUPPORT TO PRIVATE INFRASTRUCTURE PROJECTS IN EMERGING MARKETS FIGURE 2.6 difficult challenges arise during the transition from Estimated cost of mispricing and technical publicly to privately funded infrastructure, when inefficiency guarantees are most common. Even during the transi- Billions of dollars Investment tion, however, government guarantees risk simply 200 postponing the day of reckoning. Assuming that pri- vate investors cannot consistently be duped into Fiscal investing in unsustainable projects, providing guaran- burden tees imposes costs on taxpayers in the future. For this 123 reason alone governments should develop ways of quantifying all their exposures to private infrastruc- Railways | Resource ture projects and reserving for them fiscally. _los5s Two governments in the developing world-the 55 [; ;goadg- Guarantees may actually worsen and structural imbalances? macro risks. Yes Is there political stability and - o-- - Macro guarantees have little commitment to reforms? policy signaling value. Yes Are the credit markets financing - e'- Guarantees may displace private initiative infrastructure projects? or distort attitudes toward risk. No NO Selective macro guarantees may be appropriate. adjusting interest rates or exchange rates when should be deepened to provide the private sector they become unsustainable for fear of incurring with vehicles for trading in macro risks. large fiscal losses. • Macro guarantees cannot be expected to be very credible when there is a threat of large adjustments Notes in the exchange rate or inflation, when the politi- cal environment is shaky, or when government 1. Governments generally acquire equity stakes not appears likely to abuse its ultimate enforcement simply to share risks efficiently but also to achieve other role. Thus the more assurance the private sector policy objectives (such as employment generation or preser- requires, the less useful the guarantees become as a vation, national security, or other social objectives). policy device. Moreover, risk sharing is not the only form of implicit * Even if they are carefully targeted, macro guaran- financial assistance that public companies receive. tees run the risk of perpetuating the image of the Generally, public enterprises have preferential access to state as the residual absorber of risks. Moreover, by some factor or product markets, through subsidized credit shielding private agents from macro outcomes, schemes or monopoly rights. guarantees may aggravate the underlying informa- 2. This would imply using a risk-free rate for discount- tional asymmetries favoring government. ing public investments, as prescribed by Arrow and Lind * A policy of guaranteeing macro outcomes makes (1970). See Bailey and Jensen (1972) for a critical review of sense only as an interim policy while a broader set private versus public risk bearing and the implications for of reforms are being implemented that will make the social rate of discount. guarantees unnecessary in the medium term. * Rather than simply absorbing the risk itself, wher- ever possible government should strive to elimi- References nate the information and transaction cost asym- metries that would place it in a better position to Arrow, Kenneth L., and R.C. Lind. 1970. "Uncertainty and absorb risks. Financial liberalization, in particular, the Evaluation of Public Investment Decisions." 127 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE American Economic Review 760: 364-78. Developing Countries: Lessons from the Railway Age." Bailey, Martin J., and Michael C. Jensen. 1972. 'Risk and Policy Research Working Paper 1379. Washington, the Discount Rate for Public Investment." In Michael D.C.: World Bank. C. Jensen, ed., Studies in the Theory of Capital Markets. Klein, Michael. 1996. "Risk, Taxpayers, and the Role of New York: Praeger. Government in Project Finance." Policy Research Bosworth, Barry P., Andrew S. Carron, and Elisabeth H. Working Paper 1688. World Bank, Private Sector Rhyne. 1987. The Economics of Federal Credit Programs. Development Department, Washington, D.C. Washington, D.C.: The Brookings Institution. Mayshar, Joram. 1977. Taxation with Costly Administration. Eichengreen, Barry. 1994. "Financing Infrastructure in Madison, Wisc.: Social Systems Research Institute. 128 MANAGING EXCHANGE RATE- AND INTEREST RATE-RELATED PROJECT EXPOSURE Comments on "Managing Exchange Rate- and Interest Rate-Related Project Exposure: Are Guarantees Worth the Risk?" William H. Chew, Standard & Poor's Ratings, even some private sector participants, rarely identify New York and measure the risks they undertake. As a result risk allocation arrangements may not be as airtight as they Ignacio Mas presents a refreshing approach to issues appear. Risks are especially difficult to identify and that credit analysts, lenders, and project developers measure when they are covered indirectly through encounter in raising debt capital for private infrastruc- potentially sweeping pricing formulas. Power tariffs ture projects. His chapter correctly highlights several and transport tolls that are indexed to transfer curren- important aspects of guarantees, many of which need cy risk, for example, have the potential to raise prices to be more fully considered by both governments and for essential services precisely when countries are least the private sector. able to handle the increase. The existence of the guar- First, there are inherent limits to guarantees. From antee does not necessarily change the underlying risk. a practical perspective the basic premise of many guar- Third, it is difficult in practice to unbundle risks. antees is flawed. The countries perceived as being in Most agree that markets are able to handle commer- greatest need of guarantees are often precisely the cial risks well but have a hard time managing political countries in which such guarantees may have little and governmental risks. Separating these risks is often value. This is because the very government that is pro- difficult because the distinction is artificial, as differ- viding the guarantee is also the party orchestrating the ent types of risk are often intertwined. For the credit actions the guarantee is designed to cover. In the case analyst the real issue is the way in which the distinc- of sweeping events or changes, especially when part of tion between the risks is handled under project docu- a broader policy shift, governments will find it diffi- ments and the record of the host country in handling cult to enforce the guarantees. Thus guarantees, like those risks. For many projects, the distinction may be other forms of contractual credit support, are effective an invitation to debates of interpretation and thus a only when they are aligned with strong economic and source of potential risk of projects. business incentives to support them. Guarantees that Guarantees and risk covers will continue to play fly in the face of political reality, no matter how artful- an important role in raising debt capital in many ly drafted from a legal perspective, may not be worth countries. To facilitate the efficient flow of longer- much. tenor debt, practical steps must be implemented to Second, there is no free lunch. For a guarantee to move both governments and project sponsors toward have value the party undertaking it must accept the markets in which guarantees are the exception rather risk that it may be called in. But governments, and than the rule. 129 6 The Management of Contingent Liabilities: A Risk Management Framework for National Governments Christopher M. Lewis and Ashoka Mody ABSTRACT Policymakers view privatization as a way of reducing grated risk management systems will vastly improve the government's fiscal burden. But explicit and governments' ability to manage and control risk and implicit government guarantees provided as part of will enhance their efforts to improve the allocation of the privatization process often expose governments to resources in the domestic economy. considerable risk-which is rarely reflected on the Of course, the focal point of any government risk government's balance sheet. The contingent nature of management program is the systems used for account- this risk exposes governments to the possibility of sud- ing and budgeting for contingent liabilities. den and substantial obligations over a short period of Governments are often unaware of their exposure time, which could lead to severe fiscal problems. As because of their use of cash-based budgets. Cash-based the pace of privatization accelerates, governments' budgeting masks the contingent exposure and creates exposure to risk is rising, underscoring the importance perverse incentives for issuing guarantees. By not of an integrated approach to risk management. accounting for the budgetary costs of issuing guarantees For a governmental institution, integrated risk a simple cash budget encourages the expansion of guar- management involves: (a) identifying and classifying antee liabilities without requiring the government to the risks faced; (b) quantifying the government's expo- reserve against future losses. It allows political leaders to sure from these risks; (c) including those measures of increase financial assistance to target groups without risk in the budgeting process; (d) identifying the gov- being held accountable for the costs of providing the ernment's tolerance for risk; (e) establishing policies assistance, which will be realized under ensuring and procedures for structuring unexpected loss administrations. To improve the allocation of resources reserves; and (f) implementing systems for monitor- governments should follow the lead of the private sec- ing and controlling exposure over time. Use of inte- tor and move to a present value basis of accounting. 131 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE D uring the transition from public to private Other countries are facing a similar escalation of costs ownership and management, governments from deposit insurance programs. In some instances often provide various kinds of support, last minute government action averted a serious bud- including commitments to make streams of payments getary crisis (U.S. pension insurance). In other cases, in the future. Some of those payments are deferred improvements in the economy helped prevent a crisis payments, and the government in effect transfers the from occurring (U.S. deposit insurance for commer- financing off its own balance sheet as it enters into a cial banks). In many cases, however, existing and financial lease arrangement. Some commitments are growing contingent liabilities could significantly contingent, that is, they come due only if particular aggravate the next budgetary crisis. events transpire. Contingent obligations, such as guar- Drawing on recent advances in the private sector, antees, require no immediate cash outlay and are this chapter outlines a risk management agenda for therefore often favored as a method of support. national governments. It develops a framework for However, guarantees represent real liabilities and can improving the assessment, measurement, budgetary cost as much as a third of the amount guaranteed (see control, and management of risks and demonstrates Mody and Patro 1996). Moreover, these liabilities how this framework can be applied to contingent increase as government activities are moved to the pri- infrastructure liabilities. It also examines how the vate sector through privatization. Only recently have implementation of an integrated framework for risk government auditors and Treasury officials begun to management can be used to improve the ability of the recognize the continuing fiscal implications of infra- government to design programs that target specific structure privatization. risks in a transaction, allowing the public sector to Contingent liabilities arise in a variety of contexts. leverage private capital. Recently, the move to place infrastructure provision in Section 1 identifies the main components of any private hands has led to a variety of guarantees that integrated risk management system and shows how represent a significant liability for governments. In private firms use this framework to improve their own many developing countries government guarantees are business operations. Section 2 shows how this risk also used to support other private sector activities. In management framework can be adapted to the needs addition, government guarantee programs support of a government institution. Section 3 describes tools pension liabilities, export credits, and agricultural sup- and techniques for identifying risk and quantifying port. Furthermore, governments typically provide the risk exposures. Section 4 describes alternative bud- contingent support to individuals, companies, or pro- getary approaches to managing the expected payouts jects considered too risky for private financial institu- under contingent liabilities. Section 5 demonstrates tions. The full extent of these liabilities is not known, how reserves against unexpected losses enable govern- because no attempt has been made to systematically ments to manage the volatility in budget expendi- estimate them. In some parts of the world, however, tures. Section 6 highlights the advantages of a com- government guarantees may soon represent an prehensive risk management system that induces clari- unmanageable level of exposure, not only because of ty of contract design, minimizes incentives that lead their size relative to the size of the government's bal- to a call on guarantees, and implements a regular ance sheet but also because their contingent nature monitoring process. The last section summarizes the implies the possibility of sudden and substantial oblig- chapter's conclusions. ations due over a short period of time. However, governments have made little effort to develop their own systems for managing risk. As a An Integrated