/(D9(D9 j PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1990 Supplement to THE WORLD BANK ECONOMIC REVIEW and THE WORLD BANK RESEARCH OBSERVER * KEYNOTE ADDRESS A Perspective on Economic Transition in Czechoslovakia and Eastern Europe, Vaclav Klaus * Policies to Move from Stabilization to Growth, Rudiger Dornbusch * Macroeconomic Policy and Growth: Some Lessons of Experience, W. Max Corden * The Environment and Emerging Development Issues, Partha Dasgupta and Karl-Goran Maler * Regional Sustainable Development and Natural Resource Use, Peter Nijkamp, Jeroen C. J. M. van den Bergh, and Frits J. Soeteman * The Soft Underbelly of Development: Demographic Transition in Conditions of Limited Economic Change, John C. Caldwell * Population Growth, Externalities to Childbearing, and Fertility Policy in Developing Countries, Ronald D. Lee and Timothy Miller * Incentives for Small Families: Concepts and Issues, Kenneth M. Chomitz and Nancy Birdsall * Project Appraisal and Planning Twenty Years On, 1. M. D. Little and J. A. Mirrlees • Projects versus Policy Reform, Ravi Kanbur * ROUNDTABLE DISCUSSION Development Strategies: The Roles of the State and the Private Sector, Amartya Sen, Nicholas Stern, and Joseph Stiglitz FILE UOW PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1990 Supplement to THE WORLD BANK ECONOMIC REVIEW and THE WORLD BANK RESEARCH OBSERVER EDITORS Stanley Fischer, Dennis de Tray, and Shekhar Shah ED:ITORIAL CONSULTANT Paul Wolman The World Bank Annual Conference on Development Economics is a forum for discussion and debate of important policy issues facing developing countries. The conferences emphasize the contribution that relevant policy, empirical, and basic economic research can make to understand- ing development processes and to formulating sound development policies. Conference papers are written by researchers outside the Bank and at the Bank. The conference series was started in 1989. The Proceedings of the World Bank Annual Conference on Development Economics is published as a special supplement to the World Bank's professional economics journals, The World Bank Economic Review and The World Bank Research Obseruer. The Editorial Boards of the Review and the Observer do not review the Proceedings, nor are the papers selected from the conference and published here subject to the peer review process through which regular submissions to these journals must pass. Summaries are included of the floor discussions following each paper and following the roundtable discussion; they attempt to convey the sense and substance of what was discussed, interventions by participants from the floor, and responses by panelists. They have not been reviewed by the authors, the discussants, or the participants concerned. The views and interpretations expressed in this Proceedings volume do not necessarily represent the views of the World Bank or its member countries and should not be attributed to the World Bank or its affiliated organizations. All current subscribers to The World Bank Economic Review and The World Bank Research Observer receive this supplement to the journals. Subscription orders for the two journals and for individual copies of the Proceedings supplement can be placed by filling out one of the coupons at the end of this Proceedings volume. ©D 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. All rights reserved. Manufactured in the United States of America. ISSN 1014-7268 ISBN 0-8213-1607-9 Material is this journal is copyrighted. Requests for permission to reproduce articles should be addressed to the Director, Publications Department, World Bank, Washington, DC 20433, U.S.A. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the intended reproduction is for noncommercial purposes, without asking a fee. Permission is not required to make photocopies for classroom or similar use. The World Bank Economic Reuiew is indexed in Current Contents / Social e Behavioral Sci- ences, the Social Sciences Citation Index, the Journal of Economic Literature, and the Public Affairs Information Service. The World Bank Research Observer is indexed in the Journal of Economic Literature, the Standard Periodical Directory, the Public Affairs Information Service, Inc., and, online, by the Economic Literature Index and DIALOG. It is available in microform through University Microfilms, Inc., 300 North Zeeb Road, Ann Arbor, MI 48106, U.S.A. PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1990 Supplement to THE WORLD BANK ECONOMIC REVIEW and THE WORLD BANK RESEARCH OBSERVER Introduction Stanley Fischer, Dennis de Tray, and Shekhar Shah Opening Remarks 11 Barber B. Conable KEYNOTE ADDRESS A Perspective on Economic Transition in 13 Czechoslovakia and Eastern Europe V&clav Klaus THE TRANSITION FROM ADJUSTMENT TO GROWTH Policies to Move from Stabilization to Growth 19 Rudiger Dornbusch Comment, Jacques H. Pegatienan 49 Comment, Marcelo Selowsky 53 Floor Discussion 57 Macroeconomic Policy and Growth: Some Lessons of Experience 59 W. Max Corden Comment, John Williamson 85 Comment, Susan M. Collins 89 Floor Discussion 93 ii Contents SUSTAINABLE DEVELOPMENT AND THE ENVIRONMENT The Environment and Emerging Development Issues 101 Partha Dasgupta and Karl-Goran Maler Comment, Mohan Munasinghe 133 Comment, John Whalley 143 Floor Discussion 147 Regional Sustainable Development and Natural Resource Use 153 Peter Nijkamp, Jeroen C.J. M. van den Bergh, and Frits J. Soeteman Comment,KiritS. Parikh 189 Comment, William B. Magrath 195 Floor Discussion 201 POPULATION CHANGE AND ECONOMIC DEVELOPMENT The Soft Underbelly of Development: Demographic Transition 207 in Conditions of Limited Economic Change John C. Caldwell Comment, Ron Lesthaeghe 255 Comment, Susan H. Cochrane 261 Floor Discussion 271 Population Growth, Externalities to Childbearing, 275 and Fertility Policy in Developing Countries Ronald D. Lee and Timothy Miller Comment, Martha Ainsworth 305 Incentives for Small Families: Concepts and Issues 309 Kenneth M. Chomitz and Nancy Birdsall Comment, Paulina Makinwa-Adebusoye 341 Floor Discussion 347 Contents iii PUBLIC PROJECT APPRAISAL Project Appraisal and Planning Twenty Years On 351 I. M. D. Little andJ. A. Mirrlees Comment, Lyn Squire 383 Comment, Ernesto R. Fontaine 387 Floor Discussion 393 Projects versus Policy Reform 397 Ravi Kanbur Comment, Enzo R. Grilli 415 Floor Discussion 419 ROUNDTABLE DISCUSSION Development Strategies: 421 The Roles of the State and the Private Sector Amartya Sen 421 Nicholas Stern 425 Joseph Stiglitz 430 Floor Discussion 433 PROCE E D ING S O F THE W O R L D BANK ANN U AL CONFE RE NCE ON D E V E L O PM E NT E C ON O M ICS 1 9 9 0 Introduction Stanley Fischer, Dennis de Tray, and Shekhar Shah The World Bank's series of annual conferences on development economics was begun in 1989 to bring Bank staff and researchers together with academics and policymakers from around the world to discuss key issues of development. Its aim is to provide a forum for an exchange of views and ideas between those who do research on development issues and those who draw on that research to design and implement development policies and projects. The Bank's comparative advantage in conducting development research arises in part from the fact that research and operational activities take place under the same roof, often by staff who have performed both functions at some point in their careers. The vast range of countries, sectors, and policies that the Bank deals with in the normal course of its operational and research work is another source of its comparative advantage. At the same time, like almost all large organizations, the Bank must guard against a tendency to be inward looking in its analysis and prescription. These annual conferences are one of the principal vehicles through which Bank researchers and operational staff maintain contacts with the external research and policy communities. An early decision in the design of these conferences was to emphasize both the deepening and broadening of development knowledge. To achieve these dual objectives it was decided that each conference would cover several topics rather than focus on a single theme. Themes are selected in consultation with the Bank's research community to be topical and to cover a broad spectrum of interests. The second conference, held April 26-27, 1990, in Washington, D.C., focused on four pressing policy and research issues: long-term growth, project evaluation, the environment, and population growth. The wide scope of issues and ideas covered by these topics and the diversity of participants provided a major challenge: to serve participants with broad interests in development as well as experts in each chosen area. To achieve this goal we divided each topic into two sessions, a morning plenary session in which an overview of the area and key issues in the development context were presented, and a longer after- noon session in which more technical and specialized discussion could be pur- sued. In both morning and afternoon sessions, the authors presented brief sum- maries of their formal papers, which were followed by invited commentaries from other experts and then by open floor discussion. This proceedings volume, a special supplement to the Bank's research jour- nals, the World Bank Economic Review and the World Bank Research 1 2 Introduction Observer, begins in the next section with the remarks of the president of the World Bank, Barber B. Conable, who inaugurated the conference and intro- duced the keynote address by Vaclav Klaus. Dr. Klaus is Czechoslovakia's min- ister of finance and one of the architects of the historic political-economic trans- formation now under way in that country. The formal papers follow, accompanied by the text of the discussants' comments and summaries of the lively floor debates. The volume ends, as did the conference, with a roundtable discussion featuring Amartya Sen, Nicholas Stern, and Joseph Stiglitz on the roles of the state and the private sector in development. The roundtable discus- sion was designed to address some of the issues central to World Development Report 1991. Before presenting the full proceedings, we provide below brief summaries of the conference's main presentations. ECONOMIC TRANSITION IN CZECHOSLOVAKIA AND EASTERN EUROPE The conference began with Klaus's spellbinding first-hand view of the transi- tion from a command economy to a market-driven economy. In his reflections, Klaus juxtaposed what he termed as the "knowns" in the area of economic adjustment and reform with the "unknowns." He argued that as a product of their historical experience both with the mechanisms of the state-dominated economy and with previous attempts at reform, Czechoslovak reformers have concluded that partial reform in a distorted economy is worse than no reform, that transformation out of economic inefficiency must be swift and pervasive both in its fundamental recasting of property rights, prices, and incentives as well as in its pursuit of sensible policies in the fiscal and monetary realms. Klaus cautioned, however, that headlong "decentralization" without reform of the underlying structure of the market and property rights is an invitation to social chaos. Klaus emphasized that it is crucial for a generation of Eastern European leaders-who are accustomed to detailed plans and timetables-to recognize that comprehensive reform does not mean waiting for the perfect reform plan, preconceived in all its steps. Rather, he likened the reform process to a chess game in which the opening moves can be made even without looking at the board-but in which no one can know the situation after the fifteenth or twenty- fifth move. Klaus concluded his talk by reviewing the "unknowns"-questions to which he and other East European reformers most needed answers from the interna- tional development community, and in particular from this conference and its participants: In what sequence should they restructure institutions and liberalize prices? In what sequence should they liberalize foreign trade and the exchange rate? What will be the supply response to the reform measures being introduced? How rapid will it be? And finally, how do they minimize the rents that may be reaped, especially by the nomenklatura, when controls are lifted in a distorted system? Klaus's cogent presentation was received with extraordinary enthusiasm by the conference participants. Introduction 3 ADJUSTMENT AND GROWTH The 1980s were years of adjustment-for many developing countries the "lost decade." The papers presented in the sessions on adjustment and growth ask how we can make the 1990s a decade of growth. In the morning plenary session, "From Stabilization to Growth," Rudiger Dornbusch examined the key issue for many developing countries that have succeeded-more or less-in stabilizing their economies in the 1980s, but still see few signs of growth. In the afternoon session, Max Corden drew lessons on macroeconomic stabilization and growth from a major, seventeen-country, World Bank-financed study undertaken by a team led by Richard Cooper, Ian Little, Sarath Rajapatirana, and himself. Dornbusch presents his views on stabilization-which is the first, necessary step in structural adjustment-as well as on the problems of the transition from stabilization to growth. His conclusions and recommendations are an interesting mix of orthodoxy and heterodoxy. On the orthodox side, he emphasizes the key role of fiscal balance in the stabilization process, recommending also the use of incomes policy and a temporarily fixed exchange rate when the inflation rate has to be reduced rapidly. More unconventionally, however, Dornbusch urges that the exchange rate be moved quite quickly to a crawl, to prevent the overvalua- tion that has developed in programs in which the exchange rate was used as the nominal anchor. Put differently, Dornbusch sees little point for policymakers to try single-mindedly to reach zero inflation, and he recommends that policy- makers favor faster growth, even if it means inflation rates of up to 20 to 30 percent a year. Dornbusch points out that even with the appropriate macroeconomic and structural policies, growth frequently does not appear. Partly this is because of lack of demand, caused by restrictive fiscal policies and lowered real wages, and the slowness of supply to switch to export markets. In this setting, something is needed to make up the demand and to provide entrepreneurs with the confi- dence to invest. Dornbusch supports public works expenditures, which he believes should be externally financed. But he also puts considerable emphasis on the potential role of flight capital in growth, arguing that there may be dual equilibria-a low-growth equilibrium in which flight capital stays abroad, and a high-growth equilibrium in which growth and flight capital return together. How is the high-growth equilibrium to be attained? Dornbusch points to the role of an external show of support, partly through debt relief, and also through an inflow of stabilization funds. Dornbusch's interesting paper is about the high-inflation Latin American countries and Turkey, and only about them. It does not address the problems of the African countries that also face the problem of restoring growth after attempting to stabilize; nor does it take into account some of the Asian coun- tries, including Indonesia and Thailand, which succeeded both in stabilizing and in restoring growth during the 1980s. In contrast, Corden's paper draws lessons from the adjustment experiences of no fewer than seventeen countries, including some from Africa and Asia. Although Corden is characteristically cautious and 4 Introduction measured, emphasizing the diversity of experiences in the seventeen countries, he does find some broadly applicable lessons. Corden starts by pointing out that all but one of the countries (India) in the study sample experienced a public-sector-led spending boom in the period 1975-80, induced either by the availability of foreign loans or by improvements in the terms of trade. His first lesson derives from this experience: in good times, avoid euphoria and submit projects to proper cost-benefit analysis. Corden illustrates the great variety of responses to the crises that different countries faced in the early part of the 1980s with case studies of countries that on the whole adjusted well during that period: Colombia, Indonesia, Thailand, and Turkey. Colombia was slow to react to an adverse change in external conditions, but when it did react, it took drastic action in 1984. Similarly, Indonesia reacted sharply to adverse shocks, avoiding prolonged periods of overvaluation or mounting debt. But the case of Thailand shows that it is not necessary to adjust rapidly, provided a rapid decision is made to adjust and the government has the power to behave consistently to bring about that adjust- ment. The contrast between the return of growth, especially in the Indonesian and Thai cases, and its failure to return in Latin America, is worth noting. It may be partly accounted for by the much longer period of good economic management in the Asian countries before the 1980s crisis. On the relationship between inflation and growth, Corden believes that the evidence supports the general principle that high-inflation countries have had lower growth. Of course, exceptions do exist (Brazil in the 1960s). He concludes plausibly that because both inflation and growth are the outcomes of economic policy, it is likely that governments that manage policy well produce both low inflation and higher growth, and that high inflation is a sign of a government that has lost the capacity for good economic management. Corden ends by discussing the relationship between the exchange rate regime and inflation. His conclusions are again justifiably eclectic: that to assure low inflation a country has to commit itself to noninflationary monetary (and one would add, fiscal) policy, but that beyond that there is no clear evidence that a fixed exchange rate is necessary for low inflation- with the Asian countries that have flexible exchange rates and low inflation as the leading evidence. What should one conclude from Corden's evidence of the variety of experi- ences? There are few simple rules that admit of no exceptions, beyond the need to follow cautious macroeconomic policies. SUSTAINABLE DEVELOPMENT AND THE ENVIRONMENT Can development occur in ways friendly to the environment? Are some devel- opment approaches more or less harmful to the environment? What is "sustain- able development" and how can it be achieved? These questions and a host of others related to environment and development are emerging as the issues of the 1990s for developing countries and development economists. Partha Dasgupta Introduction 5 and Karl-G6ran Maler introduced this topic to conference participants in their paper, "The Environment and Emerging Development Issues," and Peter Nijkamp, Jeroen C. J. M. van den Bergh, and Frits J. Soeteman expand on the theme in "Regional Sustainable Development and Natural Resource Use." Both papers argue that the prevailing exclusion of environmental considerations in economic modeling and planning is bad for developing nations and for global welfare both now and in the future. The papers approach environmental issues from very different angles. Dasgupta and Maler argue for the inclusion of environmental factors within a more or less conventional economic framework, treating them as economic goods with positive accounting prices in national income calculations. Nijkamp and colleagues establish geographic areas, rather than nations, as the unit of analysis in their program of "regional sustainable development" and take a systems approach in which economics and the environment are viewed as two discretely functioning-albeit closely related-systems, whose competing demands need to be balanced. Natural resources, their use and preservation, are at the heart of many envi- ronmental debates. Dasgupta and Maler open their discussion by exploring the special character of natural resources. They contend that depletion and pollu- tion of natural resources, whether for direct consumption, production, or both, diminish a society's well-being even when the resources are regenerative in ways that for the most part have been ignored by the development literature and by policy and investment decisionmaking. A key point of their argument is that although societies' well-being does not necessarily hinge on preserving current stocks (or quality levels), it does demand that measures of growth and invest- ment decisions account for the sometimes hidden costs of changes in the physical environment. One cannot understand the environmental problems the world faces or design solutions for them without understanding the institutional aspects of resource use. Dasgupta and Maler emphasize that markets for environmental goods mal- function or are nonexistent in many developing country economies for a variety of reasons: because property rights are not specified, because the actors are not in contact with each other, or because one interested party enjoys a considerable and consistent advantage in deciding the course of action. These factors often combine to obscure the real prices of environmental goods and not infrequently lead to a situation in which the poor, who are often heavily dependent on natural resources, bear more than their share of the cost of distorted policies and prices. Although environmental issues in general have received much press as of late, concerns over the care of the "global commons" have been especially evident. The Dasgupta and Maler solution to global commons problems (greenhouse gases, for example) is a system of transferable, national environmental use/ abuse permits. However, as the authors themselves admit, reaching agreement on the original allocation of permits may be an insurmountable problem. 