Enterprise Risk Management result, governmental programs have been at the center Framework of some of the largest risk-related losses. In the United States, for example, the savings and loan debacle in The goal of corporate risk management today is not the 1980s cost taxpayers more than $130 billion. to manage a fixed set of risk exposures of an enter- 132 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS prise, but to determine the areas and lines of business the risk. The system allows Microsoft to focus on in which a company is willing to retain risks in order managing the business risks associated with succeed- to generate target returns. An integrated approach to ing in its core market. Bankers Trust, Chase corporate risk management helps a firm optimize the Manhattan, and Analog Devices have initiated simi- trade-off between risk and return so as to maximize lar systems. the firm's overall risk-adjusted rate of return on equity An integrated risk management process should and its shareholder value.' perform six major functions (figure 6.1): Over the past several years many large multina- * Identifying the firm's risk exposures tional firms have implemented enterprisewide sys- * Measuring or quantifying those exposures tems for risk management. Microsoft Corporation, * Assessing the firm's tolerance for risk-bearing for example, has just completed building an elaborate * Making strategic decisions on the allocation of risk management system that quantifies more than capital to support risks that are borne 144 different types of risk exposures. For each risk * Implementing risk mitigation and control mecha- identified as important, Microsoft determines the nisms to prevent unintended losses on those risks best approach for improving its management of expo- and establishing systems to continually monitor sure by insuring, transferring, mitigating, or retaining and reassess the firm's risk exposure over time. FIGURE 6.1 Integrated enterprise risk management: Optimizing enterprise returns under uncertainty Reprinted with permission from Ernst & Young LLP. oErnst & Young LLP. 133 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE Box 6.1 Management of contingent liabilities in the Philippines The government of the Philippines responded to a critical also subject to higher fees. The consultative document national power shortage by providing 'full faith and recommended withdrawing certain guarantees (such as credit" guarantees to private sponsors against the risk of the guarantee of currency convertibility if the Philippines payment default by the National Power Corporation attained investment grade credit rating and the guaran- (NPC), the public power utility buying power on long- tee of NPC payment obligations if NPC attained invest- term power purchase contracts from private generators ment grade rating). It also recommended limiting guar- under a BOT arrangement. The government waived its antees to 80 percent of total project costs in order to right to sovereign immunity, thereby accepting interna- require equity investors to bear their share of project tional arbitration in the event of a dispute. risks, developing model guarantee documents that Provision of free guarantees was crucial to the would form part of the bidding package for prospective financing of substantial generation capacity (about 3000 project sponsors, and instituting internal controls MW), which alleviated the power crisis. But it meant (including accounting for and reserving against guaran- that sponsors and lenders came to expect that such all- tees). inclusive guarantees would always be available. A set of model guarantee documents was produced Recognizing that guarantees are neither desirable nor and is now being used in specific projects. The first pro- sustainable, the government issued a consultative docu- ject to which the approach was applied was the Renon ment in March 1995, making specific recommendations Toll Road, which runs from Manila to Cavite. The key for better management of its contingent liabilities element guaranteed was the tariff formula. Since no (Government of the Philippines 1995). The government guarantee was provided for traffic or revenue volumes, acknowledged that guarantees could not be eliminated no payment obligation akin to the power purchase abruptly and that a transition was required during which agreements was incurred by the government. The guar- the legitimate risk mitigation needs of private parties antee of foreign exchange convertibility provided only would be met while an improving performance gradually for equal treatment, as specified in current Philippine allowed various elements of the guarantees to be law. The new approach is also being applied to major eliminated. power projects currently under negotiation, including A key feature of the policy was unbundling risks to the $300 million San Pascual Cogeneration Facility. In allow more flexible management (table). Certain core all of these projects the government is using the new guarantees of government obligations of "fundamental guarantee package to pare back its contingent liability rights" under a project were seen as legitimate for the and to provide a means for reducing liability even fur- government to offer to establish a record of policy per- ther when the need for a particular form of guarantee formance. Other guarantees, including the guarantee of diminishes. Discussions are continuing with the spon- currency convertibility and the risk of nonpayment of sors. The policy is also being used for new water pro- obligations by NPC, were seen as temporary and were jects coming on stream. Managing exposure under guarantees through unbundling risks Nature of guarantee Risk Core guarantee Noncore guarantee Fee chargeda Sovereign risks Concession terms, Terms define basic rules are None expropriation, tariff largely under government control formula, tax incentives Obtaining of licenses, Government commits to facilitating 25 basis points permits, right-of-way process. Risks mot fully under central government control. Foreign exchange risk Convertibility of Government assigns priority. Risk 25 basis points for the foreign exchange not fully under government control priority accorded Market risk Not under government 50 basis points initially to control. reflect commercial risk. Credit risk Transitional need to make No initial charge. Fall-away project financeable. provisions when credit benchmarks are achieved. a. Fee charged is indicative only. Source: Government of the Philippines 1995. 134 (Box continues next page.) THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS These functions are used to manage the four scarce resources on the highest risk areas. This process broad categories of risk: financial, operational, busi- yields a risk identification lattice (figure 6.2). ness, and event risk. Using a similar approach a government can assess its own risk exposures in a particular program. The advantage of this top-down approach is that the gov- Using Enterprise Risk Management to Manage ernment can focus resources on those risk categories, Government's Contingent Liabilities classes, or risk types for which exposure is greatest. This approach economizes on scarce resources and it A similar integrated enterprise risk management minimizes disruption that may be caused in the pri- framework can be adopted by any government institu- vate sector by excessive government audits. tions to help them maximize social returns. A risk-focused assessment procedure was worked out Unlike private firms, government needs to approach in the El Cortijo-El Vino toll road project in Colombia risk management from an economywide perspective. (box 6.2). The assessment determined that the greatest Implementation of a risk management system is useful exposures for the Colombian government were from the in this regard, since it provides governments with anoth- market risks associated with traffic volatility and from er tool with which to identify which risks should be construction cost overruns (figure 6.3). Early recogni- borne by the government and which should be borne by tion of these risks allowed the Colombian government the private sector. to improve its risk management techniques and contract Each of the six functional areas of a comprehen- specifications for toll road projects. sive risk management system can be implemented by Once a central government goes through the governments to improve management of their contin- process of identifying the risks it faces and gains a bet- gent liabilities, and specifically, their infrastructure lia- ter understanding of its risk exposures, the valuation bilities. Of course, implementation of the framework or quantification process can begin. A wide variety of in a particular country would require significant techniques exists for quantifying different types of adjustments to reflect the structure and dynamics of risk. The techniques used depends on the type of risk the national government, the budgetary and regulato- being analyzed. (Although this chapter addresses the ry processes, the legislative and legal environments, application of these techniques to the contingent lia- and the risks being evaluated. bilities of a government, these tools can be used to manage risk on the government's entire balance sheet.) Identifying and Quantifying the Risks Actuarial or Statistical Techniques The government's exposure to loss can arise from a Where a large body of data exists on prior losses or data wide variety of events. Attempting to account for can be augmented using statistical techniques, actuarial every source of exposure is not feasible. A systematic methods that estimate future loss patterns based on approach to identifying the principal risks is needed prior loss experience (including trends) are often used to ensure that all relevant exposures of a program can to quantify the government's exposure to loss. Actuarial be classified. techniques, which have been used to assess insurable One approach to risk assessment is that adopted by risks for almost two centuries, can be used to assess the federal regulators of financial institutions in the United magnitude of a wide variety of risk exposures. States and Europe. With limited staff resources federal Actuarial techniques use the loss history of a given regulators have evolved a top-down, risk-focused program-or comparable programs-to estimate an approach for conducting risk management examina- annual expected loss distribution. This annual expected tions of financial institutions. Regulators first examine loss distribution is then adjusted to reflect current an enterprise's general categories of risk (financial, trends in loss frequency and loss severity, as well as any business, operational, and event risks) then focus their changes in the sharing of risks between the government 135 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE FIGURE 6.2 Risk identification lattice Government risk exposure Financial risk Business risk Operational risk Event risk Market risk Strategic risk Productionis Liquidity risk Mangeen Legal risk Exgeosk Credit risk Ssesrs and the insuredt party. If the annual adjusted loss distri- econometric analysis into the parameterization of the bution is assumed to remain stable over time (adjusting expected loss distribution allows the government to: for any time trend), the distribution can be used to esti- *Model economic and financial trends that may mate the expected and unexpected costs of the program influence the pattern of losses within a program, in any given year. Discounting cash flows using a risk- free rate of interest yields estimates in current dollars.2 Box 6.2 Identifying which risks to guarantee Econometric Modeb ~ ~~~~~in the Colombian toll road project The Colombian government provided two basic forms A deficiency of actuarial models is that they do not of assurance to support the toll road project, a con- attempt to explain the patterns of loss they identifyr struction materials overrun guarantee and a traffic vol- and thus cannot be used to forecast nonlinear trends ume guarantee once road construction was finished. Under the terms of the cost overrun guarantee the gov- in loss patterns, as in the case where the risk sharing ermient would cover 1 00 percent of the cost of mater- between the government and the private sector change ial overruns that were 30 percent of the original con- over ime.This hortomin can e sinificnt, spe- struction design bid, 75 percent of the cost of material over ime. his sortcoing cn be ignifcant,C5PC overruns that were within 30 to 50 percent of the orig- cially when analyzing the performance of credit pro- inal construction design bid, and 0 percent of the cost gasthat are sensitive to economic fluctuations, of material overruns that were more than 50 percent grams ~~~~~~~~~~~~~~higher than the original construction design bid. T'he Econometric methods can be used to show how traffic volume guarantee committed the government to the expected loss distribution of a program may reimbursing the concessionaire if traffic volume falls 10 percent below the traffic volume projections agreed change over time based on the pattern of underlying to in the budget for the project. If traffic volume economic or financial factors. By forecasting future exceeded projections by more than 10 percent, the movements in these factors, econometric models can additional revenues associated would be deposited in a reserve fund used to cover future shortfalls in traffic be used to compute how these loss distributions may volume or for road maintenance and improvements. change over the life of the program. Incorporating 136 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS FIGURE 6.3 Sources of risk in the El Cortijo-EI Vino toll road project Government risk exposurel Financial risk j Business risk Operational risk Event risk Market Market risk Construction - currency of traffic cosrcinlPolitical risk risk volatility risk Inflation risk Termination Legal risk Force majeure allowing for more dynamic projections of losses vary across time with changes in the underlying and the incorporation of loss events for which economy. there is no historical precedent Over the past twenty years econometric models * Identify factors that affect loss behavior, so that have become increasingly sophisticated and powerful, actions can be undertaken to mitigate losses evolving from simple ordinary least squares models to * Improve the ability of an underwriter to evaluate logistic regressions, to nonlinear regression models the riskiness of program participants based on the and complex hazard functions. Quercia and Stegman characteristics of the participant or factors affect- (1992) provide a detailed review of the evolution of ing the participant. econometric techniques and models just within the One useful application of econometric modeling is mortgage industry. Default/prepayment models are in identifying the loss patterns associated with credit also available for small business loans, consumer loans, risk. When the government provides a direct loan or and credit card receivables. protects a third party against the default of a borrow- Both actuarial and econometric models require er, it exposes itself to the risk that the borrower will substantial data inputs on the performance of a pro- default. When it lends directly, the government also gram (or comparable program). Project finance, where faces prepayment risk exposure-the risk that the bor- deals are unique and data records are often missing or rower will repay the loan early, leaving the govern- of low quality, more advanced modeling approaches ment exposed to a loss of interest and to reinvestment are required, including stochastic simulation analysis risk. Both credit and prepayment risk can be affected and contingent claims models. significantly by conditions in the economy (such as a drop in interest rates, which usually leads to an increase in mortgage prepayments as homeowners that C have higher-coupon mortgages refinance). Econo- Contingent claims analysis is a powerful technique for metric models can be used to assess how these risks estimating the value of a loan guarantee, direct loan, 137 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE or insurance program-in isolation or as part of a This approach has been used by the federal gov- complex layering of risks. It is particularly useful emient in the -United States since 1992, when the when historical data on the performance of a program Office of Management and Budget adopted contin- are not available. Contingent claims are assets (or lia- gent claims models for deposit insurance, pension bilities) whose values on a specified future date are insurance, and mortgage guarantees to help it forecast uniquely determined by the prices of other traded budget costs during the five-year budget window and securities. The classic example is a European call beyond. Contingent claims models have also been option issued on an underlying stock-that is, an used to compute a range of expected long-term costs option to buy a stock at a specified exercise or strike for these programs, which have been published in the price on a specified date in the future. federal budget. While contingent claims models have In a seminal paper Black and Scholes (1973) not being used directly to determine the expected cash demonstrated that the price of a European call option outlays in each year of the budget window, these mod- can be valued using only the value and instantaneous els have been used as part of the federal budgetary variance per unit time of the underlying asset, the process. term of the option, and the risk-free rate of interest. To understand how contingent claims analysis is Merton (1973b, 1977) followed with a more general- used to value government guarantees, insurance, and ized theory of contingent claims pricing that allowed direct loans, it is important to first understand the for the development of new models to price all types financial equivalence of each of these instruments of assets whose payoff structure could be linked to an from the perspective of risk. When a government underlying security. Since 1973 techniques have been institution issues a direct loan, it transfers cash to the developed to value a wide array of financial and non- borrower in exchange for a promissory note of repay- financial instruments, including complex financial ment and collateral, usually in the form of a down options, corporate liabilities, third-party guarantees, payment and a secured interest in the value of the employee compensation, insurance products, and underlying asset that was purchased with the bor- more recently, the value of capital investment deci- rowed funds. If the loan were risk free-that is, if the sions, or "real options." Development of a theory of probability of a loss on the loan were zero-there rational options pricing helped foster the expansion in would be no need for the collateral interest, and the the financial markets over the past twenty-five years. government could record the full value of the loan Contingent claims analysis is also an extremely repayment as an asset on its balance sheet. Direct powerful tool for analyzing government loan guaran- loans are rarely risk free, however, as the borrower has tees, direct loans, and insurance programs. Merton the option to default on the note and transfer the (1977) used a modified form of the original Black- underlying collateral to the government. In fact, the Scholes options pricing equation to determine the borrower could be expected to default on the loan if value of deposit insurance in the United States. Marcus the costs of default (the loss of collateral and all trans- and Shaked (1984), Pennacchi (1987b), and actions costs, including penalties) were less than the Cooperstein, Pennacchi, and Redburn (1995) expand- benefits associated with continuing to make payments ed this work. The use of contingent claims analysis was on the loan. Thus, as Merton and Bodie (1992) also extended into other areas for assessing the value of showed, the issuance of a direct loan is analogous to government liabilities, including federal loan guaran- bundling two separate transactions-the issuance of a tees granted to corporations (Sosin 1980), mortgage risk-free loan and the underwriting of a put option guarantees (Foster and Van Order 1985; Cooperstein, with an exercise price of the outstanding value of the Redbumn, and Meyers, 1992; Kau, Keenan, Muller, loan and an underlying asset represented by the collat- and Epperson 1992), state guarantee funds supporting eral securing the loan: insurance company failures and federal pension insur- ance (Lewis and Cooperstein 1993; Hsieh, Chen, and Value ofRisky Direct Loan = Value ofRisk-free Loan - Ferris 1994; Pennacchi and Lewis 1994). Value ofDefaulk Put Option. 138 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS A rational borrower is expected to default on a P = Max[0, L - 71 - Max[0, L - C]. home mortgage if the value of the outstanding loan (L) exceeds the value of the underlying house by more For a reinsurer the first term in the equation is than the transactions costs and penalties (P) of analogous to being short (that is, having underwrit- defaulting. The payout of this default option is identi- ten) a call option that allows the primary insurer to cal to the government underwriting a put option on "call" on the resources of the reinsurer to pay for losses the underlying value of the house (1X, with an exercise that exceed the threshold insurance trigger. The sec- price equal to the sum of the loan and the costs of ond term in the equation is analogous to the reinsur- default (L + P1. er's being long (that is, having purchased) a put The only difference in the case of a 100 percent option that allows it to "put back" to the insurer any loan guarantee is that the transaction is unbundled. A losses that exceed the reinsurance cap. Thus the rein- private bank issues the risk-free loan, and the govern- surance contract is simply the difference between a ment underwrites a put option in the form of a loan put option and a call option written on the underly- guarantee given to the bank issuing the loan. Thus: ing exposure of the insured event, or a call spread option. If the trigger is defined as deductible a similar Value of Risky Direct Loan = Value or Risk-free Loan - argument can show how a standard primary insurance Value ofLoan Guarantee. contract can be expressed as a financial option. The techniques used to value financial options are, or then, directly applicable to the valuation of direct loans, loan guarantees, and insurance contracts grant- Value ofa Loan Guarantee = Value of a Risk-free Loan - ed in the process of supporting infrastructure liabili- Value of Risky Loan. ties. Governments can use options pricing theory to formulate a more accurate assessment of their aggre- It is straightforward to show the equivalence gate risk exposure in project finance and other areas. between the structure of an insurance policy and a contingent claim. When the government underwrites a loan guarantee the government is providing assur- AluingG rae ia ance to other parties that it will bear the risks associat- ed with borrower default. A loan guarantee is thus In 1996 the Colombian government and the World analogous to a credit insurance policy against borrow- Bank collaborated to quantify the risk exposure of er default. three project finance transactions. The purpose of the This analogy also applies to layered insurance or effort was to establish the viability of a methodology reinsurance policies. Excess-of-loss reinsurance pro- to obtain estimates of the government's exposure. To vides protection for losses (L) that exceed some trigger the best of our knowledge this was the first time that a level ( I) based on what the reinsured party can sophisticated contingent valuation methodology was absorb. Once an event exceeds this trigger the reinsur- applied to government infrastructure projects by cen- ance pays some fixed proportion of losses (L), usually tral government. up to some predetermined cap (Q on the reinsurer's A generalized form of contingent claims analysis exposure. If losses are less than the trigger, the insurer was used to evaluate three infrastructure finance pro- pays nothing. If losses fall in the range between the jects: A toll road project (El Cortijo-El Vino), a trigger and the cap, the insurer pays out the difference telecommunications joint venture (Telecom-Siemens), between the loss coverage and the trigger. If losses and an energy sector project (CORELCA). exceed the cap the insurer pays the difference between To value these transactions, the diffusion process the cap and the trigger. Using this basic structure, for all of the state variables underlying the risks in Lewis and Murdock (1996) show that the payout (1) each project was first specified. Yearly changes in of the reinsurance can be specified as follows prices (including exchange rates) and demand vol- 139 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE umes were assumed to follow a correlated lognormal of diversification, marginal risk analysis understates process. The frequency of losses as a result of event the risk exposure of each element (Merton and Perold risks (force majeure, counterparty failure, termination 1993). To compensate, any residual risk was allocated risk, and convertibility risk) was assumed to a binomi- to each risk category in proportion to the overall risk al distribution with a constant loss severity rate. The exposure (table 6.1). evolution of discretionary events, such as advertising The market risk exposure associated with traffic costs, was assumed to follow a uniform distribution. volatility and the risk of construction overruns were The means of these distributive processes were identified as the largest risks in the El Cortijo-El Vino derived from the best information available on each toll road project. The total expected loss to the project or from experience in other markets. Colombian government from these two guarantees Variance estimates were derived from an examina- was about $4.2 million. A small counterparty risk tion of the variability of the cash flows on each pro- associated with the failure of Corfigan, the reinsurer ject or, when unavailable, from the best market of the construction companies involved in the project, comparable. Covariance estimates between project was also identified. cash flows was based on best guesses or was assumed Regulatory/market risk and construction risk were to be zero. identified as the largest risks the telecommunication Using these estimates, the project used stochastic project. Regulatory/market risk exposure-stemming simulation techniques to identify the net expected loss from Colombia's deregulation of telecommunications, from each project. To provide a better understanding which ended the monopoly held by Telecom-was esti- of the decomposition of risk exposures within each mated at $10 million. The second largest risk in the project, the study also tried to analyze the marginal venture was construction risk, estimated at $9.8 million. increase in the governments exposure associated with Whether this risk is borne by Telecom or Siemens is not bearing each additional type of risk. Given the impact clear from the contracts. Telecom has nominal responsi- Box 6.3 Providing support to the Barranquilla power plant expansion in Colombia The government of Colombia supported the $755 million power purchase agreement, along with the right to expansion of the 240 megawatt Barranquilla thermal future revenues from the power from the TEBSA power plant in various ways. The new 750 megawatt plant, to the government in the event that COREL- plant will be constructed by TEBSA to provide power to CA fails. CORELCA. TEBSA, Termobarranquilla S.A., is a special * The Colombian government then provides a guaran- purpose vehicle, capitalized by the old Barranquilla ther- tee that FEN will be able to honor its commitment to mal plant, now jointly owned by CORELCA and ABB make payments under the CORELCA power pur- Distral. CORELCA is an undercapitalized, state-owned chase agreement if CORELCA defaults. power distributor on Colombia's Atlantic Coast that runs * To prevent CORELCA from failing FEN takes a sub- a narrow-margin energy distribution service. ordinated debt position in CORELCA to help ease a Government support in the expansion of TEBSA con- short-term liquidity crisis that would have forced sists of a power purchase agreement between CORELCA CORELCA into insolvency. and TEBSA, three guarantees, and a subordinated loan. * Ecopetrol, the supplier of gas to TEBSA and COREL- * CORELCA enters into a power purchase agreement CA, guarantees force majeure payments. with TEBSA, under which CORELCA agrees to The government's exposure in the CORELCA energy make capacity payments to TEBSA for the first twen- project was estimated at $67 million. In this project, the ty years of the plant's operation. As long as the plant Ministry of Finance used guarantees and subordinated is operational CORELCA has to pay a schedule of debt to support a marginally profitable energy distributor fees that start high and decline over time. (CORELCA) that, in turn, supported the development of a * The Ministry of Energy then guarantees CORELCA's new thermal power plant through a power purchase ability to make these capacity payments to TEBSA in agreement providing twenty years of capacity payments. the case of a CORELCA default. That is, FEN essen- Most of the government's exposure originated from the tially underwrites a put option giving CORELCA the fact that retail energy prices may be insufficient to support right to put the capacity payments issued under the CORELCA's operations, causing CORELCA to default. 