6 Introduction Nijkamp and colleagues argue that theoretical developments over the past two decades have solidified the notion of sustainable development and suggest a wide range of approaches. The Nijkamp framework starts from a definition of sustainable development as "a balanced and adaptive process of change . . . characterized by a dynamic Pareto-optimal trajectory in which progress in one system-that is, either the economic or the ecological-would not be to the detriment of the other system." It stresses that in the short run, environmental and economic goals, both on a regional and global scale, are often mutually conflicting. However, in the long run, consistent with the concept of sus- tainability, a situation of muitual complementarity, or "coevolution," can emerge. The Nijkamp paper uses a holistic, planning-oriented framework for achiev- ing sustainable development called "regional sustainable development (RSD)." In contrast to the general movement away from regional and national planning, the RSD framework would require a regional planning establishment that would take an integrated economic and environmental approach and have authority over a wide range of institutions and actors. POPULATION AND ECONOMIC DEVELOPMENT Scholars and philosophers have debated and analyzed the interrelationships between population growth and economic development for centuries. Progress has been made, but for many important issues the debate often seems hardly closer to resolution now than it was when Malthus first gave them intellectual legitimacy more than two hundred years ago. Do rapid increases in a country's population slow its economic development, or are people just as much an engine of growth as capital? However this debate is resolved, the world at large has long ago taken as fact that slowing population growth is at least a necessary condition for economic and social development. How, then, to achieve this goal? What are the roles of family characteristics and family planning policies in reducing fertility? How do health and population policies interact? Most important, can we "jump start" demographic transition-the move through declining mortality rates to declining fertility rates to declining population growth rates-or must we wait for history to run its course? These are important issues for the developing world in general, but they are most pressing in Sub-Saharan Africa. John Caldwell, a noted Australian economic demographer, introduced these subjects to the conference in his paper, "The Soft Underbelly of Development: Demographic Transition in Conditions of Limited Economic Change." Caldwell's paper addresses the fundamental questions of how and why demo- graphic changes occur. He seeks to understand why most Sub-Saharan African countries have not experienced the usual declining birthrates that accompany development. He also examines the other side of this coin, how demographic changes affect the pace and form of development. Throughout the paper Cald- Introduction 7 well argues forcefully that current demographic-economic theory cannot explain the national and regional experiences seen in Africa. Caldwell comes out in favor of historical analogy as a more rational guide for action. Caldwell addresses a number of other questions critical to welfare, economic development, and population policy in Africa-for example, what forces drive mortality decline, and what lessons can Africa draw from the Asian experience with family planning programs and policies? Most important from a policy perspective, he concludes that even under adverse circumstances, family plan- ning programs can speed the onset of fertility decline, but probably only if they are part of a comprehensive health program, one providing family-level counsel- ing and ready access to fertility control measures. What are the prospects for the Sub-Saharan region in the near future? Cald- well rejects recent World Bank and United Nations predictions that the region is likely to witness a widespread reduction in fertility in the 1990s. Those projec- tions, Caldwell claims, predict fertility declines perhaps twice as steep as what will occur. Furthermore, he notes, "the huge demographic, and indeed develop- mental, unknown in Africa is the impact of AIDS." The AIDS epidemic will mean that especially in Africa the past is a poor indicator of the future. Caldwell makes clear throughout his paper that current demographic-economic theory fails to portray adequately either the effects of population on economic growth or the effects of economic growth on population. He believes, however, that we can conclude that fertility decline is compatible with, and probably required for, the transition to a modern economy. Intensive, democratically applied family planning can contribute to that decline. Many countries have tried to provide incentives or coerce families into having fewer children than they want. In "Incentives for Small Families: Concepts and Issues," Kenneth Chomitz and Nancy Birdsall present an argument for public policy interventions in fertility decisions that recognizes the sensitive and value- laden nature of the issue. They argue that the incentive programs they propose do not have the drawbacks commonly associated with incentive-based policies designed to influence fertility. Incentives designed to influence fertility decisions are morally unacceptable, according to Chomitz and Birdsall, only if the popula- tion should be entitled to their benefits unconditionally or if they induce behav- ior that violates important cultural norms. They also argue that successful incen- tive programs must build on and reflect parents' underlying motives for having and rearing children. The justification for incentive schemes is that, left to their own devices, cou- ples individually have more or fewer children than society as a whole would like them to have. Chomitz and Birdsall identify two types of market failures that they believe affect fertility decisionmaking and therefore justify public policy intervention: imperfect access to contraceptive information and services; and a lack of private market mechanisms that promote contraception techniques that require only information (for example, rhythm and withdrawal). They also argue that incentive schemes, subsidies, or both could also be used to overcome 8 Introduction couples' hesitation to adopt contraceptive measures in the absence of adequate credit or insurance markets that would offset the risk of possible health compli- cations and ensuing unemployment. Chomitz and Birdsall point out that there are other types of externalities that may drive a wedge between couples' fertility decisions and the desires of the society in which they live-for example, the amount of public goods or common property, the magnitude of public transfer payments, or disparities of wealth that may affect the incremental child's productivity as an adult. An attempt to derive empirical measures of some of these factors is presented in Ronald D. Lee and Timothy Miller's paper, "Population Growth, Externalities to Childbear- ing, and Fertility Policy in Developing Countries." Lee and Miller define the externalities to childbearing as the gain or loss in utility or welfare that couples would experience if fertility decisions were decided collectively rather than independently. The authors offer calculations based on data from a number of developing countries and that of the United States to show how the externalities they identify can be quantified. Their bottom line for many countries-when externalities are summed-suggests that childbearing externalities are not terribly important for most of the cases stud- ied. Only countries with highly valued natural resources per head, such as Brazil and Saudi Arabia, showed strongly negative childbearing externalities. In con- trast, the United States has high positive externalities because it has a large defense budget, a mushrooming national debt, and a strong pension program. In what is bound to be a controversial finding, Lee and Miller conclude that those externalities associated with childbearing that they are able to analyze and quantify do not, for most countries, provide a case for public policy intervention beyond "neutral" family planning services (the provision of information on new birth control technologies, for example). The authors are careful, however, to emphasize that their analysis is much too tentative at this time to support any policy advice, either for or against, on the issue of public sector intervention in the sphere of fertility decisions. PROJECT APPRAISAL World Bank economists have frequently observed that project appraisal methods within the Bank are not applied in practice in a manner consistent with classic treatments of the subject. This concern and the central role that good project design and selection play in fostering economic development has led the Bank to take a fresh look at the principles of project appraisal and their applica- tion in the Bank. Ian Little and James Mirrlees, authors of the classic treatment of the subject, begin their paper, "Project Appraisal and Planning Twenty Years On," with a summary of the main elements of their methodology. They then go on to explore reasons for the apparent decline in the Bank's use of cost-benefit analysis. The decline in the popularity and use of the Little and Mirrlees methodology Introduction 9 is open to several interpretations. It could be that either the growing volume of lending or a perceived lowering of quality standards has led those involved in the design and implementation of projects to feel less compelled to rigor in their assessments of potential projects. Or the Little and Mirrlees methodology may itself be too cumbersome, too demanding of difficult-to-generate data, to be practical. Little and Mirrlees partially confront these alternative hypotheses in their discussion of the difficulties that have arisen in applying their recommenda- tions. They recognize that it may be difficult to obtain shadow prices, a key building block of their approach, but they argue that the attempt should none- theless be made; they discuss the relative valuations of private versus public costs and benefits of projects; and they point to the difficulty of taking account of flexibility in the design of projects. In examining the argument that uncertainty about the outcome of projects reduces the value of project appraisal, Little and Mirrlees develop a simple model that suggests that project appraisal remains worthwhile in the presence of typical amounts of uncertainty about the out- come. Little and Mirrlees, in extending their scrutiny of appraisal to practices in the World Bank, criticize some aspects of the recommended Squire-van der Tak principles, and even more the reported lack of shadow price adjustments beyond a general adjustment for shadow exchange rates. Of some comfort for those of us in the Bank, they do not find the practices in other multilateral agencies and in most governments to be any better, except perhaps in the U.K. Overseas Development Administration. In seeking to account for the decline of project appraisal in the World Bank, Little and Mirrlees point to the pressure that arose from expanded lending targets in the McNamara era in the Bank. The authors argue that organizational changes within the Bank may have compounded these problems, but admit, as well, the possibility that the recommended methods may be too complex to be consistently applied in a large bureaucracy. Little and Mirrlees conclude that "social cost-benefit analysis is not as widely, as well, or as effectively practiced as its expected net value might lead one to hope and expect." Their paper certainly should lead to further reappraisal of why that is so, and of what can or should be done to widen the method's application. Although project lending remains the central business of the Bank, in recent years a new type of lending has gained prominence: policy-based adjustment lending. About 25 percent of World Batik lending now takes the form of adjust- ment loans made not to build a road or a power generation plant but to help countries make needed policy changes. In "Projects versus Policy Reform," Ravi Kanbur examines the interrelations between project and policy loans. Funda- mentally, his paper is a plea for the use of project appraisal methods in consider- ing policy loans. As he points out, the documentation for adjustment loans frequently includes alternative projections of gross national product that could serve as the basis for the typical present value or rate of return calculation undertaken in project appraisal. Of course, as Kanbur himself would probably 10 Introduction agree, one would first have to determine whether the models that underlie these projections are sufficiently accurate to provide answers that can be used with confidence. Kanbur also discusses the impact of policy reform on the returns to projects, suggesting reasonably that policy reform that reduces economic distortions would typically increase the returns to projects (indeed, this is one of the under- lying rationales for policy-based lending). He also considers the other side of this coin, suggesting reasons why successful projects might increase the returns to policy reform by increasing supply response. Ac KNOWLEDGMENTS Several Bank staff in the Research Administrator's office and elsewhere con- tributed to the success of this conference and to the preparation of this proceed- ings volume. Paul Wolman assisted in the editing of this volume. Manny Jandu provided able and all-round support for the conference. We would particularly like to place on record our considerable debt to Celina Bermudez, who as the conference coordinator very competently organized and managed the conference logistics. It is with great sadness that we record here her untimely death in September 1990. PR OC E ED ING S OF THE WO R LD BANK ANNU AL CON FE RE N CE ON DE V E LOP MEN T EC ON OM I CS 1 9 9 0 Opening Remarks Barber B. Conable Good morning, friends. It is indeed gratifying to find what in nuclear physics you would call a critical mass of development economists here today. I think it is marvelous that we have attracted so many distinguished participants this year, and it speaks well of the reaction to last year's conference. Development is our business at the World Bank. Every year it seems develop- ment gets more complicated and more exciting. For instance, no one who attended this conference last year could have anticipated the extraordinary transformation which has since swept the world. That transformation by defini- tion goes to the heart of development research and practice. The events of the past year have posed acutely the fundamental question, how can people's lives be improved in sustainable, equitable, and manageable ways? This question is difficult to answer because of rapid change in some places and very slow change in others. Some problems appear familiar, such as the poverty, food shortages, poor health, and inadequate education that are the standard subjects of development theory and practice. Others have assumed a fresh urgency, such as revitalizing inefficient industries, making institutions and gov- ernments more responsive to public need, and introducing isolated economies to a wider range of trading partners. Comfortably familiar or newly urgent, these problems demand the illumina- tion of research. Development cannot succeed in practice if it fails in theory, and still less can it succeed in theory if it fails in practice. Operational misjudgments can affect the lives of thousands, even millions, of people. World Bank staff and I are keenly aware of the importance of the knowledge base which has to underpin our work here. I firmly believe that the understanding of development derived from Bank research and experience is at the heart of our mission, and ultimately we should be judged on the soundness of this understanding. We must be continuously tested about the assumptions underlying our approach to development and to our project and policy advice. I see Bank research as venture capital. It should be invested in ways that may enlighten our operations. Knowledge and progress march hand in hand. The purpose of this conference is to cement that relationship. In reaching out to a broad audience, we aim to stimulate debate, which will be embodied operationally both inside and outside the Bank. Barber B. Conable is president of the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 11 12 Opening Remarks Despite the pace and degree of change, some familiar issues run through events like unbroken threads. Among those with which we and others have grappled over the years are several of this conference's themes: proceeding from stabilization to growth; evaluating projects so that lenders and borrowers get the best value for their money; population growth; and the environment and sus- tainable use of natural resources. These same themes have reappeared in the dramatic changes occurring in Eastern Europe. It is fitting as well as an honor to introduce Dr. VAclav Klaus, Czechoslovakia's minister of finance. I know Dr. Klaus will feel at home in this gathering. He has a doctorate in economics from Prague University, briefly studied in Italy, and has studied in the United States, I am happy to say, at my own alma mater, Cornell. And in his own country he enjoyed a distinguished career at the Institute of Economics of the Czechoslovak Academy of Sciences, the State Bank, and the Institute of Forecasting of the Academy of Sciences, where he headed the Department of Macroeconomic Policy before emerging as a member of the Civic Forum leadership in 1989. I welcome him to this distinguished gathering. I welcome all of you to this second World Bank Annual Conference on Development Economics. PRO C E ED ING S OF THE W OR L D BANK ANN U AL CON FE RE NC E ON DE V EL O PM ENT E CON OM I CS 1 9 9 0 KEYNOTE ADDRESS A Perspective on Economic Transition in Czechoslovakia and Eastern Europe Va'clav Klaus Mr. President, ladies and gentlemen, I am-and indirectly my country is- extremely honored by being invited to be here and to be asked to deliver the keynote address to this important conference in this important institution in this important moment of history. As you know, Czechoslovakia, one of the found- ers of the Bretton Woods institutions, has reapplied for membership in the International Monetary Fund and World Bank. We have already initiated very close contacts, and we really want to be good and reliable partners, both of the institutions as well as of the individual member countries. We don't want to receive only; we want to give something, if possible, even if at the very beginning what we can offer may not be that significant compared with what we can get. As I understand it, this conference will focus on development problems, on different kinds of transitions-transition from one economic system to another, transition from adjustment to growth, transition from temporary acceleration of growth to sustainable development. My own country is in a dramatic and rapid process of transition as well. Therefore, our experience may be of some interest to you. We are trying, as I am sure you know, to realize the transition from a centrally planned economy toward a market economy. In this respect, our intentions are quite clear. We don't want to repeat our mistakes of the 1960s to introduce a hybrid between central planning and a market economy. Rather, we want to achieve the transition from a state-dominated economy toward an economy based on the private sector, private initiative, and private entrepreneurship. We don't intend to orchestrate the economy from above. We don't want to start another vicious circle of pseudo-rationalistic social engineering, based on what I might call the ambitions of irresponsible intellectuals and technocrats. We do want to achieve the transition from a non-efficient, wasteful, environ- mentally damaging economy to an economic system based on scarcity prices, on sound incentives, on transparent general rules. We want to discontinue our passive, overcautious, and defensive muddling Vaclav Klaus is the minister of finance of the Czech and Slovak Federal Republic. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 13 14 Keynote Address through based on partial amendments of the existing system. We really want a change. I even don't like to use the very-often-used term reform. We want to achieve a rapid transition from a painful restructuring period with a zero or maybe negative rate of growth toward a positive rate of economic growth very soon. We have many ambitious plans, and we have been asking ourselves all along whether it is even possible to dismantle a centrally planned economy. In addi- tion, we are not sure whether our current economic situation is in reality an asset or a liability, whether it facilitates the reform or blocks it. It is very often claimed that the Czechoslovak economy has the best starting position (as regards foreign indebtedness as well as disequilibrium in basic domestic markets) for the eco- nomic restructuring. We are afraid that this helps to create dreams that it is possible to go on, to muddle through, to improve partially the existing system. We are facing many other problems, and we are well aware of them. I will name some of them before starting to discuss what we hope to know (to some extent) and what we still don't know sufficiently. We ask ourselves how to unfold the whole process of economic transforma- tion, how to sequence it. That is what we consider the most crucial problem. Then, when the transformation process has already started, as it has in Czechoslovakia, we ask ourselves how not to lose the momentum of the reform; how to build and maintain the necessary political and social consensus; how to maintain credibility of the reform policy; how not to cross the tolerance limit of the population; how to break down the old, unproductive, collectivistic social contract-how to transform it, how to rewrite it; and how to minimize the costs of restructuring in terms of growth, employment, inflation, and so on. In raising all these questions I would like to stress that for the answers I look to existing mainstream economics, to economic studies prepared by research teams in international institutions, including-or, should I say, starting with- the World Bank. We cannot wait for the creation of the specific theory of a transition from a centrally planned economy, as it is sometimes interpreted. All the answers are available; we have to be creative enough to use them in specific cases we have to solve. We are more and more convinced that our country, or any other, is less unique than is often claimed. In some specifics there are some differences, but I am quite sure that there are more similarities than differences. The basic eco- nomic laws are valid across continents, economic systems, ideological beliefs. Maybe it's not necessary to stress this, but I remember the debates several years ago about "development economics" or the "economics of development." In any case, it's necessary to stress the point in our part of the world, where the knowledge of standard mainstream economics is very small, almost negligible. This is one of the important fields where the World Bank could help us-to increase the knowledge of mainstream economics in our part of the world would be a perfect development project. It would be very helpful. Now, I would like to stress a few points that I call the "knowns," things we Klaus 15 understand about the reform strategy, and then make some comments on the "unknowns," things we have to learn here or elsewhere. THE "KNOWNS" IN REFORM STRATEGY The following five points are what I call the "knowns," ideas that we have brought to or developed in the reform process. 1. We know very well that a partial reform is much worse than a nonreform. This is a message we learned from the partial reform in the 1960s in Czechoslovakia. This is a message we have got from very careful study of the reforms in Central and Eastern Europe in the last two decades. The partial reform in a distorted economy is a tremendous mistake. 2. When we stress a comprehensive reform, it doesn't mean that we must wait for an all-embracing reform blueprint. In my opinion, waiting for an ambitious, intellectually perfect, all-details-elaborated reform project is a suggestion to start the reform in the year 2057. It means postponing the reform process to eternity; there will never be a reform. What is even worse, when the reform process has already started, is to wait for the blueprint; it leads very quickly to a chaotic disintegration of the economy, as we see nowadays in the Soviet Union. Waiting is one way of falling into what we call in Czechoslovakia the "reform" trap. 3. A reform program means a plan for several crucial steps in a proper sequence. A reform program doesn't mean that we know all the details, that we can announce in advance all the reform steps, that we have all the data for all possible scenarios. I like to compare the transformation process with chess playing. When we want to play chess, we must know how to play. We must know how to move various pieces on the chessboard. We must know the basic opening strategies. But it's not possible to know the situation on the chessboard after the fifteenth or twenty-fifth move. I stress this because it is very difficult to explain to the politicians in Eastern Europe who are still used to detailed plans, full of data and time schedules. The politicians (and the public) must believe that we know how to play chess, that we know various opening moves and strategies even without looking at the chessboard, that we are able to respond quickly and effectively to unexpected situations. Our problem is that our politicians don't believe that we are that good at playing chess. Thus, every day I still must stress the difference between a detailed, step-by-step reform plan and a clear-cut reform strategy. 4. Traditional economic reform in Eastern Europe, we have known for decades now, is a reform trap. It is a trap for at least two reasons, one connected with the microeconomic aspect of the problem, the other with the macro- economic aspect. At the microeconomic level, the traditional reform paradigm of "decentralization" is a mistake, a trap. Decentralization essentially means shifting the locus of decisionmaking without making parallel changes in the basic characteristics of the system, starting with property rights on the one hand, and with prices and incentives on the other. Even if at first sight the idea of 16 Keynote Address decentralization in an overcentralized economy sounds rational, it is a misunder- standing. It is a trap because, all other things being equal, shifting the locus of decisionmaking to microeconomic agents without changing the basic charac- teristics of the system brings more problems than solutions. But even if the major challenges for the transformation process are micro- economic in nature, the reform process cannot succeed without sound macro- economic policy. This is something the Bretton Woods institutions stress all the time. We in Czechoslovakia know that restrictive monetary and fiscal policies are the necessary preconditions for any successful economic reform. Without them, we are in the reform trap again. We stress the macroeconomic aspects of the whole reform process very often. We made a very dramatic change, as some of you may know, in the state budget for 1990 in Czechoslovakia. The former government had proposed a state bud- get with a deficit. On Monday, December 11, 1989, the day after my appoint- ment as minister of finance, I carne in to my new office to find that on Wednes- day, December 13, I was supposed to defend the state budget in Parliament. That was a good excuse for me not to even try to read the old budget. Instead, our new government suggested a provisional budget for the first three months of 1990 and promised to prepare a new budget by March 1990. By cutting sub- sidies and by making other savings on the expenditure side of the new budget, we succeeded in reversing the budget from a deficit to a surplus. A budget surplus seems to us a very important precondition for the success of the whole reform process. At the beginning we succeeded in cutting expenditures and subsidies. This was mainly against the producers, as we put it, and until now not against the consumers, because the budget was based on cutting sub- sidies to the production sector only. This will be changed very soon. At the same time, we introduced a very restrictive monetary policy with the target zero rate of growth of money supply for the year 1990. We want to create a favorable macroeconomic environment for the whole reform process; we con- sider this crucial. 