140 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS bility for obtaining the licenses necessary to install new publishing it in the national budget, using it to estab- lines and assign operational lines to customers, while lish exposure limits or credit limits, or using it to Siemens bears the responsibility for installing the lines develop risk-adjusted performance measures. Such and switches. However, when the contract was revised measures could be used to reward programs that deliv- early in the project to account for initial delays, Siemens er social benefits with the least risk to the public bud- was held harmless for any construction overruns. Thus, get. If, for example, two programs yield the same social it was unclear which construction risks Siemens would benefits and the same expected costs, the program with actually bear in the future. the smaller variability in cash flows should receive The loss variances for each project were also ana- more budget funding and be subject to less oversight. lyzed, and scenario analyses were run to monitor how For private companies risk-based performance the risks of each project changed under varying condi- measures often attempt to measure the return generat- tions (figure 6.4). ed by a particular product line relative to the amount Scenario analysis is an extremely important tool as of capital that the product line places at risk. That is, governments review their exposure to a project finance companies look at the risk-adjusted return on capital. transaction in the context of more general fiscal poli- For national governments the driving mechanism is cies. In the toll road project, for example, such analy- the budgetary process, and risk management must sis can reveal the impact of anti-inflationary fiscal pol- focus on how the budgetary process can be improved icy on the government's exposure to traffic volume to provide stronger incentives for risk management. guarantees. Scenario analysis is also useful in analyzing Many governments face significant legal, regulato- alternative approaches to perfecting the government's ry, and political hurdles in moving from current bud- interest in a particular infrastructure project. For getary practices to a full accounting of the risks of con- example, along with a Power Purchase Agreement, the tingent liabilities. Implementing risk-adjusted perfor- Colombian Government took a subordinated interest mance measures allows governments to manage their in CORELCA. As a result, any action that is designed exposures to contingent liabilities even if an immediate to increase the value of the energy guarantee also must change in national budgetary policy is not feasible. be evaluated based on its impact on the value of the Nonbudgetary control mechanisms for contingent lia- subordinated loan granted to CORELCA. bilities could be employed during a transition to a new budgetary system, on a permanent basis for liabilities Risk-Adjusted Performance Measures: A Transition grandfathered during a change in budgetary policy, or to Budgetary Control of Risks as a permanent management solution if the govern- t^o Budgetary Control of Risks ment failed to enact a change in the budget law. These Once its risk exposure is quantified the government alternatives include publishing information on govern- can use the information as a control mechanism by ment exposures, establishing credit quotas (exposure TABLE 6.1 Expected governrment losses in Colombian infrastructure projects (millions of U.S. dollars) El Cortijo-EI Vino toll Telecom-Siemens CORELCA Type of risk road project joint venture energy guarantees Market risk 3,100 2,500 52,000 Construction risk 1,100 9,800 0 Counterparty risk 250 100 5,000 Currency risk 0 -1,300 2,000 Force majeure 200 300 7,000 Termination risk -150 200 1,000 Regulatory risk 0 10,100 0 Total 4,500 21,700 67,000 141 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE FIGURE 6.4 expected costs of a product or liability in the year in Sensitivity analysis for the Colombian toll road which the liability is issued, discounting to reflect the project time value of money. Private institutions compute vir- Percent of total expected loss tually all investment decisions, expenditures, plans, 1.5 T and budget forecasts on a present value basis. The use variance of present value accounting is especially important in 1.2 areas in which private firms are required to mark-to- market the value of assets or liabilities. But, even 0.9 / where assets and liabilities can be carried at cost, or 0.6 / Tariff w book value, present value budgeting helps in establish- / varii ing loss reserves or a capital budget. 0.3 In contrast, most government bodies account for /_^ C~~C orfigan credit and insurance products using a simple cash- 0.0 - based system of budgeting. Under a cash-based sys- -0.3 tem of budgeting, a government equates the bud- -1 0 1 2 3 getary cost of issuing financial assistance with the Standard deviation cash outlay created by the transaction in the current budget year. Thus when a government issues a direct limits), and earmarking future funds to cover guaran- loan, the entire face value of the loan at the date the tee costs. Similar performance measures can be devel- loan is issued is recorded as a budgetary cost, with oped to meet the needs of countries looking to make loan repayments recorded as cash inflows in subse- incremental changes to their budgetary policy or as a quent years. Simple cash-based budgeting thus treats mechanism for smoothing the transition to a full bud- the disbursement of a direct loan as a grant equal to getary accounting of contingent liabilities. the entire face value of the loan, with subsequent repayments representing offsetting receipts for the government. Loan guarantee and insurance programs Budgeting for Expected Costs are not recorded as costs in a simple cash-based bud- get until a claim is made at some future uncertain Governments need to make risk-return trade-offs when date. In fact, since any premium revenue from a gov- deciding which programs to fund each budget year. ernment insurance program is recorded up front in While these decisions seek to maximize risk-adjusted exchange for the insurance policy, while claims are social returns rather than financial returns, a clear not recorded until some uncertain date in the future, understanding and accounting for program costs and a simple cash-based budget may record an insurance risks is critical in making these decisions. Unlike private program as a net revenue gain. This inconsistency sector corporations few governments set aside bud- creates a budgetary incentive for policymakers to getary resources to cover the full expected costs of raise premiums rather than reduce the likelihood or financial guarantee or insurance programs, a far simpler severity of claims insured. Cash-based budgeting task than establishing reserves to cover unexpected thus misrepresents and masks the aggregate exposure costs. Instead many governments choose to budget only associated with loan guarantees and government for expected cash outlays associated with a guarantee or insurance programs and creates perverse incentives insurance program in the next budgetary period. for selecting one form of financing assistance over another. Present Value versus Casb Budget To see how these incentives skew decisionmaking, consider the different ways in which the government Private companies, especially banks and other finan- could help finance a $100 loan to a private infrastruc- cial institutions, tend to recognize the present value of ture provider. If the government provides a 10 percent 142 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS loan subsidy, the cash budget cost would be $10 in reserves up front for the expected costs of the guaran- year one. If, instead, the government provided the tee issued. loan directly, the cash budgetary cost in year one would be $1 00-the full face value of the loan. If the government agreed to guarantee a loan made by a pri- vate bank, the budgetary cost of the guarantee would To discount nominal cash flow streams to compute be zero (or negative if a guarantee fee is collected) the the present value of expected losses, private companies first year. Thus, while the economic and financial typically follow one of two procedures. Under one value of the three different forms of financial assis- approach projected cash flows are discounted using a tance are equal, a legislative body would favor the risk-adjusted discount rate based on the firm's cost of guarantee option. capital. (For more on industry cost-of-capital esti- Even more problematic, by not accounting for the mates see Fama and French 1997.) Under a second budgetary costs of issuing guarantees, a simple cash approach risk-adjusted cash flows are discounted budget encourages the expansion of guarantee liabili- using a risk-free rate of interest, usually proxied by the ties without requiring the government to reserve short-term U.S. Treasury bill rate, LIBOR, or against future losses. Without budgetary control these overnight interest rates. contingent liabilities grow, and the government's In computing a government's exposure to credit exposure to sudden increases in future budgetary costs and insurance programs using contingent claims increases. These unanticipated increases will raise gov- analysis, the second approach is used, and no consid- ernment deficits, require a realignment of budgetary eration of the appropriate risk-adjusted discount rate expenditures away from future expectations, and cre- is needed. In cases in which cash flows do need to be ate an enormous political backlash against the govern- discounted by a risk-adjusted interest rate, the govern- ment's guarantee programs. ment must determine the appropriate discount rate By not aligning the budgetary impact of direct policy. In the United States, the government uses its loans, loan guarantees, insurance, and grant programs cost of funds as the discounting factor (as reflected by with their true economic costs at the time commit- the U.S. Treasury rate with the same maturity as the ments are made, a simple cash budget creates an loan guarantee or direct loan).3 The alternative intertemporal myopia and/or moral hazard. Tracking approach considered in the United States was a "bene- the cost of guarantee claims only as the claims are fit-to-borrow" approach, in which the discount factors incurred as opposed to when the commitment was would be computed from the discount rates used by made enables political leaders to provide financial private sector agents when computing the benefits of assistance without having to account for the costs of the government program. The problem with a risk- providing the assistance, which will be realized under adjusted discounted rate approach, especially for con- ensuing administrations. This form of myopia can tracts with embedded options, is that the appropriate quickly lead to an escalation in guarantee costs as discount rate becomes a function of the riskiness of ensuing administrations increase their financial assis- the contract payouts. tance to favored parties. Only by enforcing budgetary controls at the time the financial assistance is commit- The Federal Credit Reform Act of 1990 ted can the appropriate budgetary incentives be realigned to eliminate this moral hazard. Prompted by the explosion of loan guarantees issued Use of a present value system need not affect or during the 1980s and a recognition of biases created distort cash-based estimates of the government's fiscal by a simple cash-based system of budgeting, the deficit, since the effect on the deficit is not recorded United States changed the budgetary treatment of until actual cash payments are disbursed from the direct loans, loan guarantees, and grants in 1992.4 reserve fund. Adoption of a present value method of Under the new budgeting system created by