5. We have understood that another major obstacle to a successful economic transformation is the lack of transparency of basic economic relations in a post- centrally planned economy. All economic agents at the micro level as well as the government and the architects of the reform are in reality blind because of the lack of transparency in two respects-in the field of property rights and in the field of prices. We have understood that the early, rapid transformation of property rights is absolutely crucial for the reform to succeed. Every day I come to understand more and more how important it is to start this transformation right away. This is so because the old firms don't react-don't respond-to new incentives, signals, and changes in the environment. Yet with the reform already started and rhetoric at a very high pitch, there is a chaotic, extremely inefficient, and extremely unjust privatization going on, regardless of the intentions of the architects of the reform. Thus, there is no time to wait to change the structure of property rights. Klaus 17 Under the Ministry of Finance we established a special Board for the Tempo- rary Administration of State Property and Its Privatization. The board has pre- pared the basic concept of the privatization scheme and will organize the whole privatization process. Our project has two standard stages. In the first stage is the commercialization of the existing state firms, their transformation into a "privatizable form," which means that state joint stock companies with a given number of shares will be created in the crucial part of the Czechoslovak economy. In the second stage, the shares of these companies will be sold to the public by means of auctions. Those are standard procedures, but we have one specific addition. Because of the lack of domestic capital, it will be necessary to augment the wealth of the population by distributing free a part of state property in the form of vouchers to the population at large. After that, it will be possible to start the exchange of vouchers for the shares of the state joint stock companies. I would like to stress that a very early transformation of property rights, to prevent chaotic privatization, which is going on, is absolutely necessary. There is no time to privatize 5 percent of the state property, 10 percent, or 20 percent in two, ten, or fifteen years, as in other developed and developing countries. We have to start with the bulk of enterprises in a few months' time. There is no other possibility. We also feel more and more clearly that prices must be changed at the early stage of the reform process. Only a very small part of the price restructuring can be done in the form of a centrally orchestrated price correction; most of it must be done by the invisible hand of the market after price liberalization. At the beginning we did assume that an important part of the price correction would be done by an administratively orchestrated action. We feel now that this inter- mediate step will be much smaller than the final price liberalization. Those are the "knowns," as we have derived them from our studies of the past two decades. We had a paradoxical advantage here because we were not allowed to reform in the last twenty years. It was a sort of privilege; we could concentrate on studying the reforms in other countries. We devoted our time to studying mainstream economics. Hopefully, it was a good long-run investment in human capital. PRINCIPAL "UNKNOWNS" IN REFORM STRATEGY Now I would like to talk about what we consider the five main "unknowns." These are more questions than answers. 1. We are not sure about the sequencing of the improvements of the quality of the market (that is, of the market structure) on the one hand, and the price liberalization on the other. (The sequencing between price liberalization and restrictive macro policy is clear.) In the past we stressed the role of the nurturing of the market structure as a precondition for the price liberalization. I am less and less sure about that. 18 Keynote Address 2. We are not sure about the sequencing of the institutional restructuring at the micro level and the price liberalization. It is quite clear that it would be counterproductive to adjust and liberalize prices before economic agents have the incentives and sufficient freedom to respond. We know, however, that mar- kets are unlikely to function effectively without an appropriate degree of price flexibility. We know that both must be done immediately, but unfortunately the timing cannot in reality coincide. 3. Another problem is sequencing as regards domestic institutional and price measures on the one hand, and liberalization of foreign trade and rate of exchange on the other. The flexibility of exchange rate movements and convert- ibility must be established at a relatively early stage of the reform process, and we hope to achieve it by the end of 1990, because the current rate makes the removal of import controls impossible. The exchange rate will have to depreci- ate toward a level that can be sustained. But, again, the sequencing between the foreign trade and exchange rate liberalization and domestic liberalization is another unknown variable. 4. We don't know whether we can expect a rapid (or how rapid) and positive (or how positive) supply response to the set of drastic reform measures we are introducing. (There are the first signals of output losses, of slight decline in industrial output, but we have to wait for better data because the official statisti- cal data do not measure sufficiently the growth of the private business that is under way.) 5. The final problem is more or less a political and social one, but I consider it an enormous problem regardless of all my liberal rhetoric-that is, how to minimize the ability of some individuals to reap the enormous rents that become available in the existing distorted system when the central controls are lifted. It is a dramatic problem we face every day, and I am a little uneasy about it. We could spend a much longer time discussing the problems of transition. I hope that with our closer contacts with the World Bank we will learn a lot, especially on our side, about them, and that our Czechoslovak experience will help some future reforming countries as well. Thank you very much for your attention. PROCE E DING S OF THE WORLD BANK ANNUAL CON FE RENCE ON DE VE LO PM ENT ECONOM I CS 1 990 Policies to Move from Stabilization to Growth Rudiger Dornbusch Although some discussions of stabilization see growth as more or less an assured product of appropriate stabilization policies, this paper argues that there is no guaran- tee that stabilization will lead to growth; it may result in stagnation. The paper dis- cusses the essentials of stabilization, including inflation targets, fiscal policy, monetary policy, exchange rates, and incomes policy, and suggests that two areas of structural reform-deregulation, including trade reform, and reform of the financial sector-can play a central role in the long-run success of a stabilization effort. The paper concludes that countries that have experienced protracted high inflation, financial instability, and payments crises probably will have a difficult and protracted transition to growth, and that external resources will be necessary as a continuing feature of the transition. Sustained external support in the form of long-term loans, heavily conditioned on the tangibility and credibility of domestic progress in adjustment, will help provide a bridge by which flight capital may return and foreign direct investment may be encouraged to take advantage offresh opportunities. Discussions of economic stabilization traditionally have assumed that fiscal aus- terity, competitive real exchange rates, sound financial markets, and deregula- tion provide the conditions for a resumption of growth. One must distinguish the necessary from the sufficient conditions, however. Adjustment is a neces- sary, but not necessarily a sufficient, condition for a resumption of growth, because asset holders may postpone repatriating flight capital, and investors may delay initiating projects. These factors raise an important problem of coor- dination that classical economics does not recognize. This paper starts with a statement of the problem, reviews essentials of stabilization, and then turns to the question of structural adjustment and the return to growth. I. THE PROBLEM Figures 1 and 2 show two very different cases of economic performance. In Chile (figure 1), after serious domestic and external disturbances, a long effort at Rudiger Dornbusch is professor of economics at the Massachusetts Institute of Technology and a research associate at the National Bureau of Economic Research. He is grateful to Vittorio Corbo and John Williamson for helpful comments. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 19 20 Policies to Movefrom Stabilization to Growth Figure 1. Actual and Potential Output, Chile, 1970-89 Millions of escudos 470 450 - 430- 1 00 Potential 410 - 390 - 370- 350- o 290 ~~~~~~~~~~~Actual 310 290 270- 250 1970 1972 1974 1976 1L978 1980 1982 1984 1986 1988 1990 Note: 1977 prices. Sources: Intemational Monetary Fund (IMF) data, and Marfan and Ardagoitia (1989). Figure 2. Index of Per Capita Output, Argentina, 1963-89 (1980 = 100) 102 98 _ 94- 90 86- 82 78 74 , , I , , 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 Source: IMF data. Dornbusch 21 restructuring has been paying off. In 1989, Chilean output reached the level of potential, and the scope for growth in the years ahead is substantial.1 Argentina (figure 2) represents the opposite experience. Output per capita has been declining for almost a decade and is now at the level of the early 1960s. Macroeconomic instability, particularly hyperinflation, stands in the way of normality. But even if stabilization occurs, is there any assurance that growth will resume promptly? The examples of Bolivia and Mexico suggest that it may not. Thus, even if stabilization occurs in Argentina, Brazil, or Peru (and ultimately it must), there is little assurance that the onset of austerity will not translate into protracted stagnation. The policy issue, then, centers on the following questions: * What are the essential steps that assure stabilization? * What are the key policy measures in the restoration of growth? * What is the contribution of the external environment? Specifically, what role can debt relief and stabilization loans play in supporting a program? The literature on these questions falls into two broad categories. The first is the optimistic approach, which, for example, represents the official view of the International Monetary Fund (IMF). The optimistic approach asserts that with the appropriate policies in place, stabilization will pay off rapidly in terms of growth. In support, the optimists can cite a few cases to suggest that vigorous reconstruction can give way to a period of strong growth-Brazil in 1964-67, Chile after 1983, and the Republic of Korea and Turkey at the beginning of the 1980s. The other trend in the literature, the skeptical approach, cites the experi- ence of Mexico, Bolivia, or even Chile in arguing that there is no quick step from stabilization to growth and that the transition remains difficult to understand and even more difficult to accomplish. I discuss these two lines of argument after a review of prescriptions for stabilization. II. STABILIZATION The design of a stabilization effort comprises five elements: * The post-stabilization inflation target * The extent and manner of fiscal stabilization * The appropriate monetary policy * The appropriate level of the exchange rate * The use of an incomes policy. On each of these issues there is now a significant body of evidence, and therefore it is appropriate to draw some lessons. 1. The data for potential output come from Marfan and Artiagoitia (1989), as updated by the author. 22 Policies to Movefrom Stabilization to Growth Inflation Targets Two schools dominate thought on inflation targets. The intransigent school asserts that nothing short of zero inflation is a viable policy target. The more lenient school accepts, if necessary, moderate inflation of 20 or 30 percent. The intransigents argue that only zero inflation is a stable target and that any concession on this leads to a cumulative departure. Fellner (1976) forcefully argued this view, drawing attention to the need for an explicit price target path because of the problem of "self-justifying lack of credibility" in the absence of such a commitment. This is a strong argument, and it should be the last word in an economy starting with zero inflation. But what of an economy that has fought its way down to 20 percent inflation, say, and whose policymakers now contemplate further disinflation? The policymakers face a cost-benefit issue if credibility in and of itself does not produce this further disinflation. If attaining further reduction in inflation takes protracted slack in the economy, then going all the way to zero inflation by spending an extra year or two with slack can be very costly. But it is equally clear that being too anxious to turn the corner by declaring victory over inflation too early keeps the inflationary virus fully alive and leaves the economy vulnerable to a resumption of high inflation. On the issue of inflation targets, pragmatism must prevail. Central bankers should talk about zero inflation, but they also should compromise with reality. At the margin there are tradeoffs, and pursuing zero inflation at any cost is not only socially irresponsible but also bad economics. Fiscal Policy Adjustment of the budget is the indispensable feature of a stabilization pro- gram. Protracted fiscal deficits that cannot be financed in the domestic capital market or abroad lead to high inflation and in time to megainflation or even hyperinflation. The evidence from Latin America and now from Eastern Europe in this regard is quite unambiguous. It is one thing to know what kind of deficit a stable country can run without getting into trouble; it is quite another to set the allowable deficit for a country that wants to restore stability. Hysteresis effects in this context are more than a fad; they are a live issue because the preceding period of financial instability will have semipermanently deteriorated the scope for noninflationary deficit finance. Specifically, the demonetization that is always the consequence of high inflation-whether it be by dollarization, capital flight, or flight into fully liquid interest-bearing assets- reduces for a long time the scope for noninflationary deficit finance. The size of the budget deficit that can be financed will depend on how far the financial instability has gone. If there has been hyperinflation, a budget surplus is required. If inflation reached only 50 percent, there is room for moderate deficits financed by money creation and debt finance. The size of the deficit also will depend on the inflation target. There is room for a moderate remonetization of the economy, but the scope is drastically limited. Beyond that, planned seign- Dornbusch 23 iorage revenues must be consistent with the inflation target. There is a close link between revenues and inflation, given by (1) X (ag - y) (1 - g) where ir is the rate of inflation, g is the budget deficit financed by money creation, y is the trend growth of output, and ct and j3 are parameters of the velocity equation.2 The higher the noninflationary level of velocity and the higher the response of velocity to inflation, the more inflationary is deficit finance. Therefore, it is appropriate to look at inflation in two ways: one is how to reduce inflation from high levels by restrictive aggregate demand policies and by an incomes policy; the other is what fiscal policies to put in place to finance the budget consistent with the inflation target. Fiscal adjustment should take place on several fronts. The first is the introduc- tion of a productive tax structure. A productive tax structure involves four elements: * A broad tax base, without exemptions and only a few taxes * A firm attitude toward tax compliance * Moderate, preferably uniform rates of taxation * Absence of significant subsidies of any form and establishment of efficient public utility rates. Not included here is a tax amnesty, which is often favored as part of fiscal reconstruction. Uchitelle (1989) shows the very limited success of such a measure. In Latin America, to take a specific region, tax systems are defective in every one of these dimensions. Large parts of Latin American economies-for exam- ple, agriculture in Mexico or Brazil-were exempted from taxation until recently. Tax evasion is pervasive, especially among the privileged. In Argen- tina, for example, compliance is a joke, and government after government con- dones one of the worst compliance records in the world. Argentina has just now approved a law that penalizes tax evasion, but the government is far from starting its implementation. There is much room for change. Still, improvements in fiscal administration in countries traditionally plagued by poor tax compliance-Italy, Mexico, and Spain, for example-offer grounds for hope. During periods of financial instability, public sector pricing becomes a macro- economic issue. When inflation is too high, public sector price increases are slowed down to reduce inflation. The resulting deficit creates financial problems that then are solved by emergency increases in public sector rates. This yo-yoing is extremely inefficient. Public sector prices should be set on the basis of micro- economic efficiency considerations. Any income distribution consequences 2. The equation assumes a steady state where inflation is equal to money growth less real growth and a velocity equation that is linear in inflation: V = aL + 3ir (see Dornbusch 1989). 24 Policies to Move from Stabilization to Growth should be resolved through the general tax structure. Public utility rates should be indexed on a regular basis even if that means there is more indexation and hence more vulnerability to inflation in the economy. Inflation must be stopped by a permanent balance in the budget, not by a temporary slowdown of public sector price increases. The tax rate structure in Latin America remains highly distorted. It is charac- terized by a proliferation of taxes and punitive rates for the sectors least able to evade taxation and by an excessive emphasis on regressive selected sales and trade taxes rather than by comprehensive expenditure or income taxes. Sub- sidies remain pervasive in the prices of public enterprises, in the credit market, for particular regions, and in particular sectors. The combination of punitive taxation of sectors, regressive taxes, and widespread subsidies produces a totally unproductive tax structure, a high marginal cost of revenue, and hence an almost inevitable bias toward inflationary finance in response to shocks. Reform of the tax system is essential both for economic and social reasons. Financial stability cannot come about without far larger revenues at a much lower margi- nal cost. Fiscal reform has to do with establishing a tax and expenditure struc- ture and a tax base such that the marginal cost of extra revenue is lowered. That in turn implies that extra taxes, not money creation, can become a plausible response to adverse fiscal shocks. Along with the efficiency of the tax system goes the issue of emergency taxa- tion. Take again the case of Argentina, where crises are solved by imposing export taxes and raising public utility prices. Subsequently, as inflation picks up and competitiveness deteriorates, the export tax comes off, and the utility rates are allowed to fall behind inflation. Soon the next fiscal crisis occurs, and everything starts all over again. The instability of Argentina's tax pressure is apparent in figure 3. This process destabilizes public finance, capital markets, and economic efficiency. Only a fiscal reform that provides revenue to finance the government on a steady basis can help overcome these problems. On the expenditure side, a number of reforms typically are necessary: * Efficient administration of public utility rates * Cuts in public sector employment * Privatization and closing of public sector firms * Restoration, maintenance, and investment expenditures on social and eco- nomic infrastructure. Public sectors, like attics, need occasional cleaning out. Employment in the public sector gets bloated because of patronage and poor accountability. Pro- ductivity can be raised sharply by reviewing labor requirements. The immediate savings-although not always as large as they would be in Brazil, for example, where public firms pay wages far above the industry averages-generally are worthwhile in most developing economies. There is also ample room for priva- tization. Like fashions, ideological fads change, and one can benefit from the Dornbusch 25 Figure 3. National Taxes as a Percentage of Gross Domestic Product, Argentina, 1970-87 Percent 20 19 18 17 16 15 14 13 12 11 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 Source. World Bank data. mood of the day to sell off steel mills, airlines, or telephone companies-none of which today are considered to occupy the commanding heights of capitalism. Privatization is useful for three reasons. First, the public sector does not have the managerial capacity to administer a major share of gross national product (GNP) in a cost-effective fashion. Second, the public sector does not have the investment resources required to provide all public services well. Third, reve- nues are required to avoid deficit finance. Low prices received in the privatiza- tion are a serious problem, but these low prices reflect the precariousness of the economy, and they may well become even lower if failure to privatize and thus obtain fiscal resources causes financial stability to deteriorate even further. The available resources therefore should be allocated to sectors in which private initiative is less willing or able to function. Telephone companies need not be in the public sector, but rural schools should be. Privatizing appropriately ensures that public resources are freed for important jobs and that private invest- ment is focused on upgrading infrastructure and supply of services. Privatization thus should not be an ideological issue. It helps not only to reduce the budget deficit but is essential also to finance investment. We noted above that the inefficiency on the tax side implies a very large marginal cost of resources and hence leads to shocks that translate readily into inflation. On the spending side, poor resource management cuts into both physical and social capital and reduces prospective growth and political stability. 