the guarantee budgeting simply forces agencies to set aside Federal Credit Reform Act of 1990, each of these 143 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE forms of credit was valued using a financially equiva- form of financial support on the basis of the underly- lent metric-the expected present value of future ing needs of the targeted population rather than on costs. The budgetary cost of credit is defined as the the specific budgetary treatment of alternative finan- present value, discounted at Treasury interest rates of cial structures. comparable maturity, of the expected cash outflows The Federal Credit Reform Act does have short- from the government minus the expected cash inflows comings, which provide useful guidance for future to the government.5 The shortfall between borrower budgetary reforms in the United States and elsewhere. fees, repayments, and interest and the amount needed First, coverage of unexpected losses is not included as to cover the principal of the loan and the Treasury's part of the cost of a program. This failure to incorpo- cost of borrowing represents a cost to the government. rate some measure of unexpected loss represents a seri- Likewise the difference between the fees borrowers ous shortcoming given that most loss distributions pay to the government and the cost of guaranteed associated with central government guarantees are loan defaults (and/or interest subsidies) represents a asymmetrically skewed against the government. cost. When agencies seek budget resources (budget Second, incentives remain to use "cheap" insurance authority and budget appropriations) to carry out a structures to cover loan guarantees. Government credit program in the budget process, they must esti- insurance programs are financially equivalent to guar- mate and request the full expected present value of antee programs and should be treated in a consistent future costs-including default, interest, and other budgetary framework. Third, program agencies must costs-associated with loan guarantees or direct loans make substantial investments in new information sys- to be issued in the forthcoming budget year. Funding tems technology. In the United States, new invest- to cover the expected present value of future costs is ment in information systems placed a strain on many charged against the appropriation for an agency when of these agencies. Governments adopting credit the direct loan or loan guarantee is issued and the gov- reform must recognize at the outset that funds need to ernment's commitment is extended. These costs, or be available for this investment. Finally, credit reform subsidies, must compete for budgetary resources on requires that agencies reestimate the subsidy costs of the same basis as other government spending. their programs on a regular basis so that the govern- Credit reform requires more careful record keep- ment's exposure can be recalculated and appropriate ing than a simple cash budget. Agencies must identify funding is set aside to cover future costs. Appropriate loans or classes of loans by the appropriation used to discipline is required to ensure that agencies do not fund the transaction, their maturity and date of origi- underestimate subsidy costs with the knowledge that nation, and their subsequent cash outflows and any shortfall will be made up in someone else's watch. inflows. In addition, programs are required to develop Learning from the experience in the United States, risk categories based on the characteristics that deter- New Zealand has implemented a similar budgeting mine the likelihood of default and other costs. These approach. Their program covers all contingent liabili- records are used to reestimate the value of the subsi- ty exposures (including insurance), and the govern- dies provided for loans or loan guarantees, adjust ex ment has published a present value budget for both post budgetary expenditures relative to ex ante expec- contingent and noncontingent expenditure and rev- tations, and improve the subsidy calculations for new enue flows. loans or guarantees. This tracking also helps agencies underwrite, service, and control losses on loans or guarantees. Risk Preferences and Reserve Policy The Federal Credit Reform Act significantly improved the budgeting process in the United States. In addition to budgeting for the full expected present By revealing the true fiscal implications of direct value of costs from credit and insurance programs, gov- loans, guarantees, and grants, the new budgeting sys- ernments need to set aside reserves against unexpected tem allows policymakers to make decisions on the losses. Preparing for unexpected losses prevents the 144 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS political backlash associated with redirecting scarce tools are in place the government can set reserve poli- public resources to cover the sudden increase in costs, cy based on an assessment of its aversion to making obviates the need for political battles over additional frequent requests for funding to the legislature. funding, and eliminates the perception that any sudden Distributions of potential guarantee payouts are increase in costs represents program mismanagement. complex. Rather than specifying a probability thresh- Setting up reserves to protect against such events old in terms of the probability of default, which can mitigate these problems by reducing the number would be unwieldy, common practice is to describe it of events for which the executive branch or adminis- in terms of the standard deviation of losses. tering agency needs to seek additional budgetary Depending on the type of distribution, deciding on resources to cover program costs and by reducing the the appropriate multiple of the standard deviation as size of any budgetary requests that are made. Because the threshold will result in a particular default fre- the United States government did not reserve against quency.6 Many companies set capital and reserves to unexpected losses, it incurred high political costs as a cover a two or three standard deviation movement in result of the $130 billion in losses charged to U.S. tax- their underlying risks. payers during the thrift crisis of the 1980s. Another important factor in determining the level When a private corporation examines its exposure of reserves is the government's leverage preferences, to risk, its management committee must determine the that is, the opportunity cost of holding funds in amount of capital and reserves that the company wish- reserve as opposed to spending the resources on pro- es to hold in excess of expected costs to cover unex- grams. On the one hand, holding more funds in pected losses. For an institution with multiple lines of reserve will increase the liquidity of the guarantees that business, determining the appropriate level of capital the reserve supports, increasing the value of the guar- or reserves is a complex procedure that takes into antee and allowing the government to leverage more account both the variability of losses for each product private sector funding in the guarantee program. On line and the correlation between product returns and the other hand, reserving funds in a separate account the opportunity cost of capital. Management must also reduces the amount of money available for other pub- weigh the expectations of the company's shareholders lic sector projects. If the net benefits of additional pub- and stakeholders, rating agencies, and its business part- lic spending exceed the liquidity benefits of adding to ners in determining an optimal level of capital for the guarantee reserve, the government may want to maximizing shareholder value. The level of capital or direct additional funds toward public spending. reserves held by an enterprise reflects its relative risk When a private company assesses this trade-off aversion and its ability to withstand a specific level of between holding reserves or investing in other pro- unexpected losses. Thus a firm seeking a AAA rating grams, it usually has a targeted economic return that will hold considerably more capital against unexpected helps guide its capital policy. For a government the losses (say, capital to cover a 99 percentile event over a comparable concept is social economic return. 1-year period) than a firm seeking an A rating (capital Calculating social economic return requires a com- to cover a 90 percentile event). plete asset-liability management program that goes beyond the valuation of infrastructure liabilities or Determining the Aversion to Unexpected Losses other forms of direct loans, loan guarantees, and insurance. This chapter focuses solely on reserving Setting aside reserves for unexpected losses reduces the against contingent liabilities without considering a frequency with which the executive branch needs to broader asset-liability management policy. go to the legislature for special appropriations or a special incomes bill. If the government wants to go to the legislature only once every thirty years for a given guarantee, it needs to find the level of loss protection Once a government can assess its risk tolerances and that will allow it to do so. Once the proper valuation goals, in terms of both which risks and the level of loss 145 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE it is willing to bear, it can establish reserves against factors and the joint probability distributions of unexpected losses ('risk capital") within its credit and changes in the underlying market factors affect the insurance programs.7 To do so, however, a government level of risk capital in a portfolio of risks. Examining needs to determine whether reserves will be set based these two elements allows the maximum possible loss on the additive unexpected loss exposure of each guar- within a known confidence interval to be determined antee or on a portfolio value-at-risk approach to over a given time horizon known as the portfolio's account for portfolio diversification, what the invest- Value-at-Risk (VaR). For private financial institutions ment policy of the reserves will be once they are estab- a variety of approaches are used to calculate portfolio lished, and where the reserves should reside. VaRs. The most widely referenced, although not the best, model is the RiskMetricsTM model, published by Additive versus Portfolio Reserve Requirements. The J. P. Morgan (1996).8 first decision that a government needs to consider Specifying the position sensitivities and the under- when setting up a reserve for unexpected losses is the lying variance-covariance matrix of market rate inno- measure of unexpected loss against which to capital- vations is a nontrivial exercise and requires a number ize. Under an additive reserve standard the govern- of simplifying assumptions even for actively traded ment calculates the unexpected loss exposure of each securities. For example, portfolio-normal VaR of its contingent liabilities (that is, examines the sensi- approaches assume that portfolio returns are normally tivity of each guarantee valuation to changes in the distributed. RiskMetrics™ and Delta-Normal underlying factors) independently. Then for a given approaches assume that asset returns are jointly nor- confidence level and time interval it determines the mally distributed, implying linear asset payoff profiles amount of unexpected loss it wishes to cover for each and normally distributed portfolio returns. Delta- guarantee, taking into consideration the opportunity Gamma methods assume that innovations in market cost of capital. The government then identifies the rates are normal, but that payoff profiles are approxi- average cash reserve required to fund these unexpected mated by local, second-order terms (Wilson 1997). losses. Finally, the individual cash reserve balances are Many of the criticisms of VaR models deal with the aggregated to arrive at a total unexpected loss reserve. reasonableness of these simplifying assumptions for a This additive approach for setting capital or unexpect- given application as opposed to the underlying ed reserves is supported by bank regulatory capital approach. Wilson (1997) and Duffie and Pan (1997) standards for financial institutions. provide a good summary of the advantages, disadvan- The problem with the additive approach for set- tages, and common critiques of different VaR ting unexpected loss reserves is that it fails to account methodologies. for portfolio diversification-the fact that pooling imperfectly correlated risks will reduce the variance in Value-at-Risk for a portfolio of infrastructure liabili- the expected loss of a portfolio. As a result the risk of ties. Government infrastructure guarantees can be ana- the overall portfolio will be overstated, and more pro- lyzed as contingent claims, and a VaR model can be tection against unexpected losses would be provided applied to government infrastructure liabilities. The than originally sought by the government (Merton shortcoming of most VaR approaches, including and Perold 1993). The alternative is to calculate the RiskMetricsTm, in evaluating the risks associated with aggregate loss distribution of the government's portfo- a portfolio of options is their failure to reflect the non- lio of risks, using a value-at-risk approach that incor- linear payoff functions of options. Most of these porates cross correlations between guarantee exposures approaches would thus not be suitable for calculating and then set reserves to cover unexpected losses based the VaR associated with a portfolio of infrastructure on the unexpected loss profile of the whole portfolio. liabilities. One VaR approach that attempts to incorporate the Value-at-Risk Methodologies. The sensitivity of the nonlinearity in options portfolios is the Delta-Gamma value of a portfolio to changes in underlying market approach (Wilson 1997). Unlike Delta-Normal 146 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS approaches such as RiskMetricsrm, the Delta-Gamma of policy variables to be analyzed to assess the impact approach uses a second-order Taylor series approxima- of different policy actions on the value of an existing tion of a portfolio's value function around current mar- guarantee or infrastructure liability program. ket rates to incorporate direct and cross-market convex- ity risk (the rate of change in the value of an instrument Investing reserves. One important issue in structur- given an incremental change in the underlying asset's ing reserves for unexpected loss is the investment poli- price) and vega risk (the change in the value of an cy of the reserve fund. Should the reserves be invested instrument given an incremental change in the under- in government debt securities, corporate debt, equi- lying asset's volatility). Convexity and vega risk repre- ties, or some combination? This issue has been hotly sent two of the more important risk factors in a portfo- debated in the United States, where government agen- lio of options. Assuming that market rate innovations cies typically to invest all reserve fund assets in U.S. have a joint normal distribution, the Delta-Gamma Treasury securities. Recently, many federal agencies approach solves for the VaR in a portfolio of options by have asked to be allowed to invest a portion of their searching for the market rate events that result in the reserve funds in the stock market, arguing that invest- worst VaR within a given confidence interval. As such ing in the stock market would allow them to accumu- the approach can be a useful tool for banks computing late larger reserves. the VaR of an options portfolio over short periods of One serious problem with investing reserves in the time. stock market is that funds may not be available when The approach is less useful for analyzing the VaR needed to cover losses. Consider, for example, a of government guarantees over longer time intervals, reserve fund established to pay for unexpected losses since it analyzes only how local changes in the under- on government guarantees against interruptions in lying market rate factors affect the value of an options housing construction that is invested in S&P 500 portfolio. This approach may be reasonable for com- stocks. Given its sensitivity to interest rate move- puting the one-day or two-week VaR of a financial ments, construction activity is very cyclical, falling options portfolio. It is considerably less useful for sharply during economic downturns. As construction examining the unexpected loss exposure of infrastruc- activity falls construction company earnings drop, ture liabilities over many years. increasing the probability of a company failure and a A powerful alternative VaR approach that can pro- major interruption in construction activity for pro- vide a more accurate depiction of the government's jects supported by a government guarantee. The per- longer-term risk exposure is using contingent claims formance of the construction industry is also highly analysis in concert with stochastic simulation and sce- correlated with the S&P 500 (the industry beta is nario analysis. Given an accurate contingent claims about 1.25). Therefore, if the probability of a call on model and the "true" specification of the process gov- the government's construction guarantee rises, the erning changes in the price of the underlying asset, value of reserve funds invested in the S&P 500 will Monte Carlo analysis can be used to examine the sen- fall, decreasing the value of the reserves. The more sitivities of infrastructure liability exposures to small severe the economic downturn, the more likely the and large movements in the underlying risk factors. government's guarantee will be exercised and the more Monte Carlo simulation is not commonly used by likely that the value of the reserve funds invested in financial institutions because of the massive comput- equities will be insufficient to cover unexpected losses. ing resources required to evaluate a large portfolio of In this example investing the reserve funds in equities financial options. In analyzing infrastructure liabili- actually decreases the value of those reserves compared ties, however, where the number of government guar- with investing in short-term government securities. antees outstanding in any one portfolio is more limit- The objective in investing the reserve fund should ed, Monte Carlo simulation techniques can be very be to maximize the value of the assets in the fund when effective. The combination of contingent claims pric- the costs to the government increase-that is, to invest ing and Monte Carlo simulation allows a richer array the reserve funds in assets that provide the best hedge 147 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE against the government's cost for a given return. Complementary Measures for Reducing Risk Investing the reserve fund in assets whose value is nega- tively correlated with the value of the guarantee requires Designing clear contracts, introducing incentives to very active asset management strategies, however. The reduce the incidence of calls on guarantees, and regu- government may be better served by managing its assets larly monitoring performance under the guarantee and liabilities at the balance sheet level rather than on a contracts can limit the government's exposure to risk. per program basis. To do so, the government would invest all reserve fund assets in government securities DesigningAppropriate Contracts with the same durations the loan guarantee, direct loan, or insurance programs for which the reserve is estab- A comprehensive risk management system forces lished. The government could then hedge its net bal- agents to critically assess the distribution of risks with- ance sheet position with investments that limit its expo- in a particular direct loan, guarantee, or insurance sure to those macroeconomic risks that the government program (box 6.4). The recent debate over the provi- deems consistent with public policy. Of course, invest- sion of catastrophic disaster assistance in the United ing in government securities in this manner is equiva- States highlights the importance of a comprehensive lent to reducing the government's gross debt position, risk management system. Over the past five years implying the need to examine reserves policy as a gov- insurance companies in the United States have recog- ernmentwide asset-liability issue. nized that they are overexposed with respect to prop- Investing construction guarantees reserve fund erty damage from natural disasters. Recognition of assets in government securities would provide a hedge this overexposure led to many early legislative initia- for the government, since rising interest rates would tives by the insurance industry calling for the federal cause the value of the construction guarantees (and government to provide direct insurance or reinsurance costs) to rise and the price of government securities to for disaster coverage. As the debate over the govern- fall. The government may thus find it advantageous to ment's role in disaster risk evolved and the issue was fund any guarantee costs by issuing new cheaper debt more narrowly defined as an incomplete market in the instruments-in effect, substituting for the securities intertemporal smoothing of large idiosyncratic risks, in the reserve fund. If all of the government's guaran- however, the U.S. government recognized that provid- tees are in an external currency, the government then ing a mechanism for financing only the higher layers could purchase currency forward to hedge against its of disaster losses provided a more targeted and effi- net currency exposure. cient solution (Lewis and Murdock 1996). The government also needs to decide if it will The process of comprehensive risk management hold its offshore in a foreign currency or domestically also forces a government agency to ensure that any in the domestic currency. In the case of project guarantee or assistance has clearly defined terms that finance guarantees the same logic that applied to the are aligned with the agency's management objectives. investment policy of reserves applies to the manage- The contracts in the Colombian Telecom joint ven- ment of foreign exchange risk. If the project finance ture allocate construction risks clearly. However, when guarantees are denominated in dollars, the govern- the contract was restructured after an initial construc- ment should consider investing the reserve fund in tion delay in the project, Telecom assumed all of the dollar assets and possibly keeping the reserve offshore costs-leaving Siemens with the same net present to circumvent convertibility risk issues. This policy value benefits as in the original contract. Management would greatly enhance the market's value of the guar- of the contract sent a signal to Siemens that Telecom antee and provide the government with greater lever- will bear a larger portion of the construction risks age from the guarantee program. However, any deci- than envisioned in the original contracts. sion on the location of the reserves must be made in When the management of government assistance the context of the government's broader foreign cur- deviates from the terms of the assistance being provid- rency risk management program. ed, the government is perceived to provide an implicit 148 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RisK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS guarantee. Although an implicit guarantee does not engineering specifications (a public good provided contractually obligate the government to provide free) and then provided a more narrowly defined guar- assistance, where the public believes the government antee, thereby obtaining a more targeted structure. will step in to provide assistance when needed an Because guarantees and insurance can be narrowly tar- implicit guarantee becomes explicit. Examples of geted they can be used to get the private sector to implicit guarantees abound, including the "too-big-to- absorb as much risk as possible. fail" and 100 percent depositor protection concepts Where the private sector is better able to under- for deposit insurance in the United States and federal write and service the underlying risks but some gov- support of government sponsored agencies in most ernment assistance is needed, public-private risk- countries. sharing is often the best solution. In this case pro rata Faced with implicit guarantees the government guarantees and insurance in which the private sector should either make the guarantee explicit and manage and the government share all losses on a particular risk the guarantee as an assumed risk or explicitly deny equally are often the best form of assistance, since the any obligation and willingness to provide assistance firm shares an equal percentage of the losses across all when needed. By doing neither the government rein- types of risk. Risk-sharing provides the private entity forces the perception that an implicit guarantee will with an incentive to price the coverage appropriately, be honored and increases the political pressure to sup- ensuring the government that the private sector will port the provision of government assistance in the not shift additional risks to the taxpayer. event that support is needed, while maintaining no Other risk-sharing mechanisms within and control over the management of this conditional between classes of risk are also feasible. However, they exposure. (For more on implicit guarantees see Kane usually require more government oversight and more 1996). government underwriting expertise. The government must first assess which party (public or private) has the best access to information Box 6.4 needed to objectively and most accurately assess the Improving risk management riskiness of the underlying risks. The government on the Colombia toll road project must then assess which party is in the best position to monitor, control, and service the risks once they are In soliciting bids for the Colombia toll road project the government asked prospective concessionaires to underwritten. If the government is in the best posi- bid on construction projects based on only a prelimi- tion to underwrite the risks directly, direct credit nary set of engineering designs. Recognizing that these should be considered, with credit assistance targeted designs provided insufficient detail, the government granted cost overrun guarantees that would compen- to the area of concern. The government should then sate the concessionaire for cost variances within a determine whether it also has the information and wide band around the submitted bid. While the guar- antee served the purpose of attracting qualified bid- skills to most effectively monitor and control the risks ders, the structure of the guarantee allowed the con- or whether a private servicer should be employed to cessionaire to extract a near certain rent from the gov- service the loans. Where the government delegates ser- ermnent of about 35 percent of the original bid costs. service the loans. Where the government delegates ser- After critically assessing the risk transfer associat- vicing, it must have the systems for monitoring the ed with these toll road projects and quantifying the performance of the servicers. risks in the El Cortijo-EI Vino project, the Colombian government changed its toll road guarantee program. Even if the government has the best access to The government now commissions more detailed engi- information on a particular risk, it may choose to pro- neering studies before it solicits bids to limit the uncer- tainty inherent in the bidding process and provides a vide assistance in the form of a guarantee targeted at a narrow guarantee. The new policy was less expensive specific layer instead of providing direct credit, since a than the old one but provided the same benefit