26 Policies to Movefrom Stabilization to Growth Fiscal adjustment, both on the spending and tax sides, must achieve a better mix of equity and efficiency. The mix has deteriorated dramatically in many countries in Latin America in the past decade. Real payments to pensioners in Argentina, for example, have declined to only half those of wage earners. At the same time, too many people are on pensions because the system permits retire- ment at an absurdly low age. As a result, younger workers qualify for pensions and then work in the underground economy while older pensioners are put into distress. A more sensible systern is to raise the age of eligibility for retirement and to use the savings to support more viable payments to older pensioners. The same type of problem emerges in the public sector-too many people, paid too poorly and too randomly. The result of this misadjustment is poor performance. The state's poor perfor- mance as a supplier of services in turn "legitimizes" tax evasion as a form of revolt against the state. This vicious circle must be broken. The solution is not to abolish the state but rather to have it function better-more efficiently and more equitably. Tight Money or Tight Budgets? The typical stabilization program for an economy emerging from high infla- tion involves freezing wages and prices and making little fiscal adjustment. For the first month or two this program is successful, partly because expectations of a freeze will have led to prior price hikes. Soon the freeze wears off, however, and the deferred rise in public sector prices, the lifting of export taxes, and real appreciation combine to erode the budget position. Then, in phase two, policy- makers implement tight money. This gives an unsustainable program another few months of life, but of course it also increases public indebtedness sharply. Next, in phase three, the problems with the program become widely perceived, and debtors plead their distress due to high real interest rates. When tight money goes, the house of cards collapses; the exchange rate collapses, inflation surges, and real interest rates turn very negative. Soon, another stabilization is under construction, ready for the spring, tottering in the summer, and blown away by fall. The important lesson to draw is this: tight money is not a substitute for a balanced budget. Real interest rates ultimately should be low (New York plus 3 percent), and the only way such a situation is sustainable is by basically sound fiscal and real exchange rate policies. We return to the financial market issue below. Here we simply note that tight monetary policy-realized real interest rates of 30 or 40 percent-is a signal of serious misalignment in the budget, the real exchange rate, or both. Realized real interest rates at such levels, in the presence of domestic debt, soon give rise to fiscal problems. Exchange Rates At the outset of a stabilization program, quick success in disinflation is critical in gathering the political capital for further progress on more basic adjustments. Dornbusch 27 Fixing the exchange rate can help achieve this objective, more so the higher the initial inflation, and hence the more the level of the dollar serves as an indicator for economywide pricing. In fact, a fixed rate is far more preferable than a floating rate. Under a floating rate the inevitably tight money position will drive up interest rates sharply and thus attract capital flows that will lead to sharp real appreciation. Real appreciation is undesirable because, ultimately, it has to be undone. This is not a negligible issue, because if disinflation succeeds, it will be very inconvenient to have a large devaluation just as the program gathers momentum. Political resistance to the devaluation in turn forces policymakers into a high interest rate policy to defend a basically overvalued exchange rate. There should be a premium on removing pending problems, not on creating new ones. The key issue then is to select the initial level of the exchange rate so that even with moderate inflation for a few months, the real exchange rate is not over- valued from the start. Moreover, very soon, exchange rate policy should shift from a fixed rate to a crawling peg to offset inflation differentials and maintain competitiveness. Once again, politically, it is extremely inconvenient to shift to a crawling peg (often seen as the source of inflation) just as inflation comes down. But if the decision is postponed too long, the real exchange rate becomes starkly overvalued, and the program ultimately fails. Chile's experience illustrates these problems. Figure 4 shows Chile's real exchange rate. Following the coup of 1973 and a period of fiscal stabilization, Figure 4. Index of the Real Exchange Rate, Chile, 197690 (1980-82 = 100) 120 110 100 90 80 70 60- 50 l l 1976 1978 1980 1982 1984 1986 1988 1990 Source: Morgan Guaranty Trust Company, unpublished data. 28 Policies to Movefrom Stabilization to Growth the Chilean currency was placed on a tablita (schedule), and in 1978, when inflation was still above 20 percent, the currency was fixed to the dollar. As a result, the real exchange rate appreciated steadily and vastly. Inflation did come down over the next two years, but not fast enough to avoid a dramatic over- valuation. In 1981, net exports reached a deficit of 8.2 percent of gross domestic product (GDP) against an average of a surplus of 0.8 percent in the 1970s. For a while, the deficits were financed in the world capital market, but by 1982 confidence and credit had withered, and the policy collapsed. Chile's exchange rate policy in the post-1983 stabilization was far more appropriate. Inflation targeting became more pragmatic, and the real exchange rate was pushed steadily into more competitive ranges. As a result, a steadily strengthening traded goods sector could support a sustained growth in the econ- omy. The strength of the traded goods sector in turn translated into moderate real interest rates, thus facilitating management of the external debt and of the domestic budget. Figure 5 and table 1 show the case of Turkey. The initial exchange rate policy, following the 1980-81 problems, supported the strong restructuring and recov- ery of the economy shown in table 1 (see Celasun and Rodrik 1989; Sareacoglu 1987; and Dervis and Petri 1987 on the Turkish stabilization). In 1989-90 this sound exchange rate policy has given way to a dramatic real appreciation. This may well become a Chile-style problem, particularly because Turkey does not have preferential European Community (EC) access and Figure 5. Index of the Real Exchange Rate, Turkev, 19 76-90 (1980-82 = 100) 150 140 - 130 - 120- 110 100 90 80 1976 1978 1980 1982 1984 1986 1988 1990 Source: Morgan Guaranty Trust Company, unpublished data. Dornbusch 29 Table 1. Restructuring Success, Turkey (average annual growth and shares in percent) Share of GDP Measure ofsuccess 1973-79 1981-87 1980 1987 GDP 5.1 5.6 100.0 100.0 Exports -1.1 24.6 7.3 21.3 Imports 1.8 12.0 15.4 22.9 Manufacturing value added 4.9 8.4 22.4 26.0 Source: Organisation for Economic Co-operation and Development data. because serious competition from Eastern Europe in the European market is a certainty. In fact, the slowdown in growth and the widening current account imbalances already indicate major problems. When fiscal austerity reduces demand, full employment growth requires an offsetting mechanism for crowding-in. A competitive real exchange rate does provide such a mechanism. This may not be the case in the short run, as discussed below, but in the medium term it does work. Incomes Policy The discussion of exchange rates has already introduced the topic of incomes policy. Here, we go a step further to raise two questions. First, is incomes policy an important ingredient in stabilization? Second, how should it evolve in the course of stabilization? Without fiscal austerity stabilization cannot start. Without incomes policy it is unlikely to succeed. Incomes policy is necessary as a coordinating device when wage and price setting is not fully centralized. Because of built-in inflation expectations in contracts, adjustments have to be made. It is also important to intervene to stagger wage and price setting over time so that different contracts are spread across various points along the adjustment cycle. In principle, all this could be accomplished by enough austerity and tightness of aggregate demand. But if substantial inertia prevails, via implicit or explicit indexation, incomes policy can help reduce the unemployment cost of indexa- tion. To take the extreme example of an economy in which all wages and prices are both fully flexible and entirely forward looking-and thus capable of falling into line on the mere announcement of a credible program-is not realistic. Thus, incomes policy comes to play its role by shifting all wages and prices to a new regime. But although temporary incomes policies, including wage-price controls, are useful, their perpetuation is certain to create deep problems. Ample examples exist. Policymakers thus should move quickly to a system of indexation of public sector prices, the exchange rate, and wages. The temptation to postpone the shift to a crawling peg exchange rate is often responsible for an ultimate over- valuation of the exchange rate. Wage indexation on a semiannual or annual basis will create a new inertia around a low inflation rate. Far from being a 30 Policies to Movefrom Stabilization to Growth source of inflation, wage indexation protects the economy against rapid infla- tionary escalation provided that monetary and fiscal policies are sound. (It is understood that real exchange rate changes and changes in real public sector prices must be purged from the indexation formula.) If monetary and fiscal policies are not sound, nothing can protect against inflation. Indexation has gotten a bad reputation in Latin America because it has been blamed for the instability of inflation. There is no merit to that argument, because it implicitly assumes that in the absence of indexation real wages would have adapted more easily to the shocks of the 1970s and 1980s. An argument to the contrary is that real wage resistance would have translated into faster and politically more troublesome wage adjustments in response to shocks. Next we turn to a discussion of the supply side, which serves as a background for the review of structural adjustment. III. THE SUPPLY SIDE The starting point for discussion is an aggregate production function. The determinants of output are the available labor force, N, the capital stock, K, and the state of knowledge and institutions captured by the parameter A. (2) Y = AF(K, N) The basic approach to growtlh relies on a production function in the tradition of Solow-Dennison growth accounting (the distinction between GNP and GDP is omitted here): (3) y = a + (1-ae)k + ain where lowercase letters a, k, and n represent growth rates, and a is the share of labor in income. Estimates of the sources of growth have been collected by Chenery, Robinson, and Syrquin ( 1986) and are shown in table 2. The data reflect the significant role of total factor productivity, the catchall for the poorly understood mechanics of economic growth. Table 2. The Sources of Growth in Developing Countries (average growth rates in percent) Chenery sample Korea, Rep. Source of growth of 20 countriesa 1963-73 1973-86 Value added 6.3 9.5 7.8 Totalfactor input 4.3 5.4 4.1 Capital 2.5 3.2 2.2 Labor 1.8 4.1 3.8 Totalfactor productivity 2.0 4.0 2.4 a. Sample of twenty developing countries in various time periods. Source: Chenery, Robinson, and Syrquin (1986, table 2-2); and Song (1990, table 5-5). Dornbusch 31 The growth accounting approach can be expanded in a direction that high- lights three aspects of factor inputs: the available supply in the economy, the efficiency with which a given supply is allocated, and the level of utilization of the given supply. For simplicity, let X refer to an index on the interval 0-1 for the degree of utilization. And let E be an index that measures the extent to which distortions in the allocation of resources impair the efficiency of factor utiliza- tion and hence their productivity, again on a range 0-1, with unity representing the undistorted economy. Moreover, let these efficiency and utilization indexes be common to both capital and labor.3 Then the growth equation becomes (4) y=0+ an+(1 -a)k; 0=a+x+e In this form we can separate out five sources of growth in income. In addition to technical progress and increasing capital intensity, we now identify as sepa- rate contributing factors both the efficiency of resource allocation and the level of utilization. Cyclical recovery, for example, would yield transitory extra growth over and above what factor accumulation gives, as would an improve- ment in the allocation of resources. The central point of this decomposition is to highlight that capital formation is only one avenue to growth. In view of the scarcity of saving available for capital formation, increased attention must focus on improving productivity. Moving a step further, we note that capital formation relies on domestic saving or a noninterest current account deficit. Rewriting the growth equation we have (5) y = a + can + r(s +) where s is the national saving rate, X is the noninterest current account deficit expressed as a fraction of GDP, and r is the marginal return on capital formation.4 Equation 5 highlights the role of domestic saving, s. Higher saving rates finance capital accumulation and growth. But the equation makes the important point that the immediate impact of saving on growth is minor. Assume that the return to capital is 10 percent. Raising the saving rate by 5 percentage points of GDP will then raise the growth rate of output by only 0.5 percentage points. Of course, the compound growth effects of an extra 0.5 percent growth are consid- erable, but only in the long run. Recent literature on growth economics has struggled with the fact that empiri- 3. Specifically, we now have Y = AF(EXK, EXN), which, with linear homogeneity, becomes Y = AEXF(K, N). 4. The distinction between GNP and GDP arises from net foreign assets. Capital formation, AK = S + NICA, has as a counterpart national saving and noninterest current account deficits (NICA). The growth equation for GNP (Z) then can be written as z = (1 - K)(a + x + e + czv) + rs + K(r - r*)X, where K is the ratio of net foreign liabilities to GNP, X is the noninterest current account deficit, and s the national saving rate. The rate of interest on net foreign liabilities is r*, and the marginal return on home capital formation is r. 32 Policies to Movefrom Stabilization to Growth cally total factor productivity growth accounts for so much of growth and is so poorly explained (Romer 1989a; Helpman 1988). One important direction for further studies of sources of growth is in the scale of the market and related externalities (see Romer 1989b; Murphy, Shleifer, and Vishny 1989a, 1989b). The growth accounting framework leads to a number of policy-oriented questions: * Is there a link between econornic policies and total factor productivity growth? * Is there a link between policies and the national saving rate? * What kind of policies will assure the full utilization of resources? * What kind of policies assure that national saving is invested at home rather than abroad and that foreign saving will become available? These questions are naturally familiar from the discussion of structural adjust- ment and stabilization. They center on the issue that a country must use the limited availability of resources most effectively; sound regulatory and trade policies are at issue here. They also deal with the need to mobilize effectively domestic saving and to create an environment in which it will be invested at home. That has to do with a stable, productive financial framework for eco- nomic development. IV. STRUCTURAL ADJUSTMENT Two areas of structural reform are singled out here for special attention: deregulation, including trade reform, and reform of the financial sector. Both areas are focal points of adjustment efforts, and structural adjustment in both fields can play a central role in tlhe long-run success of a stabilization effort. Deregulation and Trade Reform Growth accounting consistently shows that most of growth in per capita income is not explained by capital accumulation but by growth in total factor productivity. It is appropriate therefore to ask whether a country can identify policies that would lead directly to a more efficient use of resources. Deregula- tion and trade reform can play that role. The effect of an improved resource allocation, by trade liberalization or by deregulation, can be represented as a gain in productivity (see, for example, Easterly 1989). Suppose the production function for output is linearly homoge- neous in capital, labor, and intermediate inputs, H: (6) Q = F(K, N, H) The value-added function, Y, can then be written as (7) Y = 0(p) G(K, N) where p measures the real price of intermediate goods. A decline in the real price of intermediate goods because of competition or reduced costs of transborder Dornbusch 33 shipments therefore operates in the way of technical progress by shifting out the aggregate production function. Another way in which a more open competitive market or improved trade opportunities translate into productivity gains can be represented in a model that places importance on the variety of intermediate products available to firms. In the formulation of Romer (1989a), emphasis is placed on the size of the market in sustaining the profitable production of specialized intermediate goods. Because of the presence of fixed costs, the larger the market, the larger the range of specialization that can take place. Let the production function for final goods be (8) Y = Nl-Y Exa where x denotes the quantity of each intermediate good.5 Let there be M inter- mediates, and assume that it takes one unit of labor to produce a unit of the intermediate. The labor requirement for intermediates, N,, then is N1 = Mx, and that leaves NF = N - N1 of labor for final goods production. Therefore, we can rewrite the aggregate production function for final goods as (9) Y= (N- NI)1- N,Ma The point of the Romer formulation is to highlight that, in addition to labor, input variety (proxied by M, the number of different inputs) is a determinant of the level of output. A larger and more open market increases the aggregate output not because of scale economies to labor but because it allows the produc- tion of a larger variety of specialized inputs. But gains also result from the more traditional economies of scale that stem from declining average variable cost attributable to wider markets. Raising the scale of operation of individual firms is in this case the source of gain in produc- tivity. Worldwide operation for firms with scale economies raises their produc- tivity and frees resources as firms merge into more efficient units. De Melo and Robinson (1990) emphasize the correlation between growth rates and the growth of total factor productivity and interpret one of the channels as export- led growth, which provides the resource base for imports of capital goods. Pecuniary externalities become available in export-led development that acceler- ates growth over what the classical growth model allows. Opening of markets that are closed by licenses or by government monopolies or restrictions thus provides an important source of productivity growth. In fact, aggressive deregulation may well be one way to achieve a Schum- peterian change (Schumpeter 1934, pp. 64-66): "Development in our sense is a distinct phenomenon.. . . it is spontaneous and discontinuous change in the channels of the flow, disturbance of equilibrium, which forever alters and dis- 5. For simplicity we assume that the quantity of each intermediate good used in the final good is the same so that xi = x. This symmetry result would emerge if the production of each intermediate good had the same constant unit labor cost. 34 Policies to Movefrom Stabilization to Growth places the equilibrium state previously existing. . . . Development in our sense then is defined by the carrying out of new combinations." In Schumpeter's analysis, development originates in the following: * The introduction of a new good * The introduction of a new method of production * The opening of a new market * The conquest of a new source of supply of raw materials or intermediate goods * The carrying out of the new organization of any industry. Deregulation and trade reform may be effectively the instruments that take an economy out of the trap of slow growth toward an acceleration of growth that then develops its own dynamics and financing. Even though the search for productivity growth is essential and obvious, caution is required when trade reform is at stake. The elimination of obstacles to trade-the movement away from a system of quotas and licenses that effectively closes the economy, as in Chile or Mexico-invariably spills over into a large increase in imports. The beneficial effects on exports are much slower to appear, because although inputs become more readily available and technology improves, exports do not rise immediately even if a real depreciation is under- taken. Without real depreciation, exports will scarcely help pay for the higher imports. If real depreciation is not possible, then liberalization should occur in two rounds. First, the country should move from quotas and licenses to a uniform, high tariff of, say, SO percent. Later, when the economy booms, and the external balance can support liberalization without the risk of an exchange crisis, tariffs can be taken down to 10 percent. Such a policy does change radically the openness of the economy, because tariffs allow competition at the margin, whereas quotas and licenses prevent such competition. But at the same time, a two-round liberalization policy avoids the grave risk of an exchange crisis. When Chile liberalized imports almost fully in the late 1970s (and overvalued its currency), import levels exploded and could not be financed. The exchange rate collapsed, and another stabilization had to be undertaken. Similarly, in Mexico, when the country moved from a closed economy almost immediately to a tariff of only 10-15 percent, import levels increased very sharply, the trade surplus disappeared (see figure 6), and the exchange rate thus became overvalued. Incomes policy packages and a concern for inflation now make it impossible to devalue. As a result, very high real interest rates are being used to defend the premature liberalization. The policy is clearly unsustainable unless capital inflows, fostered by modernization and the free-trade agreement with the United States, provide the financing. Reform of the Financial Sector Fiscal mismanagement and the resulting financing of deficits by persistently large negative real returns on assets ultimately cannot fail to divert savings abroad and reduce investment. Once again, Argentina is an example. Figure 7 Dornbusch 35 Figure 6. Trade Balance, Mexico, 1985-89 Bilions of U.S. dollars 2.6 2.4 2.2 2 1.8 1.6- 1.4- 1.2- 1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 1985 1986 1987 1988 1989 1990 Note: Billion = 1,000 million. Source: Banco de Mexico data. Figure 7. Index of the Real Value of an Investment, Argentina, 1983-89 (1983= 100) 170 160 150 140 130 Atv 120 110 100 90 70 _ 60 50 . Passive 40 30 - .- 20 * 10 1983 1984 1985 1986 1987 1988 1989 1990 Note: Active rate is the lending rate; passive rate is the deposit rate. Source: Data from Fundaci6n de Investigaci6nes Econ6micas Latinoamericas (Argentina). 36 Policies to Move from Stabilization to Growth shows the cumulative real value of an investment in Argentina's financial market at the active and passive (lending and deposit) rates. Starting from 1983, the real value of an investment would have declined to only 5 percent by 1989. A country that runs a financial system with dramatic negative rates of return, on average, cannot expect to retain saving or investment. The variability of real rates adds to the loss because it forces everybody to become a speculator in a negative-sum game. The policy package of international institutions rightly emphasizes the inevi- table need for budget balancing and for competitive real exchange rates. But it also gives strong emphasis to the need for positive real interest rates and, more generally, to the need to abolish financial repression (see Polak 1989; World Bank 1989b; Gelb 1989; and Molho 1986). The evidence in support of this policy recommendation on positive real interest rates is less decisive than that in support of competitiveness and a balanced budget. Two arguments ordinarily are presented for the positive real interest rate recommendation: * Positive deposit rates mobilize saving. Specifically, with positive rates there are higher saving rates, and saving will be efficiently channeled by financial inter- mediaries rather than going into goods or dollars. • Positive real active rates assure a higher quality of investment and therefore higher growth rates of output. The World Bank has expounded the view that positive real interest rates and financial liberalization can help promote growth. World Development Report 1989 (World Bank 1989b) reports evidence of a positive relation between real growth and real interest rates. A study by Polak (1989) similarly concludes that real interest rates have a positive effect on growth. Table 3 presents their evidence. In both cases averages of growth rates for a sample of thirty-three developing countries were used in a cross-section regression. The results for World Devel- opment Report 1989 further allow for a shift dummy variable to separate the 1965-73 period from the 1974-85 period. Both studies support the view that a 5 percentage point increase in the real interest rate raises the real growth rate by an entire percentage point. If these results are at all representative, they of course Table 3. Effects of Real Interest Rate on Growth of Real Gross Domestic Product Source Constant r dum R2 World Bank (1989b) -0.12 0.2 -0.02 0.45 (-2.5) (5.2) (-3.4) Polak (1989) 5.21 0.21 0.32 (1.5.3) (4.5) Note: r = real interest rate on deposits; dum = dummy variable for 1974-85. The t-statistics are reported in parentheses. Dornbusch 37 Figure 8. Real Deposit Rates and Growth, Fourteen Asian Economies, 1970-79, 1980-86 Real deposit rate (average percent for period) 6 5 _ 4 0 3 00 2 0 I 0 0 0 0 0 0 0 o -1Z -2 0 -3 0 0 -40 -5 -6 -7 -8 -9 -10 -11 0 -12 __ _ __ _ __ -2 0 2 4 6 8 Real per capita GDP growth rate (average percent for period) Note: Data are one observation in each period for each of the fourteen Asian economies listed in footnote 6, less one observation not available for China in the 1970-79 period. Source: Olita and Faber (1989, tables All-2 and All-7). have extraordinary implications for growth policy. The evidence in support of a linkage between real interest rate and growth is less strong, however, than the World Bank or Polak would lead us to believe. Gelb (1989), on whose research World Development Report 1989 is based, is in fact far more circumspect than the report itself. Persistently large negative real deposit rates misdirect saving. Similarly, ran- dom and priceless allocation of investment has negative consequences for the productivity of resources. Most of the evidence about the harmful consequences of misdirected capital market policy come from the outliers-countries that have vastly negative asset returns. Once these outliers are isolated, the evidence no longer supports the claim that positive real interest rates help growth. Figure 8 and table 4 support this view. The data shown here are the averages (1970-79 and 1980-86) of per capita growth rates of real income and real deposit rates for fourteen Asian economies. Regression analysis using twenty- seven data observations (two subperiods, fourteen economies, less one observa- tion not available for China in the 1970-79 period) yield no significant evidence of an effect of real interest rates.6 6. The regressions use all observations for the periods 1970-79 and 1980-86 reported in Okita and Faber (1989). The economies included in the sample are Bangladesh, China, Hong Kong, India, Indo- nesia, Korea, Malaysia, Nepal, Pakistan, Philippines, Singapore, Sri Lanka, Taiwan, and Thailand. Excluded because of lack of observations for real deposit rates were Burma, Fiji, and Papua New Guinea. 38 Policies to Movefrom Stabilization to Growth Table 4. Real Interest Rates, Saving, Investment, and Growth Rate Constant r K2 S/Y 22.7 0.23 -0.033 (12.4) (-0.41) I/Y 25.4 0.21 -0.031 (17.2) (-0.47) Ay- n 3.9 0.05 -0.03 (7.7) (0.34) Note: S/Y is the saving rate, II Y the investment rate, and Ay - n the growth rate of per capita income. The variable r denotes the real deposit rate. The t-statistics are reported in parentheses. Source: Okita and Faber (1989). In the sample of Asian countries there is no correlation between saving rates and real interest rates, between investment rates and real interest rates, or between per capita growth rates and real interest rates. With so striking an absence of any real interest rate effects in this particular sample, we return to the World Bank data. Figures 9-12 show the data for the investment rate, the growth rate, and the real interest rate (the data are shown in the form 1 + growth rate or 1 -b real interest rate). As in figure 8 above, these data represent period averages. Even when real interest rates were averaged over the nine- and eleven-year periods, for several countries the rates were strongly negative. In looking further at the evidence, we want to separate two issues. First, do positive real interest rates have all the positive effects predicted above, or do they apply only to growth, to investment, or to saving? Second, are adverse effects caused by a regime of negative real rates or by isolated instances of very negative rates? On the first question, the World Bank sample indeed confirms a positive effect of real deposit rates on investment and growth. But, interestingly, there is no significant effect on saving. A key part of the story is missing, and therefore one must ask whether this does not seriously limit any policy implications. To test the second hypothesis--the impact of outliers-a dummy variable was used for countries that had more than three years of strongly negative real interest rates (of less than -10 percent). The results are shown in table 5. Note that the effect of positive real interest rates on growth continues, although the dummy for large negative real interest rates is insignificant. More- Table S. Effects of Sporadic Negative Real Interest Rates on Real Per Capita GDP Growth Rates Constant r DUMI DUM2 I/Y R2 0.88 0.16 -0.007 -0.01 0.36 (44) (2.63) (-0.67) (-2.24) 0.88 0.12 -0.003 -0.002 0.16 0.49 (44) (2.27) (-0.26) (-3.70) (3.94) Note: DUMl refers to cases in which there are more than three instances of very negative real rates; DuM2 is a dummy for the 1965-73 subperiod; r is the real interest rate on deposits; and I/Y is the investment rate. The t-statistics are reported in parentheses. Sources: Author's computations based on World Bank (1989b); Polak (1989). Dornbusch 39 Figure 9. Investment-GDP Ratio and Growth, Thirty-three Economies, 1965-73 Investment-GDP ratio 0.4 0 0.38 - 0.36 - 0.34 - 0.32 - c 0.3 0 0.28 - O o 0.26 0 0 O O 0.24- 0.22 0 0.2 _0 0 Q0 0.18 0 0O 0 0.16 0 oO 0 0.14 0 °0 0 0.12 0 O 0.1 I 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1 + Real per capita GDP growth rate Note: Data are for Algeria, Argentina, Brazil, Chile, C6te d'lvoire, Ecuador, Ghana, Jamaica, India, Korea, Malawi, Malaysia, Mexico, Morocco, Nigeria, Pakistan, Peru, Philippines, Portugal, Senegal, Sierra Leone, Singapore, Sri Lanka, Taiwan, Tanzania, Thailand, ISnisia, lTrkey, Uruguay, Venezuela, Yugoslavia, Zaire, and Zambia. Source: World Bank data. Figure 10. Investment-GDP Ratio and Growth, Thirty-three Economies, 1974-84 Investment-GDP ratio 0.45 0 0.4 - 0.35 - 0.3 -0 0 d co 0.25 O0 0 &- 0 0 0 0 CP ~ 0.2 0 O 0 00 0.15 0 0.1 _ 0.05 il0 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1 + Real per capita GDP growth rate Note: Data are for thirty-three economies listed in note to figure 9. Source: World Bank data. 40 Policies to Move from Stabiliza;tion to Growth Figure 11. Real Interest Rates and Growth, 7hirty-tbree Economies, 1965-73 1 + Real interest rate 1.1 1.05 -0 0 0 00 °° 680 00 0 0 0.95 0 cO 0 0.9 0 0 0.85 0.8 0 0.75 0 0.7 . . . . . 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1 + Real per capita GDP growth rate Note: Data are for thirty-three economies listed in note to figure 9. Source: World Bank data. Figure 12. Real Interest Rates and Growth, Thirty-three Economies, 1974-84 1 + Real interest rate 1.06 0 0 1.02 0 0 00 0.86~~~~~~~~ 0 0 0 0.98 0 0 0.94 0~ P 0 00 0 0.9 0 o 0.86 0.82 0 0 0.78 0.74 0 0 0.7 I 0.95 0.97 0.99 1.01 1.03 1.05 1.07 1 + Real per capita GDP growth rate Note: Data are for thirty-three economies listed in note to figure 9. Source: World Bank data. Dornbusch 41 over, when the investment rate is added as an explanatory variable in the growth equation, the coefficient on the real interest rate remains positive. As Gelb (1989) has noted, the real interest rate must proxy some growth effect different from those identified above. Thus the evidence does not support the view that positive real interest rates promote saving or that a linkage between real interest rates and investment raises the growth rate. Financial intermediation-mobilizing saving and financing domestic in- vestment-makes an important contribution to development. Specifically, for a given rate of saving, it can increase the share that is kept in the home country, and it can raise the efficiency with which the savings are allocated among alter- native investment projects. Curiously, too much financial liberalization may be at cross-purposes with precisely these objectives. Insistence on a full range of capital flight products (such as dollar deposits in domestic banks) or on high real interest rates is likely to be destructive of financial stability and productive investment. V. FROM STABILIZATION TO GROWTH The neoclassical growth models, or the modern versions that highlight exter- nalities, focus on trend growth. They describe economies in which flexibility of relative prices assures full utilization of resources along the path of potential output growth. Policymakers do face these issues, implicitly at least, because the policies they set determine in the long run an economy's incentive structure and hence performance. But the more obvious issue is the short run, in which lack of any growth, certainly in per capita terms, is the most striking challenge. Less than full resource utilization and slow growth, or no growth at all, has to do in part with the level of aggregate demand. After expansionary government policies cease driving an economy, it experiences great difficulty in shifting to a new regime in which growth in the traded goods sector and internal demand, including investment, become engines of growth. The difficulty in restoring growth-quite obvious in Bolivia or Mexico, for example-involves three sets of issues: * Budget correction will have reduced real wages and hence internal demand. Without internal demand firms do not invest. Resources that are freed by fiscal austerity do not find their way automatically into exports or import substitution. * If the exchange rate is highly competitive, this implies that the real wage is very low. Strongly competitive real exchange rates do ultimately support strong export growth. Chile after 1983 documents this, as does Turkey in the early 1980s. But in the short run, real depreciation does exert a contractionary effect on demand (see Lizondo and Montiel 1989 for a review). * Firms' willingness to invest in export expansion or in domestic import sub- stitution depends on their confidence that the regime will not revert to unsound 42 Policies to Move from Stabilization to Growth Table 6. Index of Economic Indicators, Mexico, 1981-87 Indicator 1981 1983 198S 1987 Y 108 102 108 106 YPC 105 94 96 91 Manufacturing Y_ 106 96 106 105 Wm 103 75 71 61 E o105 91 92 85 Note: Index: 1980 = 100. Y = real GDP; YPC = per capita real GDP; Y_ = real GDP in manufacturing; W_ = real product wage in manufacturing; E_ = employment in manufacturing. Source: Banco de Mexico data. economic policies. As discussed in the next section, if there is no front load- ing of incentives, the option to wait may be the best investment. Yet front loading of incentives is difficult because it involves still further redistribution of income. Stabilization may be inevitable, but it is not a ticket for prosperity. Table 6, showing data for Mexico, documents these problems. The risk that stagnation will follow stabilization is thus very grave. Although the problems of stabilization are being recognized, official institu- tions still offer an overly optimistic outlook. The IMF'S rendition of stabilization and adjustment, for example, portrays a rosy scenario (Kahn and Knight 1985). But close inspection of the IMF model reveals that all the crowding-in problems discussed above are solved by assumptions: investment is assumed to rise spon- taneously in response to structural adjustment; real depreciation drives growth immediately; and whenever the economy deviates from full employment the growth rate responds positively to the gap by an unexplained mechanism. In practice, of course, none of these assumptions hold. Unless the export sector rapidly becomes a strong, driving force, growth will not come. If domestic demand is the source of growth, then external constraints soon become binding (see Dornbusch and Edwards 1989). Stabilization often fails, after a while, because of income distribution issues and recession, because the financing for supply side policies that raise growth cannot be marshaled, or because the trimming back of credit growth and the devaluation produce a deep recession and no investment boom-not in the first year and not for many years. If the private sector does not respond with investment and capacity expan- sion, and if confidence and inflation fears bar a public sector expansion, then the policymaker becomes the proverbial emperor without clothes. That is, although the policymaker has sharply increased profitability in the traded goods sector, the profits are expatriated, and there is neither growth nor equity. A simplistic response to this problem is that policy is simply not credible, and that therefore, to no one's surprise, it fails to deliver on its promise. But that response is either tautological or foolish. One should not presume that the Dornbusch 43 market automatically solves the coordination problems involved in the return of flight capital or the resumption of investment. Ultimately, growth will return if the adjustment-induced pricing of resources is competitive by world standards and if incentives are present to save and to keep savings at home. Stabilization and adjustment have to accomplish this. The rest is a slow building of confidence that will develop the political will to go on and resist the (futile) temptation to change course and reverse policies. But there is some room to dampen the undesirable effects of adjustment in the short run, and there is critical room to think through the question of why the return of capital flows is so tricky. A cushion in the short run can be provided by well- designed public works. One form of such public investment is through emer- gency funds that finance local projects and thus provide a shock absorber to the income effects of real depreciation. If the projects are financed externally, and if they have, as they should, little direct import content, then they can help avoid the decline in internal demand. Such a project is being undertaken very effec- tively in Bolivia in the form of the Emergency Social Fund (see World Bank 1989a). The other support for a return of confidence and thus growth has to come from the external side. Domestic production and investment have to become sufficiently safe for people to repatriate their assets and risk their wealth in production at home rather than keeping assets abroad. We now turn to this key problem. VI. THE WAITING OPTION The return of stability requires external resources to support confidence in the exchange rate and make available resources for growth. There are two sources of external resources, debt reduction or a return of flight capital. I concentrate here on the critical question of incentives for the return of flight capital. For some countries in Latin America external private assets are of extraordinary size, certainly more than sufficient to underwrite stabilization and growth if only they could be mobilized. A common problem in the aftermath of stabilization is the lack of capital reflow (this section is based on Dornbusch 1990). Moreover, even if capital does return it is placed in liquid form in financial markets rather than in plant and equipment. Investors have an option to postpone the return of flight capital (see table 7). They will wait until the front loading of investment returns is sufficient to compensate them for the risk of relinquishing the liquidity option of a wait- and-see position (the option value of the waiting approach has been used in this context by van Wijnbergen 1985 and by Tornell 1988; Blejer and Isze 1989 develop an argument similar to that presented here). Real investment is slow to resume because of residual uncertainty whether stabilization can in fact be sustained. Assume that an economy (say, Mexico's) has two states of the world. In the 44 Policies to Move from Stabilization to Growth Table 7. Cumulative Estimates of Capital Flight (billions of dollars) Years Argentina Brazil Mexico Peru Venezuela 1979-82 5.8 25.3 n.a. 20.7 22.4 1983-87 24.8 35.3 3.3 18.9 6.8 n.a. Not available. Sources: Cumby and Levich (1987); update by the author from IMF data on balance of payment statistics. good state the return on an investment is ra. In the bad state it is rb. Investors have the option to invest abroad (say, in Miami) at r*, or at any time to make an irreversible investment in Mexico. Their evaluation of states follows a Markov process: in a bad state there is a probability q of persistence and (1 - q) of a shift to a favorable state. As a sharp simplification, once a favorable state prevails, it is expected to last forever. Investors are assumed to be risk-neutral. How much of a premium, k, over the Miami return is required for an investor to go ahead and invest in Mexico rather than to wait and see, maintaining the option of investing only when the favorable state is verified? The required front- end premium is (9) = [q/(RX -q)] (r -rb) where R* = 1 + r*. This formulation has two key features. First, it confirms Bernanke's (1983) "bad news" principle that the option value of waiting depends only on the bad news, not on the good news, because investors can take advan- tage of good news situations by investing even late. Second, if bad states are persistent, the premium nears the present value of the differential, (r* - rb)/r*. Thus, persistence translates into a sizable front-end premium required to bring about immediate investor commitment. The ideas can be carried a step further if we assume that there is a link between the front-end premium and the extent of capital reflow. Such a relation can exist either because a reflow reduces the probability of a bad state or because it raises returns in unfavorable states and makes conditions more attractive. We assume then that 4 = 0 (K, . . .) with a larger capital inflow, K, reducing the premium-that is, 4' (K) < 0. The excess return on assets in Mexico, m, is taken to be exogenous to the reflow. The criterion for the excess return in Mexico required to induce repatria- tion now becomes m > ¢(K). It can be readily shown that there are two equi- libria. In one case, when the domestic rate of return is insufficient to warrant the risk of repatriation, no capital comes in. In the other case, because enough capital returns, the risk is low, and therefore the required excess return falls off to nothing. The question then is how to trigger this "good" equilibrium. How can governments reassure investors? The common answer is, by bring- ing about a "credible" stabilization. If real depreciation is not sufficient to bring about investment, the government faces a very awkward position. Income is being redistributed from labor to capital, but because the real depreciation is not Dornbusch 45 sufficient, the increased profits go the way of capital flight. Labor obviously will insist on reversing such a policy. Uncertainty is an important feature in under- standing the relationships between real exchange rates and capital flight and in understanding post-stabilization difficulties. The options of postponing repatriation and of postponing investment in plant and equipment, in export markets, or simply in working capital are too valuable, and hence growth does not return. The discussion of the option value of waiting (and the associated credibility issue) highlights one way in which the competitive model fails to address the transition from stabilization to growth. Stabilization by itself is not enough to trigger a virtuous circle. There is a need for a coordination mechanism that overcomes the competitive market tendency to wait. Political Economy The point can be taken a step further to bring in political-economic considera- tions. There are economic equilibria and there are political equilibria. Open economy issues must be modeled with both in mind. An extraordinarily large adjustment in real wages may set the economic incentives right, but in doing so it may also bring about a political situation that is not comforting for investors. Similarly, the direction and even size of required economic adjustments are understood, but politically these are not possible. What then? What markets consider a sufficient policy action may simply be beyond the political scope of democratic governments. In fact, if governments went far enough to create the incentives that would motivate a return of capital and the resumption of investment on an exclusive economic calculation, the implied size of real wage cuts might be so extreme that on political grounds, asset holders might consider the country too perilous for investment. In the aftermath of a major macroeconomic shock, competitive markets by themselves may be unable to restore a good equilibrium. The option value of a waiting approach highlights the critical leverage that developed countries can employ in underwriting (on a heavily conditioned basis, the more so the more effectively) stabilization loans. With such loans in place, private market participants feel comfortable in repatriating their assets. The repatriation in turn ensures that the loans will not effectively be drawn on (just as in the case of a bank run) and that growth resumes. A minimal step in that direction is for industrial countries to support the complete suspension of exter- nal debt service-to commercial banks and to official creditors-for a substan- tial period. Work by the League of Nations in the 1920s provided such pro- grams, and the same are required today (see League of Nations 1926a, 1926b, 1946). VII. CONCLUSIONS Countries that have experienced protracted high inflation, financial insta- bility, and payments crises will not find their way back to growth easily. Their 46 Policies to Move from Stabilization to Growth economies need to achieve not only fiscal reconstruction by thorough budget balancing but also a far-reaching institutional reconstruction that involves a financial system able to provide efficient intermediation and a regulatory and trade regime that helps allocate resources to maximize productivity. When external resources are in short supply, making the most of a country's resources through better allocation of resources is the only way of raising the standard of living. Fortunately, in the aftermath of mismanagement, the scope for such productivity enhancement is often substantial. Economic reconstruction is the work of a decade or more. There are no greater dangers than complacency with an initial stabilization, which leads even- tually to a reversal of sound exchange rate and fiscal policy. Chile's new demo- cratic government seems to be keenly aware of the need to nurture ar:i foster the stabilization in place today. In contrast, Turkey's government is allowing a dramatic slippage of progress achieved in the first part of the 1980s. Reconstruction is necessary, but it is not sufficient. Public external support ultimately must become part of the effort. External support in the form of long- term stabilization loans, heavily conditioned on accomplishment and continuing effort, can help build the bridge by which flight capital returns and foreign direct investment is encouraged to take advantage of fresh opportunities. REFERENCES Bernanke, Ben. 1983. "Irreversibility, Uncertainty, and Cyclical Investment." Quarterly Journal of Economics 98, no. 1: 85-106. Blejer, Mario I., and Alain Isze. 1989. Adjustment Uncertainty, Confidence, and Growth: Latin America after the Growth Crisis. International Monetary Fund Work- ing Paper 89/105. Washington, I).C. Celasun, M., and Dani Rodrik. 1989. "Debt, Adjustment and Growth. Turkey." In Susan Collins and Jeffrey Sachs, eds., Developing Country Debt and Economic Perfor- mance. Chicago: University of Chicago Press. Chenery, Hollis, Sherman Robinson, and Moshe Syrquin. 1986. Industrialization and Growth: A Comparative Study. Nlew York: Oxford University Press. Cumby, Robert, and Richard Levich. 1987. "On the Definition and Magnitude of Recent Capital Flight." In D. Lessard and J. Williamson, eds., Capital Flight and Third World Debt. Washington, D.C.: Institute for International Economics. de Melo, Jaime, and Sherman Robinson. 1990. "Productivity and Externalities: Models of Export-Led Growth." World Bank Country Economics Department Working Paper wps 387. Washington, D.C. Processed. Dervis, Kemal, and Peter Petri. 1987. "The Macroeconomics of Successful Develop- ment." In Stanley Fischer, ed., National Bureau of Economic Research Macro- economics Annual 1987. Cambridge, Mass.: MIT Press. Dornbusch, Rudiger. 1989. Exchange Rates and Inflation. Cambridge, Mass.: MIT Press. Dornbusch 47 1990. "The New Classical Economics and Stabilization Policy." American Eco- nomic Review 90, no. 2 (May): 143-47. Dornbusch, Rudiger, and Sebastian Edwards. 1989. "Macroeconomic Populism in Latin America." National Bureau of Economic Research Working Paper 2986. Cambridge, Mass. Easterly, William. 1989. "Policy Distortions, Size of Government and Growth." National Bureau of Economic Research Working Paper 3214. Cambridge, Mass. Fellner, William. 1976. Towards a Reconstruction of Macroeconomics. Washington, D.C.: American Enterprise Institute. Gelb, Alan. 1989. "Financial Policies, Growth, and Efficiency." World Bank Country Economics Department Working Paper wps 202. Washington, D.C. Processed. Helpman, Elhanan. 1988. "Growth, Technological Progress, and Trade." Austrian Eco- nomic Papers 15, no. 1: 1-12. Khan, Mohsin, and M. Knight. 1985. Fund-Supported Adjustment Programs and Eco- nomic Growth. IMF Occasional Paper 41. Washington, D.C. Layton, Walter, and Charles Rist. 1925. The Economic Situation of Austria. [Report Presented to the Council of the League of Nations.] Geneva: League of Nations. League of Nations [Sir Arthur Salter]. 1926a. The Economic Reconstruction of Austria. Geneva. 1926b. The Economic Reconstruction of Hungary. Geneva. 1946. The Course and Control of Inflation. Princeton, N.J. Lizondo, Saul, and Peter Montiel. 1989. "Contractionary Devaluation in Developing Countries: An Analytical Overview." IMF Staff Papers 36, no. 1 (March). Marfan, Manuel, and P. Artiagoitia. 1989. "Estimacion del PGB Potencial, Chile 1960- 1988." Coleccion Estudios Cieplan, no. 27 (December). Molho, Lazaros. 1986. "Interest Rates, Saving, and Investment in Developing Coun- tries." IMF Staff Papers 33, no. 1 (March). Murphy, Kevin, Andrei Shleifer, and Robert Vishny. 1989a. "Income Distribution, Market Size, and Industrialization." Quarterly Journal of Economics 104 (August): 537-64. . 1989b. "Industrialization and the Big Push." Journal of Political Economy 97, no. 5: 1003-26. Okita, Saburo, and Michael L. 0. Faber. 1989. The Asian Development Bank in the 1 990s: Report of a Panel. Manila: Asian Development Bank. Polak, John. 1989. Financial Policies and Development. Paris: OECD Development Cen- ter Studies. Romer, Paul. 1989a. "Capital Accumulation in the Theory of Long Run Growth." In Robert Barro, ed., Modern Business Cycle Theory. Cambridge, Mass.: Harvard Uni- versity Press. . 1989b. "What Determines the Rate of Growth and Technological Change?" World Bank Country Economics Department Working Paper wPs 279. Washington, D.C. Processed. Sareacoglu, R. 1987. "Economic Stabilization and Structural Adjustment: The Case of Turkey." In V. Corbo, M. Goldstein, and M. Khan, eds., Growth-Oriented Adjust- ment Programs. Washington, D.C.: IMF. 48 Policies to Movefrom Stabilization to Growth Schumpeter, Joseph. 1934. The Theory of Economic Development. New Brunswick, N.J.: Transactions Books reprint, 1983. Song, Byung-nak. 1990. The Rise of the Korean Economy. Oxford, England: Oxford University Press. Tornell, Aaron. 1988. "Credibility and Irreversible Investment and the Tobin Tax." Columbia University Economics Department. New York. Processed. Uchitelle, Elliot. 1989. "The Effectiveness of Tax Amnesty Programs in Selected Coun- tries." Federal Reserve Bank of New York Quarterly Review 14, no. 3 (Autumn): 48-53. van Wijnbergen, Sweder. 