to the contingent guarantee can be more narrowly focused at concessionaire. The change made the Colombian toll road project more efficient-delivering a higher risk- the market failure. In the Colombian toll road, for adjusted rate of return by reducing the government's example, instead of providing direct financing for the risk of delivering a fixed benefit. toll road construction, the government purchased the 149 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE The tools and techniques associated with risk government. They can place restrictions on the use management are also helpful in analyzing the struc- and investment policy of reserves that are held by the ture of government programs that share responsibili- guaranteed party to ensure that the value of the ties between the federal and state level. In the United reserves is unimpaired during periods in which a loss States, for example, several programs combine the event is likely. They can structure the government's national government's ability to redistribute resources support to promote pro rata risk-sharing, where a pri- across economically diverse regions with the ability of vate party shares risk equally with the government for state and local governments to identify investment some, or all, types of loss. Since the private party in needs at the local level. The national government this transaction bears the same risk per dollar of expo- funds the program, while state and local government sure as the government, public-private risk-sharing provide the underwriting and administrative function. allows the government to benefit from the private sec- This federal-state partnership is a potentially powerful tor's pricing of risks. Finally, the government can levy combination that is analogous to a parent company risk-based guarantee fees that both reduce the bud- providing a guaranteed source of financing to a sub- getary cost of issuing guarantees and improve the sidiary established to perform a particular service. alignment of incentives between the guaranteed party Such federal-state partnerships are not without and the central government. (Fees can be estimated risks, however. If the federal government providing using the techniques identified earlier, including con- the funds does not retain oversight of the underwrit- tingent claims techniques.) ing function, the national budget remains at risk. But Limiting the ability of private agents to shift addi- if the federal government is overly prescriptive in set- tional losses to the government reduces the budgetary ting regulations for the program, the flexibility of the costs of issuing guarantees and enhances the allocation state and local entity to identify specific needs in the of scarce budgetary resources by limiting rent-seeking local community is reduced. The goal is to reach the behavior. optimal trade-off between the delegation of project selection and federal oversight of the underwriting Monitoring Performance and Reestimating Risks performance of the state facility (box 6.5). Once the government has implemented budgetary the Frequency and Financial Impact of and reserves systems for its contingent liabilities and Minimizing ' ' decided which risks it chooses to cover, it should com- Calls on Guarantees... umunicate these decisions and risk management guide- Governments need to implement strong risk manage- lines to the agencies responsible for implementing the ment programs to limit their contingent liability guarantee programs. The government should evaluate exposure to additional loss shifting by the guaranteed the performance of agency personnel based on their party. Mitigation actions attempt both to reduce the ability to meet these goals. In this way the govern- frequency of the government's losses and to minimize ment can obtain a proper alignment between govern- the financial impact of those guaranteed events that ment risk management objectives and the perfor- do occur. Risk controls attempt to minimize the abili- mance of the agencies administering the programs. ty of the guaranteed party to shift additional risk to To be effective comprehensive risk management the government (through moral hazard, adverse selec- system must implement systems for monitoring the tion, and other forms of distribution shifting). changing risk exposure of its portfolio. As experience Governments can reduce their contingent liability has shown, techniques for assessing risk are only as exposure to risk in many ways. They can require the good as the information on which the models are guaranteed party to hold a certain amount of capital based. Over time institutions change, markets evolve, or collateral to serve as a first-loss protection barrier and new information on risk exposures is obtained. In for the project, thereby aligning the guaranteed party's many instances risks that were previously unknown or incentives to remain vested in the project with the unquantifiable are revealed through a series of loss 150 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS Box 6.5 Revitalizing urban areas through federal-state partnerships in the United States Under the Community Development Block Grant program has enabled state and local govemments access (CDBG) program in the United States, states and local to up-front financing for their development projects. communities receive federal grants to help finance com- The federal government used two risk mitigation munity development projects designed to transform techniques. First, it collateralized the loan guarantees with abandoned urban neighborhoods into viable local com- a state's ability to receive future block grant funds. If a munities for mixed-income households. The program state triggered the federal guarantee by defaulting on a also provides grants to support the financing of capital loan, the state would lose access to all future block grant projects designed to create new centers of economic funds until the loss on the loan was repaid by the federal activity in areas that have become economically government using that state's future block grant money. depressed. Opportunities for investment are selected by The federal government also established strict criteria, the state and local community and are financed with the based on project underwriting performance, for which federal grants. states and local communities could participate in the loan The CDBG program is an important part of the guarantee program. These oversight functions were seen effort to revitalize urban centers in the United States. as critical to the success of the overall program. State and local communities were often unable to use the The CDBG program has its shortcomings. Since the CDGB program, however, because investments in local funds are not an entitlement, collateralizing the loan community revitalization required a large up-front infu- guarantee against a discretionary source of sion of funds, not a gradual annuity in block grant fund- Congressional funding provides a very limited form of ing. To address this concern the U.S. Congress enacted collateralization. Nevertheless, the program provides a the Section 108 Community Development Block Grant good illustration of the power of federal-state partner- Loan Guarantee Program, which provides federal loan ships, the problems faced in structuring a federal-state guarantees on development loans obtained by the state risk-sharing agreement, and the risk management tools and local communities for economic development. The needed to assess the risks of each entity's exposure. events, leading to radical changes in risk assessment. Federal Credit Reform Act also required agencies to The governments thus need to have a systematic implement systems for computing reestimates on a approach for quickly incorporating new information timely basis as part of the overall budget process. on its changing risk exposures into its pricing of new Under this system federal agencies are supposed to contingent liabilities and for making adjustments to reassess the expected costs of each year's activity in the expected costs of previously issued liabilities (rees- their credit programs at regular intervals and use this timates). Development of these systems improves the information to alter the expected costs for newly government's accounting of expected loss and limits issued direct loans or guarantees.9 Furthermore, if the opportunities for moral hazard, adverse selection, and change in the expected costs of previously issued other means of shifting additional risk to the central direct loans or guarantees is significant (that is, it devi- government. ates from prior estimates by more than 5 percent), the To comply with the tracking requirements man- agency is required to seek additional budgetary dated by the Federal Credit Reform Act, government resources to fund the additional exposure. Similar sys- agencies in the United States were forced to update tems can be implemented in other central govern- their outdated budget and accounting systems. This ments for all forms of contingent liabilities. improvement in information processing and tracking systems led to a substantial increase in both the quan- tity and quality of information available on govern- Conclusion ment programs-information that policymakers have used to guide future reforms or program develop- The explosion of infrastructure liabilities has created ment. Although the costs associated with implement- the need for risk management techniques with which ing new risk management systems were significant, to manage governments' exposure to contingent liabil- the benefits associated with better risk processing sys- ities. Because guarantees involve no immediate cost to tems are believed to have exceeded the costs. The the government, they do not appear in the govern- 151 DEALING WITH PUBLIC RISK IN PRIVATE INFRASTRUCTURE ment accounts, and funds are not budgeted to cover the variance of the expected claims distribution. However, a them. This failure to account for guarantees leaves more accurate approach also would incorporate higher governments vulnerable to large unexpected demands moments of the loss distribution since the actual loss distri- on their resources. It also skews government decision- bution associated with many risks are not asymmetrically making in favor of guarantees over subsidies, since distributed around the expected value (or mean). Note, this guarantees require no legislative approval and approach is comparable to a discounted cash flow analysis, funding. where r is the risk-free rate of interest. Quantifying the value of guarantees using enter- 3. Technically, a more appropriate implementation of prise risk management techniques can help govern- this concept would use the Treasury rate with a maturity ments reduce risk, improve project and contract comparable to the duration of the federal guarantee or design, and reduce the incentive to offer guarantees direct loan. rather than subsidies. Moreover, by budgeting for 4. The Federal Credit Reform Act did not change the expected losses and setting aside reserves against unex- budgetary treatment of insurance programs, creating a clear pected losses, governments can avoid potentially seri- inconsistency in the U.S. budget. However, the Office of ous fiscal problems and the political backlash that Management and Budget in the Executive Branch has occurs when contingent liabilities come due. endorsed budgetary reforms designed to end this anomaly By implementing an economywide risk manage- and the Congressional Budget Office, the General ment system, governments can manage risk from all Accounting Office, and the Congressional Research Service sources of revenue and expenditures as part of a have all acknowledged that putting insurance programs on a broader risk management strategy. Adopting such a consistent basis is the next major budgetary reform. system will provide governments with a valuable tool 5. Note, by using the Government's cost-of-funds to with which to better allocate scarce resources and risk discount expected future costs, the United States creates a within the economy. disconnect between the budgetary costs of a program and the costs that should be estimated as part of any benefit-cost analysis justifying the program, which would be estimated Notes using a (higher) private sector discount rate. 6. For a normal distribution, which is the most likely This paper may not represent the views of Ernst & Young distribution for the overall portfolio exposure, the relation- LLP or the World Bank. The authors thank Clemente del ship between variance and expected distribution function is Valle and the editors for valuable contributions. The well known. authors retain the responsibility for all errors and omissions. 7. While we limit our discussion on the establishment 1. The risk-adjusted rate of return represents the differ- of reserves to all credit and insurance programs, given the ence between the rate of return earned on an investment focus on infrastructure liabilities, the principles discussed and the risk-free rate of interest less the market's premium here apply more broadly to all government risks. for bearing the risks associated with the investment. 8. As an illustration, J. P. Morgan's RiskMetricsTM VaR 2. For example, the general formula for an insurance formula can be expressed as follows: policy, which we demonstrate below as equivalent to a com- VaR = p; / I G [)t. bination of guarantees or options, can be expressed as fol- where p is a constant representing the desired one-tail confi- lows (Patrick 1990): dence interval for the standard normal distribution, I is the Premium = f(Expected Loss Distribution) N x N annualized covariance matrix of security (or guaran- (1 - EL)(1 - r) tee) returns, o is the N x I vector of portfolio position where EL is the insurer's expense loading-enabling it to weights, and At (or r) is the time interval expressed as a cover its administrative costs-and r is the targeted eco- fraction of a year (J.P1 Morgan 1996). nomic rate of return. The term f(.) prices the risk of the 9. Guidance provided by the Office of Management insured portfolio based on the expected loss distribution. At and Budget has indicated that regular intervals should a minimum, f(.) incorporates the mean expected loss and translate to every year for the first five years of a program 152 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS and then every fifth year after year five or when there has Lewis, Christopher M., and Kevin Murdock. 1996. "The been an identifiable material change to the risk exposure of Role of Government Contracts in Discretionary the program. Reinsurance Market for Natural Disaster." Journal of Risk and Insurance 63: 567-97. Marcus, Alan, and Israel Shaked. 1984. "The Valuation of References FDIC Deposit Insurance Using Option Pricing Estimates." Journal of Money, Credit and Banking 16: Black, F., and M. Scholes. 1973. "The Pricing of Options 446-60. and Corporate Liabilities." Journal of Political Economy Merton, Robert C. 1973a. "An Intertemporal Capital Asset 81: 637-54. Pricing Model." Econometrica 41: 867-87. Cooperstein, Richard, George Pennacchi, and E Stevens . 1973b. "Theory of Rational Option Pricing." Bell Redburn. 1995. "The Aggregate Cost of Deposit Journal of Economics and Management Science 4: Insurance: A Multi-Period Analysis." Journal of 141-83. Financial Intermediation 4: 242-71. Merton, Robert C. 1977. "An Analytical Derivation of the Cooperstein, Richard, F. Stevens Redburn, and Harry G. Cost of Deposit Insurance and Loan Guarantees: An Meyers. 1992. "Modeling Mortgage Terminations in Application of Modern Option Pricing Theory." Turbulent Times." AREUEA Journal 19: 473-94. Journal of Banking and Finance 1: 3-11. Duffie, Darrell, and Jun Pan. 1997. "An Overview of Value . 1990. Continuous-Time Finance. Cambridge, at Risk." Journal of Derivatives 4: 7-49. Mass., Blackwell Publishers. Fama, Eugene, and Kenneth French. 1997. "Industry Costs Merton, Robert C., and Zvi Bodie. 1992. "On the of Equity."JournalofFinancialEconomics43: 153-193. Management of Financial Guarantees." Financial Foster, Chester, and Robert Van Order. 1985. "FHA Management87-109. Terminations: A Prelude to Rationale Mortgage Merton, Robert C., and Andre Perold. 1993. "Management Pricing." AREUEA Journal 13: 273-291. of Risk Capital in Financial Firms." Harvard Business Government of the Philippines. 1995. "New Policy on School Working Paper 93-038. Cambridge, Mass. Guarantees for Private Infrastructure Projects: A Mody, Ashoka, and Dilip Patro. 1996. "Valuing and Consultative Document." Manila. Accounting for Loan Guarantees." World Bank Research Hsieh, Su-Jane, Andrew H. Chen, and Kenneth R. Ferris. Observer 11: 119-42. 1994. "The Valuation of PBGC Insurance Using an Pennacchi, George. 1987a. "Alternative Systems of Deposit Option Pricing Model." Journal of Financial and Insurance: Pricing and Bank Incentive Issues." Journal Quantitative Analysis 29: 89-99. of Banking and Finance 11: 291-312. J.P. Morgan. 1996. "RiskMetrics: Technical Document." . 1987b. "A Reexamination of the Over- (or Under) New York. Pricing of Deposit Insurance." Journal of Money, Credit Kane, Edward J. 1996. "Difficulties in Making Implicit and Banking 19: 340-60. Government Risk-Bearing Partnerships Explicit." Pennacchi, George, and Christopher Lewis. 1994. "The Value Journal of Risk and Uncertainty 12: 189-99. of Pension Benefit Guaranty Corporation Insurance." Kau, J., Donald Keenan, Walter Muller, and James JournalofMoney, CreditandBanking26: 735-53. Epperson. 1992. "A Generalized Valuation Model for Quercia, Roberto, and Michael Stegman. 1992. Fixed-Rate Residential Mortgages." Journal of Money, "Residential Mortgage Default: A Review of the Credit, and Banking 24: 3. Literature." Journal of Housing Research 3: 341-79. Lewis, Christopher, and Richard Cooperstein. 1993. Sosin, H. 1980. "On the Valuation of Federal Loan "Estimating the Current Exposure of the Pension Guarantees to Corporations." Journal of Finance 35: Benefit Guaranty Corporation to Single-Employer 1209-21. Pension Plan Terminations." In Ray Schmidt, ed., The Wilson, Thomas C. 1997. "Calculating Risk Capital." In Future of Pensions in America. Philadelphia: University Carol Alexander, ed., The Handbook of Risk Management of Pennsylvania Press. and Analysis. New York: John Wiley & Sons. 153 MANAGING GOVERNMENT EXPOSURE TO PRIVATE INFRASTRUCTURE PROIECTS Comments on "The Management of Contingent Liabilities" Clemente del Valle, Director General, Ministry The lack of historical data on the occurrence of of Finance and Public Credit, Colombia events that are being guaranteed against limits the use- fulness of actuarial or econometric methods for mea- Mody and Lewis present a useful approach to the suring risks and expected losses. The Colombian gov- management of contingent liabilities. Their proposed ernment finds it more useful to use a model based on methodology shows how to identify and value contin- contingent claims theory and Monte Carlo simula- gent liabilities and outlines procedures on how to tions. This allows projections to be made based on incorporate their costs to government, with special multiple scenarios with different probabilities in order emphasis on budgetary aspects. This methodology to determine the probability of bad states of the and these procedures need to be formalized and insti- world. The government is trying to make the model tutionalized to ensure their sustainability over time. more user friendly. Better measurement of losses and The,Colombian case provides a good example in sev- the probability of their occurrence will improve the eral areas. structure and coverage of guarantees. The role of a good regulatory framework in mak- Fiscal discipline in the use of guarantees is a top ing guarantees unnecessary is well recognized. Good priority of the Colombian government. However, the project and contract design can also help reach this proposal to provision for guarantees and to establish a goal. If guarantees are necessary, however, it is impor- fund is not always politically or financially feasible. For tant to have a public institution or entity entrusted this reason the government is exploring other comple- with establishing policies on guarantees. In particular, mentary ways to provide discipline. First, a recent law this entity needs to define guidelines on the distribu- establishes limits on the ratio of interest payments to tion of risks by sector between the government and current savings (60 percent) and on the ratio of debt private firms. It also needs to unify criteria across sec- stock to current income (80 percent) for all public tors and across the various levels of government. In entities. This obliges the entities to reflect the impact Colombia a commission comprising the Finance and of guarantees. It can also be used in conjunction with Planning Ministries plays this role. The commission is the proposed guarantee fund, thereby ameliorating its championing a law that requires public institutions, impact on the entity's cash flow position. Second, especially at the municipal level, to formally record where it is not possible to provision for guarantees at important obligations, to value the guarantees, and to the time they are given, future obligations should be reserve against the contingent liability. The law also programmed and budgeted at least one year in envisages the creation of a national fund for this advance. Third, guarantees should be clearly accounted purpose. for. An interesting alternative to the guarantee fund, at 154 THE MANAGEMENT OF CONTINGENT LIABILITIES: A RISK MANAGEMENT FRAMEWORK FOR NATIONAL GOVERNMENTS least from a liquidity point of view, is the standby cred- ances, and contingent state outcomes is sparse. While its of the World Bank, which are being proposed in the Lewis and Mody show in their examples from Tobiagrande-Puerto Salgar toll road. Colombia that it is possible to value real guarantees, it A sovereign asset-liability management system can is not clear exactly how the estimates were arrived at help ensure debt sustainability as part of an economic or how robust they are. This may give a false air of development strategy. Such a system allows integrated specificity to the analysis. While governments need to treatment of the risks associated with both explicit recognize that contingent liabilities are capable of and contingent liabilities within an economic frame- analysis they should also be aware that these analyses work, and can be implemented without moving to the themselves are subject to uncertainties and can require ambitious schemes of Australia and New Zealand, expensive and time-consuming but nevertheless inex- where balance sheets and statements of profit and loss act estimates. are produced. The systems used in Ireland and In outlining guidelines for incorporating contin- Belgium, which provide an institutional framework gent liabilities into the government budgeting process, for the modern and efficient management of risk, may the authors show that cash budgeting leads to signifi- prove more fruitful. The Colombian government, cant distortions in government liabilities and to biases with the help of the World Bank, is developing such a in the types of government support used. The tempta- system. Although the process is slow, it should help tion for governments to provide guarantees without reduce the abuses of the current system and gradually budgeting for their costs is apparent. But Lewis and be adopted at all levels of the public sector. It is cru- Mody go beyond this point to recommend establish- cial to create a culture of risk awareness in govern- ing reserves for unexpected losses in the same fashion ment, in which the potential impact of risk is recog- as a risk-taking corporation. This may not be feasible. nized. Doing so will create incentives to rationalize, Governments may find it impossible to self-insure value, control, and manage risks in an integrated way, against catastrophic losses, or they may find that the which will require a significant investment in human backlash by voters, foreign capital providers, and cred- capital. it rating companies may be unacceptably high. Another strength of the paper is that it suggests ways in which a risk management system can improve David L. Roberts, Duff & Phelps Credit Rating contract design and project management. By under- Company, New York taking careful studies of the risks associated with the toll road project before the bidding, the government As Lewis and Mody note, the risks that many govern- of Colombia reduced risks for both the private and ments assume in order to attract private investment in public sector. Conducting an analysis of risk will infrastructure can by quantified, introduced into the ensure that governments understand the risks they are budgeting process, and reserved against. Where this is taking on. Even implicit government guarantees, done, projects will be pursued only when both social which arise in the case of large banks, large construc- and private ex ante rates of return are positive, and the tion firms, and politically sensitive projects, can also risk of large adverse shocks to the budgets and prove costly, as both the government and the private economies can be minimized. sector may be uncertain of the government's support. Few would dispute that governments need better Lewis and Mody have shown that both implicit ways to account for the contingent liabilities they and explicit liabilities can be appropriately priced, undertake, and few would take issue with the theoreti- budgeted for, and managed. If they could also show cal approach outlined by Lewis and Mody. The prob- that risk management can be done reliably, quickly, lem is how to implement the theory in practice, when and comprehensively, the prospects for improved information on probability distributions, price vari- infrastructure finance would be markedly improved. 155 World Bank Latin American and Caribbean Studies Viewpoints Series Latin America after Mexico: Quickening the Pace by Shahid Javed Burki and Sebastian Edwards Poverty, Inequality and Human Capital Development in Latin America, 1950-2025 by Juan Luis Londofio available in English and Spanish Pobreza, Desigualdady Formacion del Capital Humano en Am/rica Latina, 1950-2050 by Juan Luis Londofno Dismantling the Populist State: the Unfinished Revolution in Latin America and the Caribbean by Shahid Javed Burki and Sebastian Edwards Decentralisation in Latin America: Learning through Experience by George E. Peterson Urban Poverty and Violence in Jamaica by Caroline Moser and Jeremy Holland La pobreza urbanay la violencia en Jamaica by Caroline Moser and Jeremy Holland Prospects and Challenges for the Caribbean by Steven B. Webb Black December Banking Instability, the Mexican Crisis and Its Effects on Argentina by Valeriano Garcia The Long March: A Reform Agenda for Latin America and the Caribbean in the Next Decade by Shahid Javed Burki and Guillermo E. Perry Crime and Violence as Development Issues in Latin America and the Caribbean by Robert L. Ayers Dealing with Public Risk in Private Infrastructure Edited by Timothy Irwin, Michael Klein, Guillermo E. Perry and Mateen Thobani Proceedings Series Currency Boards and External shocks: How Much Pain, How Much Gain Edited by Guillermo E. Perry Annual World Bank Conference on Development in Latin America and the Caribbean: 1995 Edited by Shahid Javed Burki, Sebastian Edwards, and Sri-Ram Aiyer THIE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433 USA Telephone: 202-477-1234 Facsimile: 202-477-6391 Telex: MCI 64145 WORLDBANK MCI 248423 WORLDBANK World Wide Web: http://www.worldbank.orgl E-mail: books@aworldbank.org 9 780821 3403014 30 rnwr8fF4 .1 Y *a~rPWV RRI FP TI-IF Afl.,I F 0%i)II-IF / S9 70-821 340301l-