1985. "Trade Reform, Aggregate Investment, and Capital Flight: On Credibility and the Value of Information." Economic Letters 19. World Bank. 1989a. "The Bolivian Emergency Social Fund." World Bank Population and Human Resources Department. Washington, D.C. Processed. 1989b. World Development Report 1989. New York: Oxford University Press. PR OC E ED ING S OF T HE WO R L D B AN K ANN U AL CON FE RE NC E ON DE VE LO P MEN T E CON OM IC S 1 9 90 COMMENT ON "POLICIES TO MOVE FROM STABILIZATION TO GROWTH," BY DORNBUSCH Jacques H. Pegatienan Rudiger Dornbusch argues forcefully, and in my view convincingly, that "appro- priate prices" and "competitive markets" brought on by standard adjustment policies may not lead automatically and instantaneously to growth. In support, he points out that the low aggregate demand created by restrictive financial policies and contractionary devaluations stunts domestic markets and growth. In addition, he notes that reviving domestic investment in plant and equipment through spontaneous capital reflows and an increase in private investment requires politically unsustainable depreciations in the real exchange rate. Dorn- busch concludes that external finance may be necessary to help create the appro- priate climate for the repatriation of flight capital and for the resumption of growth. The paper raises three questions around which my comments are organized: * How appropriate are the prescriptions of standard adjustment policy? * What is the importance of the human resource factor in the transition to growth? * Is the real exchange rate policy the main determinant of capital flight? My comments on these questions focus on Sub-Saharan Africa. I. How APPROPRIATE ARE THE PRESCRIPTIONS OF STANDARD ADJUSTMENT POLICY FOR SUB-SAHARAN AFRICA? No doubt adjustment policies are necessary, but the real issue here is the nature of the adjustment. I do not think that one can adequately solve the problem of growth by assuming systematically, as Dornbusch does, that a public spending boom is the exclusive source of the initial disequilibrium. Even if a spending boom were the immediate cause, disaggregation should show that public consumption and public investment may have different effects on the inflation rate and on the balance of payments because of the crowding-in effects public investment has on private investment and on the supply side of the economy. Jacques H. Pegatienan is maitre de conferences at the Centre Ivoirien de Recherches Economiques et Sociales and a professor of economics at the Universite Nationale de C6te d'lvoire. i) 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 49 50 Comment The inflation rate is the main target of standard stabilization policies. I wonder why this should be so, given the experience of the West African Mone- tary Union (WAMU), for example. In Cote d'lvoire, an influential country in the WAMU, we have had predominantly single-digit inflation that has been consis- tently below 6 percent. Even in the double-digit inflation episode of 1974-80, the rate never rose above an average of 17 percent. Netting out the inevitable and uncontrollable imported inflation shows that autonomous inflation is low. Thus, given that both autonomous and total inflation rates in C6te d'Ivoire are well below Dornbusch's 50 percent benchmark for most African countries, why not monetize fiscal deficits, especially when they are caused by productive public investment? To that extent, the "zero inflation target" policy option would indeed unnecessarily forgo growth. Can we maintain external competitiveness by other means when nominal exchange rate manipulation is institutionally impossible, as in the wAMu? Reducing production and trade taxes is a powerful substitute for nominal deval- uation because it improves after-tax profit rates and lowers the cost of inter- mediate inputs. The short-run tradeoffs with reduced fiscal revenues should not be underestimated here, however. Incomes policy was available until recently in countries such as Cote d'Ivoire, where the Phillips curve was generally very flat, because of the weakness of African trade unions and the lack of indexation for African workers. (In the highly segmented labor market of C6te d'Ivoire, the situation is radically differ- ent for the large community of European workers.) But the efficacy of incomes policy, as an ingredient of exchange rate policy, may fall drastically if the recent political turmoil significantly raises the slope of the Phillips curve. High real interest rates are a prominent feature of standard adjustment poli- cies, as an instrument to mobilize saving. My strong skepticism about the utility of high real interest rates grew even stronger after I read this paper's additional counter evidence. Liquidity constraints are so strong in Sub-Saharan African rural economies that high real interest rates may play a significant role only for the minority middle- and high-income groups of countries and individuals. Not only does the interest rate elasticity of saving remain low at this stage but high real interest rates squeeze small- and medium-scale firms further, as working capital becomes more costly and scarce. When a disequilibrium of external origin persistently reduces domestic real income, as it has in Sub-Saharan Africa for the last decade, growth-oriented countercyclical and expansionary policies are indicated. Yet the standard adjust- ment programs supported by the International Monetary Fund and the World Bank have been procyclical in effect, and thus they have unnecessarily amplified the initial disequilibrium. To the extent that their procyclical nature makes them unable to generate growth and hence to raise the debt repayment capacity of developing economies, these programs may end up unnecessarily increasing the debt burden of African economies. Pegatienan 51 II. How IMPORTANT IS THE HUMAN RESOURCE FACTOR IN THE TRANSITION TO GROWTH? I have always believed that economics was primarily about people producing goods and services for people's enjoyment. Too often, however, we are told that economics is more about disembodied markets monitored by faceless and uni- dentified mechanisms. I had hoped that Dornbusch's skillful presentation of the lessons taught by the growth accounting approach would do more justice to the human resource factor, which he recognizes elsewhere in his paper as a major determinant of growth. Instead, he comes back to the usual story of allocative efficiency, trade liberalization, competitive markets, and sound incentive sys- tems. He takes for granted not only that all economic agents are adequately informed and willing to take the risks of transferring resources away from unprofitable activities toward more profitable ones but also that they have all the skills and technological abilities to convert displaced resources into new goods and services at lower cost, as a Schumpeterian innovator would do. In competitive markets, a cost curve is also the supply curve. Thus, the business of equating the price to the marginal cost is both an economic and a technological task, and it requires the producing firm and its labor force actually to have the skills and technological expertise implied by the ruling cost structure. Accordingly, aligning domestic prices with international ones requires develop- ing countries instantaneously to domesticate and use the best skills and technol- ogies available in world markets. This policy recommendation takes for granted either that the local speed of learning new skills and technologies is infinite or that world markets spontaneously and costlessly provide the appropriate know- how. Neither assumption is warranted. In Sub-Saharan Africa disillusionment with public enterprises is deep indeed. This is not so much because Africans harbor substantial doubts about the intrin- sic value of government intervention but because they see that technological considerations in government intervention have been overshadowed by perva- sive corruption, nepotism, and politicized decisionmaking. The idea that the private sector should take over tasks that governments cannot perform compe- tently and equitably is one that I can endorse comfortably. But I believe that Dornbusch has not shown how a mere switch from public to private ownership will transform enterprises into engines of growth. Part of the reason for this shortcoming is that Dornbusch, along with other advocates and practitioners of standard adjustment policy, takes for granted that the market will ensure the existence and availability of the skills and techno- logical abilities required for effective private ownership. He assumes that the market will automatically generate the required supply of risk-taking African entrepreneurs presumably born with the appropriate managerial mentality. But these conditions do not and cannot obtain in most African countries without delay and without government intervention and investment. Hence, I fear that 52 Comment privatization means that foreign financial and technical control over the business sector-which is already overwhelming in French-speaking Africa-will rise beyond tolerable levels. This itself will increase risks and economic uncertain- ties, and, to the extent that it does, it may make privatization counter- productive. Finally, incentive systems are the responsibility of governments. They work through institutions built and monitored by civil servants, whose welfare is generally the target of most fiscal adjustments. How can one expect a govern- ment to perform more efficiently while adjustment measures demoralize and demobilize the very civil servants who are to implement them? There is a serious inconsistency here, unless one assumes that extended-family-burdened and liquidity-constrained civil servants enjoy tightening their own belts. The efficient resource allocation and the alignment of domestic prices with international standards will not automatically promote growth if firms, the labor force, and civil servants do not have the required skills and technological abilities, if they lack the appropriate managerial mentality, and if they are unwilling to take risks or to make sacrifices for the prosperity of others. Markets are prompted by human needs, skills, technological abilities, and risk-taking attitudes and actions. We must acknowledge that the "invisible hand" that coor- dinates these markets must work through a visible, palpable human agency. A necessary condition for growth to resume, the human resource factor, must not be routinely assumed; it must be fostered. III. Is EXCHANGE RATE POLICY THE MAIN DETERMINANT OF CAPITAL FLIGHT? Dornbusch assumes that those who hold assets abroad are either present or potential investors who are forced to invest abroad by noncompetitive rates of return caused by overvalued real exchange rates. This view is simply not correct for countries such as C6te d'Ivoire, where nationals who hold assets abroad are virtually all civil servants. The origin of their assets is the public budget-not private business profits. The assets are hidden abroad and are likely to stay there for safekeeping, not because of the market and credibility considerations Dorn- busch cites. That is why I am very skeptical that external loans will promote the repatria- tion of flight capital. Unquestionably, external loans are needed to finance the necessary countercyclical expansionary policies. One might hope that increased democracy and the growth of responsible national spirit will counteract the pervasive Swiss bank account syndrome. But the resumption of growth also requires the quick emergence of a new domestic managerial mentality and of risk-loving Schumpeterian innovators. Thus, although very stimulating, Dornbusch's analysis and policy recommen- dations fall short of being fully relevant for promoting growth in Sub-Saharan Africa. Human resource constraints there will prevent "appropriate" prices and "competitive markets"-induced incentives, per se-from performing their pre- scribed role as engines of growth. PRO C E ED ING S OF THE W OR L D BANK ANN UAL CONFER EN CE ON DE V EL O PM ENT E CONOM I CS 1 9 9 0 COMMENT ON "POLICIES TO MOVE FROM STABILIZATION TO GROWTH," BY DORNBUSCH Marcelo Selowsky Rudiger Dornbusch discusses an important topic, one that certainly concerns some of us who are working on Latin America. The issue is whether stabiliza- tion guarantees an automatic and speedy recovery of output, let alone sustained growth. We all have to admit that lags in recovery and investment are still not understood well. In discussing these lags it is useful to address the following questions: (1) What structural measures are needed to complement stabilization to support a supply response in the economy? (2) What are the sources of growth during the transition? (3) What determines the waiting period of investors and entrepre- neurs? and (4) What role do external financing and debt reduction play in supporting and speeding the recovery process? My discussion focuses primarily on the first and last of these questions-that is, on issues of design and sequencing of reforms that may speed recovery and that may avoid reversals; and on the role of external financing in supporting each stage in the transition-that is, the precise complementarity between domestic reforms and external financing at each stage. This latter point is important because some may argue that if investment has not yet recovered, there is no need for external financing. I believe, in contrast, that external financing is crucial even before the recovery of investment and growth. This is a point that Professor Dornbusch does not fully develop in his paper. I. STAGES IN THE ADjUSTMENT PROCESS We can look at the adjustment process as comprising several stages-of which recovery of growth is a culmination. For a nation's economy to recover growth, both its citizens and foreigners must recover the desire to invest in that economy. In addition, the social and private rates of returns on that investment must be high and must be expected to remain high. These two preconditions create the potential for sustained growth. In earlier work I have called this Stage III of adjustment (Selowsky 1990). Thus, to reach the potential for growth, an econ- Marcelo Selowsky is chief economist, Latin America and the Caribbean Regional Office of the World Bank. ( 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 53 54 Comment omy must have gone through two earlier stages successfully-recovery of macro- economic stability (Stage I); and liberalization and deregulation of the economy (Stage 11). Fiscal Adjustment Stage I of adjustment, restoring macroeconomic stability, calls for reducing the domestic sources of financing the public deficit. Printing of money and crowding out of domestic capital markets must be sharply reduced. This reduc- tion is required to reduce inflation and its variability and is a prerequisite for stopping capital flight. Thus, Stage I calls for a major fiscal adjustment. But crucial to restoring stability is not merely the size of the fiscal adjustment but also its quality. An increase in the primary fiscal surplus must be undertaken with efficient revenue and expenditure measures, without introducing new dis- tortions or cutting high-priority public investment and targeted social programs. These specific fiscal reforms must be perceived as irreversible and must be strong enough to face the pressures of future parliamentary, municipal, or presidential electoral campaign periods, when public accounts tend to deteriorate. Ensuring an effective and sustained fiscal adjustment requires thorough clari- fication and rationalization of the relations between the central government and central bank, on the one hand, and provincial governments, state banks, the social security system, and public enterprises, on the other. This calls for impor- tant institutional reforms. The non-transparency of these relations and their pernicious effects during election campaigns have been at the root of the unsolved fiscal problems of Argentina and Brazil. Moreover, the sudden emer- gence of strong fiscal deficits during the recent preelectoral periods in Costa Rica and Uruguay also shows how fragile progress can be. Thus, unless governments support the fiscal adjustment with appropriate institutional reform and coor- dination, the adjustment will not be perceived as sustainable and permanent. This itself will influence the waiting period of investors and the reflow of capital flight. In brief, the main challenge during Stage I is this: to reduce macroeconomic instability sharply, the domestic sources of financing the deficit must be cut swiftly. Yet it takes time to increase a primary fiscal surplus efficiently. It takes time to implement a good tax reform, to privatize inefficient public enterprises, and to reallocate expenditures toward targeted social services. Thus, we have a timing problem-a tradeoff between reducing macro instability swiftly and achieving a high-quality increase in the primary fiscal surplus. It is here that external financing becomes crucial; it can bridge the difference between the quick reduction required in the domestic sources of financing the deficit and the gradual but sustained increase in the primary surplus. Because it allows an efficient fiscal adjustment-one that is perceived as per- manent and less susceptible to reversal-external financing reduces the lag in the response of investors to further reforms. This will become clearer in what I turn to next. Selowsky 55 Liberalization and Deregulation of the Incentive System Stage II involves structural reforms that improve resource allocation and the mobility of factors. Liberalizing trade, deregulating the financial sector, decon- trolling prices, and opening of sectors previously reserved for the state are typical examples of such reforms. In Latin America, we have seen that undertak- ing these reforms prematurely-without first firming up the fiscal reforms of Stage I-is risky. We have seen, for example, that the private sector will not react to improved incentives and will not reallocate resources if it believes the incentives may be reversed by fiscal problems. The reason is that many of the distortions being removed in Stage II were introduced originally because of fiscal problems. Export taxes and high mandatory reserve requirements in the bank- ing sector are typical examples. Thus, the supply response to Stage I-type reforms will depend on the expected permanency and irreversibility of the fiscal reforms undertaken in Stage I. In the end, external financing influences the supply response in Stage II because it supports a "high quality" fiscal adjustment during Stage I. In Stage II we still do not expect a strong recovery of the desire to invest in the country. Nor do we expect a capital reflow. Investors are still waiting to see whether the reforms will stick. But we may expect a reallocation of resources, improved capacity utilization, and increases in the internal efficiency of enter- prises as a result of increased competition. These factors were important sources of growth in Chile during the 1985-88 period, for example, when net private investment still remained low. They could be important as well in flexible econ- omies such as Brazil's, where it is easy to move resources and adapt capacity between the nontraded and traded sectors. But increased capacity utilization and readaptation require higher levels of imports. Countries will not be able to finance increased imports with their exports alone and simultaneously serve their entire external interest bill. Such nations require external financing to bridge this gap, particularly if their exports respond with a lag to the improved domestic incentives. External financing thus has a crucial role in financing the recovery of imports during this stage of structural reform. Recovery of Growth and Debt Reduction It is only after Stages I and II are complete that we can expect a significant reflow of capital and recovery of the desire to invest. Only then will external financing feed into additional investment, supplementing national saving. Capitalizing a fraction of contractual external interest payments and reducing debt service are alternative mechanisms of external financing. And we have learned that reducing debt and debt service has additional benefits. For example, reductions in the public component of the external debt diminish the likelihood that the government will have to raise taxes to serve that debt; this increases the expected private returns to capital and thus encourages the desire to invest in Stage III. It also reduces uncertainties in exchange rate and domestic financial 56 Comment markets by providing a permanent and predictable foreign exchange cash flow. In Mexico, domestic interest rates went down 20 percentage points after the debt deal. II. CONCLUSIONS I share Dornbusch's emphasis on the critical role of external financing- including multilateral financing-in supporting the adjustment process and the eventual recovery of growth. But I must note that his paper does not spell out fully the different mechanisms by which such assistance supports each stage during the adjustment and hence may contribute to shortening the waiting period for investors and capital repatriation. To put it in his terms, we need a fuller theory of (1 - q), the probability of a shift to a favorable "state." Also, we cannot expect a favorable state to last forever; the probability of reversals must be introduced. This is a key element, in my judgment. Most important, (1 - q) is a function of past events-that is, it depends on a recursive process of past actions and reforms. In my discussion 1 have tried to identify the different stages in the adjustment process, to suggest the manner by which each stage can influence the lag time of investors, and to identify the role of external financing in supporting the pro- cess. In other words, I have attempted to see how external financing at each period (t - 2, t - 1, and so on) may affect the value of (1 - q) at period t. Dornbusch emphasizes the stability of the exchange rate and the amount of resources available for growth as the factors influencing investors' decisions. But I would give equal weight to the quality and sustainability of the fiscal adjust- ment in conditioning the attitude of investors, because quality and sustainability are the main factors influencing the probability of reversals. I am suggesting, then, that a large part of external assistance should be directed toward Stage I reforms-that is, toward helping countries undertake deep and durable reforms in the public sector early in the adjustment process. REFERENCE Selowsky, Marcelo. 1990. "Preconditions for the Recovery of Latin America's Growth." Finance and Development 27, no. 2 (June): 28-31. PRO C E ED ING S OF T HE W OR L D BANK ANN UAL CON FE RE NC E ON DE VE LOP MEN T EC ONO MI CS 1 9 90 FLOOR DISCUSSION OF THE DORNBUSCH PAPER Dornbusch thanked both discussants for their comments, and the chairman invited observations and questions from the floor. A participant noted that India seems to represent a counterexample to Dornbusch's scenario for the transition from stabilization to growth; despite the fact that in terms of efficiency and distortions India's policies are arguably "all wrong," its economy has been grow- ing by approximately 5 percent per year. Another participant, referring to Pega- tienan's criticism of Dornbusch's paper for slighting the elements of human capital involved in productivity growth, suggested that perhaps there was room to broaden the Dornbusch scenario, given that Dornbusch has suggested that productivity growth depends 50 percent or less on capital. Dornbusch acknowledged that his paper did not specifically address India-or Africa, where inflation is not a central problem-and suggested that his main focus was Latin America, his most familiar ground. He pointed out that he regards Latin America as something of a testbed for the problems of the next fifteen years in Eastern Europe, where questions of privatization, hyperinfla- tion, fiscal policy, and exchange rates are becoming crucial. Poland, for exam- ple, in many ways is recapitulating the experience of Argentina in terms of political traditions and economic problems. Dornbusch disagreed with Selowsky's recommendation for heavy and early support for fiscal adjustment, suggesting that he believes that current interna- tional lenders lack the authority-as exercised by League of Nations commis- sioners in post-World War I reconstruction in Austria, for example-to ensure local government compliance with budget balancing and debt reduction targets. More crucial than a perfectly balanced budget, Dornbusch argued, is support at the right time for the transition from stabilization to growth by the World Bank and International Monetary Fund. Thus, for the Bank and the Fund, it should not be so much a question of trying initially to create a favorable climate for capital reflows from nonresident investors as it should be of responding pos- itively and significantly-whether this is sooner or later-to evidence of substan- tial achievements in stabilization and structural reform. In response to a participant who suggested that Dornbusch's suggestion of a 50 percent uniform tariff still represents a major bias and that this bias would impede a necessary export response and require keeping real wages low, Dorn- busch cautioned against excessive trade liberalization. He noted that it is one This session was chaired by Leopoldo Solis, director general of the Instituto de Investigacion Economia Social de Mexico. ( 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 57 58 FloorDiscussion of Dornbusch Paper thing to liberalize imports of capital goods and intermediate goods, but Milky Way candy bars are a high-gross item in Mexican imports, for example, and no one would argue that they are essential to growth strategy. Hence, a 50 percent uniform tariff may go halfway toward opening trade; it introduces the notion of competition but avoids the risk of having to cut wages to pay for nonessential imports. Dornbusch cited as impressive the 1958 liberalization in the Federal Republic of Germany, which took place at a time of a domestic boom and current account surplus. That is the golden time to liberalize, he argued, because "you can pay for it and you want the [imported] resources." Dornbusch concluded that one might like to liberalize earlier, but "if you can't pay for it, don't risk the inflation [and macroeconomic instability] that would come from an exchange rate collapse. " PR OC E E DING S OF THE W OR L D BANK ANNUAL CON FE R EN C E ON DE V EL O PM ENT EC ONO MI CS 1 9 9 0 Macroeconomic Policy and Growth: Some Lessons of Experience W. Max Corden This paper asks how macroeconomic policies affect growth. It draws on the experiences since 1974 of seventeen developing countries-Brazil, Cameroon, Chile, Colombia, Costa Rica, C6te d'Ivoire, India, Indonesia, Kenya, the Republic of Korea, Mexico, Morocco, Nigeria, Pakistan, Sri Lanka, Thailand, and Turkey. First, the paper looks at the effects of the foreign-financed public spending booms of the 1 970s, which tended to destabilize economies and lead to debt crises, initially raising and then lowering growth rates. Next, the paper compares reactions to the 1980-82 crises and adjustment poli- cies in four countries that were relatively successful-Colombia, Turkey, Thailand, and Indonesia. The paper reviews the varying experiences with inflation of the countries in the study and assesses the effects of inflation on growth. The generally low-inflation countries experienced short bursts of high inflation caused by external shocks. Finally, the paper analyzes the relation between inflation and exchange rate regimes and describes the exchange rate policies of Indonesia and Mexico in detail. The paper draws a number of policy lessons-notably the need to do cost-benefit analyses for public investments, the need to avoid "euphoria" when the economic outlook appears favor- able, and the need to react speedily to crises. Nominal exchange rates should be adjusted to avoid unstable real rates, and a flexible exchange rate policy should be combined with a commitment to a noninflationary monetary policy. How do macroeconomic policies affect growth in developing countries? Can we derive some lessons of experience to guide us in determining which policies are relatively more favorable to growth? To gain some insights into this subject, I survey seventeen developing coun- tries, most of which have gone through episodes of public spending booms, crises, and adjustment since 1974. I am able to take this overview because this paper is a by-product of a World Bank research project on macroeconomic policies, crisis, and growth in the long run, which involves comparative studies of seventeen developing countries over a long period (usually since the W. Max Corden is professor of international econormics at the Paul H. Nitze School of Advanced International Studies of Johns Hopkins University and a consultant to the World Bank. He wishes to acknowledge comments on an earlier draft by Vittorio Corbo, John Cuddington, and William Easterly. ( 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 59 60 Macroeconomic Policy and Growth mid-1960s).1 My aim is to see what we can learn about the effects of macro- economic policies on growth, both in the short and long terms. However, I do not try to explain why governments pursued various macroeconomic policies at different times; I am not engaging in political economy. The methodology of this paper is eclectic. Unlike much generalizing and model-building in academic development economics, the present study is not based on the experience of only one or a few countries. Nor does it take an econometric approach that conceives of many countries as representing, in some sense, a common population. Rather, it attempts to take into account the detailed experiences of a large number of countries.2 This approach is charac- teristic of much World Bank research. Of course, in a conference paper one can include only so much detail, evidence, and qualification. Thus the "lessons" I relate at the end of each section of this paper should be understood as based on much more evidence than I can present here. Similarly, they should be seen as subject to numerous qualifications that I also cannot elaborate here. In short, I am posing these lessons as a basis for discussion. The paper draws extensively on the experiences of sixteen of the countries in the World Bank study: Brazil, Cameroon, Chile, Colombia, Costa Rica, C6te d'lvoire, India, Indonesia, Kenya, Mexico, Morocco, Nigeria, Pakistan, Sri Lanka, Thailand, and Turkey. The seventeenth country in the World Bank study is Argentina, but it is such a special case that I mention it only sporadically here.3 I have brought in Korea, however, so the total number of countries studied is still seventeen. Korea is notable because it has had experiences similar to those of many of the other countries, but, as is well-known, its policy reac- tions and growth outcomes have been exceptionally favorable. Table 1 presents some basic data on growth rates, inflation rates, investment ratios, and external debt for these seventeen countries.4 In the last eight years or so, the academic literature on macroeconomic poli- 1. I am indebted to the authors of the forthcoming country studies, as well as to numerous published papers on some of these countries. Here I would particularly like to note the country studies contained in Sachs and Collins (1989) and Sachs (1990). The work of synthesis, drawing on the various country experiences, is still in process, and is being done by Richard Cooper, Ian Little, Sarath Rajapatirana, and myself. This paper has drawn on our ongoing work, and I am indebted to my co-authors. The views and judgments expressed here are purely my own at this time, and may not coincide with my views or those of my co-authors when this work is completed. 2. 1 am in sympathy with Reynolds (1985), who has sought an understanding of growth through in- depth historical studies of many countries and has looked for some comparisons and generalizations from these. He argues that "it is wrong to regard cross-section analyses as a satisfactory substitute for longi- tudinal studies" (p. 13). 3. Inevitably, I shall refer to Argentina in the section on inflation. 4. All large developing economies (except Egypt, Korea, and Venezuela) are included in the project, nine in all. A large economy is defined as any nonsocialist developing economy with a GDP greater than US$30 billion in 1987 (billion equals 1,000 million). Medium-size economies (US$10 billion to US$29 billion) are Cameroon, Chile, Morocco, and Nigeria. Small economies are Costa Rica, Cote d'lvoire, Kenya, and Sri Lanka. Corden 61 Table 1. Growth, Inflation, Investment Ratios, and Debt: Seventeen Developing Countries (percent) Gross domestic Total external 1980-1988 investment as a debt as a Average annual Average annual percentage of percentage of growth in per rate of inflation GDP GDP Country capita GNP (CIi) 1980 1988 1980 1988 Korea, Rep. 6.9 8.7 31.7 32.6 48.7 22.0 Thailand 4.1 5.9 26.4 25.8 25.9 36.1 India 3.7 9.4 22.8 21.5 11.9 22.3 Pakistan 3.7 7.2 18.5 17.6 42.5 45.6 Cameroon 3.1 9.5 18.9 14.0 36.8 33.7 SriLanka 2.8 13.0 33.8 22.8* 46.1 75.1* Indonesia 2.6 10.0 24.3 28.0 28.0 69.0* Turkey 2.1 50.1 21.9 19.0 34.2 57.7* Morocco 1.8 8.1 24.2 24.2 56.2 105.9* Colombia 1.3 23.2 19.1 17.5 20.9 46.5* Chile 0.8 21.9 21.0 15.9 45.2 96.6* Brazil 0.6 212.2 22.9 16.7 30.6 30.7 Kenya 0.3 10.9 30.0 19.0* 51.2 71.3* Mexico -0.6 74.5 27.2 16.7* 30.3 58.0* Costa Rica -0.9 28.3 26.6 23.9 59.5 100.0* C6teD'Ivoire -3.3 6.4 28.2 11.6* 58.3 161.8* Nigeria -3.8 17.9 20.5 8.1* 8.9 107.1* Note: GNP is gross national product; GDP is gross domestic product; cpt is consumer price index. Countries are listed in order of per capita GNP growth. Countries in which the Investment-GDP ratio fell by 10 percent or more between 1980 and 1988 are marked with an asterisk in the fourth column. Countries in which the debt ratio increased by more than 20 percent of GDP over the period are marked with an asterisk in the last column. Source: World Bank data. cies in developing countries has blossomed. Yet this literature has been domi- nated by attention to a few Latin American countries-particularly Argentina and Chile but also Bolivia and Brazil. For all four of these countries, economists have been primarily concerned with analyzing the processes of reducing or attempting to reduce high inflation. By broadening the scope of investigation and giving a little more attention to the non-Latin American countries, one is reminded, for example, that inflation is not a general characteristic of develop- ing countries. Section I of the paper deals with the public spending booms that destabilized various developing countries at different times, usually between 1974 and 1980, and that together with terms of trade and interest rate shocks brought about fiscal imbalances and accumulations of debt that caused the subsequent crises and need for adjustment. Section 1I deals with the crises of the early 1980s and subsequent macroeconomic adjustments. Section III discusses the relation between inflation and growth, and section IV treats exchange rate policy. My major interests are in the implications of various macroeconomic policies for growth in each of the seventeen countries and the lessons that may be derived from looking closely at their experiences. 62 Macroeconomic Policy and Growth I. THE PUBLIC SPENDING BOOMS A Stylized Story of a Boom Let us begin with a simple stylized story of a public spending boom. Of course, this story actually varies greatly between countries, and I will consider these variations later. In our hypothetical case, first, there is an increase in public spending, whether on consumption or investment, caused above all by the ready availability of funds from the world capital market. The extra spending goes both toward home-produced goods and toward imports. The higher demand for home- produced goods increases domestic output, given that there is some slack in the domestic economy. This is the Keynesian effect, which manifests itself in a short-term rise in the growth rate. (The Keynesian effect on output may last for a time after the growth rate declines, but it will be temporary when the boom itself is temporary.) In addition, extra demand for imports worsens the current account. The adverse effect of the spending boom on the budget is moderated by higher tax revenue resulting from domestic output expansion and by the fact that increased imports bring in mnore revenue from taxes on trade. In due course-usually very quickly-domestic prices rise in response to the higher demand. Given a fixed nominal exchange rate (or one that is not depreci- ated much in response), this produces an appreciation of the real exchange rate, with the usual expected effects. The higher domestic costs, including wages, squeeze the profitability of exports. In addition, domestic demand switches toward exportables and toward imports, especially imported inputs into domes- tic production. Hence the current account worsens further. The precise effects depend, of course, on such matters as the composition of extra government spending-in particular its import content (or, more gener- ally, its tradable content). This in turn may depend, among other things, on how much is extra investment spending and how much is consumption spending. Its subsequent effects-its effects on real output and income, and hence on the medium- and long-run rates of growth-are of particular interest for this paper and are discussed below. Obviously they depend, above all, on how much has been extra investment spending, how efficient the investment has been, and what debt burden has been incurred. Inevitably, the real appreciation would have to be reversed once the spending boom comes to an end, and this can also be regarded as an important medium-run effect. Because of the Keynesian output and income expansions, private consump- tion spending will rise. Private investment spending will rise if there is a favor- able expectations effect. At the same time, or possibly before the spending boom was embarked on, there may have been a rise in export income caused by improved terms of trade. This may have led directly to higher private incomes, which are then spent and have the effects just discussed-Keynesian expansion, real appreciation, increased imports, and a squeeze on those exports that did not benefit from the Corden 63 boom. The rise in export income will also have led to higher government reve- nues, which are then spent, as before, with the usual effects. The real apprecia- tion resulting from such an export-financed spending boom leads to the familiar Dutch disease-a squeeze on nonboom ("lagging sector") tradables. When the spending boom is financed out of the gains from an export boom there will not be an increase in foreign debt. Indeed, the export boom may exceed the spending boom so that reserves are actually built up, as happened in a few countries. In other countries the spending boom was greater. The Booms: What Actually Happened The stylized story gives a general indication of what happened in the seven- teen countries, but there were great variations (for other comparative studies of booms, see Gelb and others 1988 and Cuddington 1989). The common feature is that all except Chile and India had a public investment boom at some stage during the 1974-81 period, and in almost all cases this led to increased foreign debt. The fact that this was so common is quite striking, especially considering that during the same period, developed countries were generally going through reduced growth and were trying to restrain public spending.5 The explanation of the booms is a matter of political economy, and here each country appears to have its own story. Yet there must be some common factors-the most obvious being the easier availability of funds on the world market. Despite some commonalities, the booms differed greatly in timing, magni- tude, and details. India did not have a boom and borrowed very little. Did India thereby miss opportunities? Chile experienced only a private investment and consumer durables boom. (The debts incurred by private borrowers in Chile were taken over by the government because of the crisis and pressure from foreign creditors, so the net result for public external debt was much the same as if the boom had been public originally.) In the three oil-exporting countries-Indonesia, Mexico, and Nigeria-the booms were clearly initiated in response to higher actual or expected export income. In coffee-exporting Kenya the situation was similar. In Cameroon a modest investment boom-in particular, investment in oil exploitation- followed the coffee boom and preceded the oil boom. Morocco's public invest- ment expenditures had already started to increase before the phosphate export boom but then greatly increased as a result of this boom. In Costa Rica and C6te d'Ivoire also, the public spending booms started before the export booms but were indeed kept going longer than otherwise as a result. It should be pointed out that Mexico actually had two spending booms. The larger one was in 1977-81 and was clearly initiated by expectations of high oil- export income. The smaller and earlier boom, during 1972-76, however, can be S. This very factor caused demand for funds to decline and real interest rates to be low, and so made banks more eager to lend to developing countries and made governments of developing countries more ready to borrow. 64 Macroeconomic Policy and Growth explained in domestic political terms and by the ease of foreign borrowing. It was not associated with an actual or prospective export boom. Colombia is a special case. It had an export boom in coffee but not a public investment boom at all during the export boom period. Rather, the gains went mostly to the coffee producers; the monetary effects were partly sterilized; and the budget deficit was reduced. Colombia's story is resumed later, but it should be noted now that Colombia firnally did have a public spending boom-one that came as a reaction to the decline in the coffee price and that was designed to stabilize domestic demand. The accounting so far leaves six countries that had public investment or spending booms of some kind but no export booms at all during the period. Sri Lanka's boom was particularly big and is discussed below. Turkey had its boom roughly from 1974 to 1977, encouraged both by the ease of borrowing abroad and by a 1971-73 boom in remittances from workers abroad (this was similar in effect to an export boom). Pakistan had a public investment boom in 1974-76 and was able to finance its current account deficits through cheap loans from some of the oil-producing Gulf countries. Korea increased its investment ratio (private and public) from 1974, and in 1978 and 1979 it went into a short-lived "maxiboom" resulting from a major investment drive in heavy and chemical industries. Even cautious Thailand followed the fashion and went into a modest foreign-financed public investment boom in 1977-78. Brazil maintained a high level of spending despite the adverse effects of the oil price rise, and it borrowed heavily, but the investment ratio only rose slightly in 1974 and 1975; much of the foreign-financed investment was done by parastatal enterprises. Many of the countries built up big debt burdens during this period. India was an exception. Other notable exceptions were Cameroon, Colombia, and Nigeria-all countries in which the export boom exceeded the spending boom and in which reserves were built up. Pakistan also did not really develop a debt problem because the concessiona[l nature of its loans made interest payments low and because the nation borrowed only modestly. Implications of a Boom for Growth This paper began by asking how macroeconomic policies have affected growth. The public spending booms leading to fiscal deficits that have been described here were indeed the initiating macroeconomic policies for the whole long episode I am discussing. How has growth been affected? For many countries the data tell a simple story. During the booms, when the investment ratio rose, the rate of growth of output also rose, and because much of the boom was financed by foreign borrowing, spending grew even faster. Later, after the crises began, both the investment ratio and growth fell, often drastically. As for retained national income, because of debt-service payments, the rate of growth-and sometimes even the level-fell even more. What, then, has been the net effect of the whole episode on growth and on real incomes? First, we must distinguish effects of investment booms on rates of growth Corden 65 from effects on levels of output. It is really rather misleading to focus on growth rates. Suppose there is an investment boom in year t1, and this bears fruit in a higher output level from year t3 onward. The rate of growth will then rise in year t3 and fall back again to its trend value the next year. It is not a failure of the investment boom that it has not led to a permanent rise in the growth rate. Second, we should distinguish boom effects on "Keynesian growth"-or demand-determined growth-from effects on growth of capacity. It seems obvious that the increases in the growth rates in the years of the public spending booms were Keynesian. They had nothing to do with the productivity of the investment and its possible effects on long-term growth. It is not necessarily an achievement for the macroeconomic policymakers to bring about growth of this demand-led kind because such growth is not sustainable. As I have just sug- gested, a Keynesian demand expansion leading to output levels above trend-a boom-for several years may well cause the rate of growth to rise at first and then to fall and become negative in the later years of the boom. The issue is whether a boom yields fruit later. Let us focus on this and hence on the possible effects on incomes and output. One possibility is that public investment is efficient in an ex post facto sense. Perhaps output rises as a result (when converted into tradables), and it rises sufficiently to finance the debt service and something extra. In that case it will have had a favorable effect not only on output but also on national income (that is, after net factor payments). The spending will temporarily have raised the rate of growth, and it will have made the country permanently better off. The rise in debt, even possibly in the ratio of debt to gross domestic product (GDP), should not be a problem. The fact that the so-called resource transfer resulting from the investment becomes nega- tive once the investment ceases should not be a matter of concern.6 It is also possible that the investments could have been efficient in an ex ante sense-that is, they were efficient, given available information and reasonable expectations at the time, allowing for risk. But because of unexpected adverse terms of trade and interest rate movements they turned out to be uneconomic. This might be a good description of what happened in many cases, although some of the investments were clearly not efficient even in this sense. In any case, when investments proved inefficient it could still mean that output growth was boosted for a time, and levels stayed higher than they would have been otherwise. But a debt problem was generated or increased, forcing reductions in absorption below levels that could have been sustained in the absence of the earlier boom. Thus the effect on retained income (gross national 6. When the public investment boom is financed from a country's own export income boom, the story has to be a little different. To judge the efficiency of the domestic investment, the stream of extra output has to be set against the earnings that could have been obtained by investing the gains from the export boom abroad. If domestic investment is inefficient there will have been the same sort of loss, as discussed in the text, but it need not lead to or help to bring about a debt crisis in the direct sense. It will, however, still reduce the country's creditworthiness (compared with investment abroad), and hence it will have an indirect effect. 66 Macroeconomic Policy and Growth product) was negative even though the effect on output (gross domestic product) may have been somewhat positive. In addition, the debt crisis had indirect adverse effects on outputs and real incomes, and possibly growth later, and these must be attributed to the earlier boom, which contributed to the buildup of debt. First, in some cases (notably Brazil) the crisis led to an increase in trade restrictions. In other cases (notably Mexico, Morocco, and Turkey) the crisis led to trade liberalization, so that there was presumably a positive productivity-and possibly growth-effect. Second, the crisis forced reductions in imports that caused outputs of import- using industries to decline and, given some immobility of labor, at least transi- tional unemployment. Note that there is empirical support for the proposition that over a longer period (1965-88) countries that on the average had high investment ratios tended also to have high growth rates. Yet for the 1975-88 period, a clear time- series relationship between investment ratios and rates of growth of output a few years later-when the fruits of the investment would be expected-cannot be found. But this is no surprise when we recall that during the boom period, investment ratios and rates of growth were high, and during the crisis and early adjustment period a few years later both were low (the earlier high growth being explained by Keynesian effects and the later low growth by various factors, including the consequences of the inefficient investment earlier). This has also meant that there is a tendency toward a negative cross-country relationship between investment booms in the 1974-80 period and rates of growth from 1982. Lessons of the Boom It is important to learn the lessons of this 1970s boom because some of the circumstances that gave rise to it could come about again. The lessons should be remembered whenever the economic situation for a country appears to be really favorable. When the crisis comes it is too late to avoid pain. How much macroeconomic destabilization can be tolerated? Suppose a coun- try looks as though it is moving into a foreign-financed investment boom, public or private, and all the evidence suggests that it is soundly based. If debt- financed, the expected interest rate over the relevant period is low relative to expected returns. Alternatively., the boom may be the result of direct private investment, which is unsubsidized and sound on proper cost-benefit calculations (though these may not necessarily be calculations that take into account the considerations discussed here). But the boom is inevitably temporary. Some macro-destabilization is then inevitable. During the boom the current account will go into deficit (or greater deficit than before), the real exchange rate will appreciate, and there will be some domestic inflation, all features described in the stylized story earlier. It is important that price rises are perceived as temporary adjustments, so that inflationary expectations are not generated. Later these effects will tend to be reversed. This prospective destabilization has Corden 67 to be recognized-and should be foreseen-but it is not sufficient reason to try to kill a boom completely. Nevertheless, there is then a need for stabilizing or smoothing policies-notably reducing public investment when the boom is a private one, or moderating public and private consumption spending when the source of the boom is the availability of funds from abroad for investment. The adverse (Dutch disease) effects on the tradable industries of temporary real appreciation are a particular problem; they strengthen the case for stabilizing policies. Unsound investment spending shows the need for a cost-benefit ap- proach. Our study of the seventeen countries shows that some public spending decisions were unsound on the basis of knowledge and reasonable expectations at the time. Some projects were bound to be unsound because of the speed of the decisionmaking processes, the rapidity with which the programs themselves were implemented, and the massiveness of the investment. Many of the invest- ment booms and borrowing splurges took place in very short periods and cre- ated debt burdens that have affected the countries adversely for many years. The obvious need is for a cost-benefit approach, preferably one that is institu- tionalized. This lesson is as important for creditors (private or official) who are eager to lend as it is for the governments. In particular, when the investment is in nontradables, lenders and governments need to take into account the likelihood of a later real depreciation. The need to reverse the resource transfer-as new capital inflow declines and especially as interest and dividend payments have to be remitted-means that there will have to be some real depreciation later, at least relative to what might have happened otherwise. If new investment took this into account it would tilt the pattern of foreign-financed investment in the direction of tradables. At the same time, cost-benefit analysis that makes some use of shadow pricing would tilt new investment away from highly protected (usually import-replacing) industries. The net effect would be that investment in export industries, or in direct or indirect inputs into exports, would be favored. Beware of euphoria. It is well at this point to recall the experiences of three countries-Chile, Sri Lanka, and Cote d'Ivoire. In Chile there was massive private borrowing during a very short period- 1980 and part of 1981. This much-analyzed episode teaches at least two lessons. The first is how quickly problems can be created that take years to solve. The second is the need for caution against "euphoria." In this case, it was a euphoria of the international capital market and of domestic investors about Chilean economic prospects as a result of policy changes believed to be favorable to private enterprise, and the building up of confidence in the private sector that this situation would be sustained. Sri Lanka's boom began in 1977 when a new, liberalizing government came into power. From 1977 to 1982, public spending rose from 28 to 41 percent of GDP. The previous government, like the Indian government, had been reluctant to borrow abroad and possibly could not get much from official sources. When a government came into power that liberalized trade, reduced subsidies, and 68 Macroeconomic Policy and Growth liberalized the financial sector in a country that had been very protectionist and regulated, funds from official sources, including the World Bank, became readily available. Here again, as with Chile, we had euphoria, now on the part of the government itself, which undertook an ambitious investment program (mainly a single irrigation project), and on the part of official lenders. Finally, C6te d'Ivoire teaches similar lessons. In 1974 public investment was 11 percent of GDP, and by 1978 it was 21 percent. Until 1974 Cote d'Ivoire was considered a great success story because it had foilowed policies that were sympathetic to private enterprise and foreign investment and had attained the highest per capita growth rate in sub-Saharan Africa. This all fell apart because of a massive public investment boom embarked on in 1974-before the coffee boom began-whose consequences are still being felt in a very high debt burden and low growth rate. The foregoing are extreme cases of euphoria in which governments had a sound record or appeared to be following or planning orthodox policies. Another well-known example is that of Mexico in 1980-81, when the spending boom far exceeded the oil export boom. In general, euphoria was characteristic of the whole period among many countries that had export booms, or countries that found it easy to borrow abroad even without such booms. There are lessons here not only for various governments but also for potential foreign creditors. They should be remembered whenever a government, its poli- cies, and the country's economic prospects look "really good." II. CFLISIS AND ADJUSTMENT The broad characteristics of the crises that befell almost all of the countries in the study from about 1980 to 1982 (poorer terms of trade, higher real interest rates, decline of foreign lending) are well known and will not be described here. But once one gets beyond aggregate data and simple econometrics, one becomes aware of the enormous differences between countries-in the extent of the crises; in the precise origins of the crises (terms of trade, interest rates, decline of lending, and internal factors); in the timing; and, above all, in the policy reac- tions. The onset of the crisis was always a decline in the availability of new loans from the private capital market, and the countries that avoided an immediate crisis were principally those that could and chose to continue borrowing for a time. The timing of the terms of trade shock depended crucially on whether the country was an oil exporter or importer, or a coffee exporter. Pakistan was the one country in our project that did not really have an actual or potential crisis of the kind that affected the other countries, because the adverse effect of the oil price rise was offset by the boom in remittances from workers in the Middle East and because Pakistan did not have a significant debt-servicing problem. For this paper one would ideally like to determine how the immediate policy Corden 69 reactions affected later growth. Two aspects were probably critical. First, one might presume that if there is continued borrowing to maintain consumption levels (as in Brazil) or to finance doubtful investment programs (as in Sri Lanka), the increased debt burden would have an adverse effect on future growth, as discussed earlier. Second, many countries tightened import restrictions. This is a natural reaction at a time of balance of payments crisis, but, if maintained, it is harmful for the growth of exports and economic growth later. The question then is whether the tighter restrictions lasted for some time. In some countries (Mexico, Morocco, and Nigeria) structural adjustment programs led later to major trade liberalization measures that have more than reversed the earlier increases in restrictions, but this was not so in other cases-notably Brazil and Colombia. Actual short-term outcomes depended not only on the policy reactions but also on the initial shocks and their effects. Countries with flexible economies and flexible policies-Indonesia, Korea, and Thailand are examples-managed to recover quickly or keep their domestic recessions short. Going beyond the crises to subsequent developments, some countries went through major adjust- ment processes-Turkey from 1980, Morocco from 1984, Colombia from 1984, and Nigeria from 1986. Usually these involved a significant devaluation and subsequent trade liberalization. Before discussing a few countries in more detail, I would like to suggest a general lesson. Unfavorable shocks and surprises can hardly be avoided, and it is difficult to conceive of crisis management that does not involve at least some pain. The most that some countries could do was to postpone the pain. The main lesson is that countries should aim to establish favorable initial conditions in good times. They should avoid big spending booms; keep their debt ratios fairly low, which is primarily a matter of pursuing conservative fiscal policies; and keep their reserves high, especially when there is heavy dependence on a few volatile exports. They should make their economies and policy reactions as flexible as possible. Some initial conditions can be influenced in advance, but once the crisis comes, all the initial conditions are beyond control. Four Cases of Crisis Management and Adjustment Let us now look briefly at the experiences of four countries that have been reasonably successful and see whether there are any particular lessons that emerge. The greatest success, Korea, is not discussed here because it is so well known (see Aghevli and Marquez-Ruarte 1985, and Collins and Park 1989). Colombia. When other countries were in crisis, Colombia was not, but Col- ombia did experience a later, modest crisis of its own making. In 1978, when the coffee boom came to an end, Colombia had no crisis because debt was low. In fact, very sensibly, Colombia had reduced debt and had built up reserves during the boom. In 1979 Colombia started a public investment boom to counteract any recession the fall in coffee prices might cause. This boom involved foreign 70 Macroeconomic Policy and Growth borrowing. The move could be regarded as a sensible countercyclical policy, and for a while the country could afford it. But in 1981 coffee prices collapsed, and gradually Colombia's Keynesian strategy got out of hand, especially in 1983, when a serious recession emerged. Colombia continued its policy of stimulating the domestic economy by fiscal and monetary expansion into 1984, with more and more public sector and current account deficits, supporting the policy by a fall in reserves and by heavy borrowing. In 1984, however, Colombia undertook an orthodox adjustment program, involving fiscal discipline and substantial depreciation, and this program has been so successful that by 1986-88 the current account was roughly in balance. In retrospect, it appears that Colombian policymakers failed to depreciate the exchange rate sufficiently from about 1981, bearing in mind the real apprecia- tion during the boom and the need to stimulate demand when the boom came to an end. In due course, depreciation would have fostered exports and provided some of the additional demand for home-produced goods that the fiscal and monetary expansion had provided instead. Still, as balance of payments prob- lems developed in 1984, the government took quick, appropriate, and quite drastic action. That is an example to other countries: mistakes are made even in well-managed countries, but the art is to recognize them early and act with sufficient firmness. Turkey. Turkey had its spending boom, its debt crisis, and its recession relatively early. The crisis had nothing to do with the oil price rise but rather was the result of excessive borrowing earlier and other domestic problems. In 1980 Turkey adopted a major (classic) structural adjustment program that was badly needed in its very distorted economy. The World Bank-its advice and its money-was much involved. Details need not be given here other than to note that a big devaluation took place. That year was still a year of low growth and of exceptionally high inflation (more than 100 percent), a by-product of price and exchange rate adjustments. But from 1981 on, Turkey became a big success story, with a rapid growth of exports and average per capita growth from 1981 to 1987 of 3.2 percent. The devaluation clearly played a major role here. This probably is the main lesson of this case, although there were also some favorable market developments, and some degree of trade and financial sector liberalization must have helped as well. But there has not been a very significant overall fiscal adjustment. There was capital inflow from official sources, offsetting the decline in private capital inflow, so that Turkey could continue to run substantial current account deficits even though it had faced a debt crisis. This was an important feature of the Turkish case. Between 1981 and 1987, new funds were more or less suffi- cient to pay for the growing interest bill so that no outward resource transfer was needed. The result was that the ratio of debt to GNP rose from 34 percent in 1980 to 62 percent in 1987. But with the great growth in exports, the ratio Corden 71 relative to exports fell, and this must have helped in maintaining Turkey's perceived creditworthiness. It is still an open question how beneficial for growth of national income this official capital inflow will turn out to have been. It depends on how the funds have been spent. But insofar as the inflow contributed to the improvements in structural and exchange rate policies that led to the export boom (through conditionality and through avoiding crises), its effects may have been very positive. The policy improvements would not have happened without the 1978-80 crisis, so that in this instance at least the response to the crisis was clearly good for the level, and probably the growth rate, of real income.7 This has not generally been true in Turkey or in other countries. More often, balance of payments crises have led to the imposition or tightening of import restrictions that are not removed when the balance of payments problem disappears. Thailand. Thailand was as seriously affected by the second oil shock as by the first. The second shock cost Thailand about 4 percent of GDP. The nation borrowed its way through the immediate period and made gradual adjustments. Indeed, gradualism is a characteristic approach of Thailand's policymakers. As a result, the growth rate stayed high. But the adjustment policy bore fruit. By 1986 the current account deficit had disappeared, and the fiscal situation was thoroughly under control. Of course, the 1986 oil price decline helped. The slow adjustment involving large current account deficits for some years was possible because Thailand had maintained creditworthiness (so that capital inflow from private creditors never ceased) and, more important, because the nation had obtained strong support from official lenders. There were two rea- sons for this favorable view held by foreign lenders, private and official. First, Thailand started off with a relatively low debt-GDP ratio. Second, after a short deterioration, inflation was low, sound policies were being embarked on, and an export boom appeared to be under way. The real exchange rate had been mildly appreciating, mainly because the baht moved with the dollar, but in 1984 there was a 14 percent depreciation relative to the dollar to offset this, and from 1985 the real rate depreciated further, moving with the dollar. Thailand thus suffered a considerable external shock, and yet it was able to get over it without a crisis comparable to that of so many other countries, and without a severe recession. Initial conditions-a relatively low debt ratio and a tradition of fairly conservative management-helped, as did the underlying high growth rate. And even though the policy response was rather slow, it was credible on the basis of previous experience. Arguably, it was a stabilizing policy, which might be favorable for long-term growth. On the other hand, Thailand may have run undue risks by delaying full adjustment for some time but was rescued by the 1986 oil price decline. 7. The authors of the forthcoming country study on Turkey for the World Bank project, Ziya Onis and James Riedel, suggest in their draft that this was indeed the main effect of the crisis on long-term growth. 72 Macroeconomic Policy and Growth Indonesia. Indonesia suffered two shocks. One was in 1982-83 and was caused by the world recession, which brought about an oil price fall and declines in the prices of other export commodities. The other was in 1986 and was caused by the further, massive, oil price fall. The responses to the two shocks must be analyzed together. The two terms of trade shocks cost Indonesia about 10 percent of GNP. The responses were orthodox, impressive in their speed and magnitude, and bore fruit in a reduced current account deficit, although Indonesia is still relying heavily on new funds from official lenders, including the World Bank. There was severe fiscal retrenchment. From 1980 to 1987 the per capita growth rate was modest (1.5 percent) because of the two shocks and possibly because of the policy responses, although the long-term (1965-87) per capita growth rate of 4.5 percent has been notably high. Compared with Thailand's, Indonesia's total shock was bigger and its response quicker. Like Thailand, Indonesia has essentially (though no longer formally) tied its currency to the dollar, but it has been much more ready to devalue substantially, and it did so in response to each of the two shocks. These devaluations had significant effects, especially because of the tight monetary policy. From 1986 on Indonesia has taken major structural adjustment measures-trade liberalization, deregulation, and so on-and 1987 saw the beginning of a boom in manufactured exports. Decisive orthodox macroeconomic and structural policies have been success- ful in this case-with big fiscal retrenchment and big devaluations following quickly after a shock. Since 1986, trade and other liberalizations have contrib- uted to a nonoil-export boom. One could argue that Indonesia, like Turkey, derived a favorable by--product from the crisis (the 1986 crisis, this time), in that the crisis induced long-overdue structural adjustments, notably trade liberaliza- tion. This is likely to have beneficial long-term growth effects. The debt-GNP ratio rose from 28 percent in 1980 to 69 percent in 1988, but Indonesia avoided a debt crisis because of strong support from official lenders, which also ensured Indonesia's ability to continue borrowing on the private capital market. The Four Countries: An Overview Colombia and Thailand were slow to adjust to their adverse shocks. They could afford to adjust slowly because their debt ratios were relatively low. Thailand was able to borrow readily because its relatively sound macro- economic policy record indicated that the government was in control of the country's policies. To a lesser extent, but for the same reasons, Colombia was also able to borrow, but, in addition, it was able to draw on ample reserves built up during the boom. Colombia's ability to borrow from the private market was damaged by the spillover effects of the debt problems of other Latin American countries. Yet Colombia was unique in Latin America in never having called for a Corden 73 rescheduling of its debts, and one would expect a "rational" private market to have taken more account of this. Colombia did encounter a crisis, whereas Thailand avoided one because of its adjustment policy and with help from the 1986 oil price decline. In neither country did the situation get out of hand, though both might have acted sooner. Both countries effected real devaluations. Colombia seems to have made rather limited structural adjustments on the micro side and is still quite protectionist. Thailand has also made adjustments, typically in a gradual way, but in any case it was a much more open economy. The current success of Thailand seems to make a case for gradualism in adjustment policies, but such gradualism may be appropriate only in circum- stances such as Thailand's-when there is no crisis situation, when distortions do exist but are not as major as in so many other countries, and when there is a tradition of firm conservative financial management. One should not advocate such gradual macroeconomic adjustment for some other countries (particularly Argentina and Brazil) because their authorities lack credibility. That is, people would not believe that the proclaimed path of adjustment would be followed consistently. One should also distinguish gradualism in a planned adjustment program from slowness in making a decision to adjust adequately over a period. Provided there is no credibility problem and new borrowing remains possible, a case for gradualism can be made. Turkey suffered a major debt crisis, responded with drastic measures of an orthodox kind, and achieved favorable results for growth. But, in comparison with heavily indebted Latin American countries, Turkey had more foreign funds available, and this undoubtedly eased the macroeconomic adjustment process. Thus Turkey has not really solved its fiscal problem, and it has now become a high-inflation country, a matter discussed below. The case of Turkey suggests a possible link between structural adjustment policies, foreign borrowing (especially concessional borrowing), short-term growth, and long-term growth. Structural adjustment policies bring in more foreign funding, concessional or otherwise, and this has a beneficial effect on short-term (Keynesian) growth. It depends on the extent to which these funds are used for investment and the efficiency of the investment, whether this raises long-term growth. In addition, for well-known reasons, the initiation of signifi- cant structural adjustment policies (including policies leading to real devalua- tion) is likely to have a directly favorable effect on long-term growth. Indonesia has also made important structural adjustments and has also bene- fited from continued availability of foreign funds, mainly concessional. Like Turkey, Indonesia has acted quickly and drastically, but Indonesia has been far more effective in its macroeconomic stabilization. Although Indonesia should have made its microeconomic structural adjustments sooner, one can hardly fault its response to the crisis, because the crisis resulted not from an earlier borrowing boom but principally from the unexpected 1986 oil price decline. It is too early to observe the effects on medium- or long-term growth in Indonesia. 74 Macroeconomic Policy and Growth III. INFLATION AND GROWTH The Effects of Inflation on Growth What are the effects of inflation on growth? Inflation, of course, is not a macroeconomic policy but rather the outcome of policies, or possibly of the failures of policies. This subject is large, but it cannot be avoided if we want to assess the relationships between growth and macroeconomic policies. Here, to start, I will focus on the role of inflation as a tax. It is probably the key aspect for the countries that have gone or are still going through significant inflationary episodes, such as Brazil, Mexico, and Turkey. Inflation results primarily from monetization of fiscal deficits. It is a tax that has distorting effects that not only lower the level of real income but also lower the rate of growth by reducing the productivity of investment. Inflation's uncer- tainty effects may also discourage investment. But the effects of the inflation tax on the rate of growth can only be assessed in relation to the alternatives. If government investment expenditure were reduced to avoid a deficit that would otherwise have to be monetized, the rate of growth would (or might) also be affected adversely. If there were a switch from money financing to domestic debt financing, domestic private investment would be crowded out. "Ordinary" taxes, such as taxes on trade, might replace the inflation tax, but these would also cause distortions that would lower the efficiency of investment and hence the rate of growth. Increased taxation of investment goods or of corporations would reduce private investment directly. If the alternative were more foreign borrowing, there would be future costs. Only if the alternative consisted of reductions in government consumption or private consumption-the latter attained through reduced transfers or higher ordinary taxes that were not very distorting-would the growth effect of choosing the inflation tax be clearly adverse. Despite this long list of ways in which various possible alternatives to inflation can lower the growth rate, an inflation tax is surely never part of an optimal tax structure that is carefully thought out and that takes into account the need to minimize distortions and attain desirable redistributive and investment effects. It is usually the tax of last resort.A3 In the high-inflation countries it is the outcome not of policy but of policy failure. The costs of a given rate of inflation can be reduced by indexation of various kinds, but this is likely to raise the rate of tax required to attain a given revenue. It is hard to believe that more efficient ordinary taxes are not available. Furthermore, the costs of inflation rise over time, and this is usually not taken into account when the inflationary process is-by default-embarked on. First, 8. "The method is condemned, but its efficacy, up to a point, must be admitted. A government can live by this means when it can live by no other. It is the form of taxation which the public finds hardest to evade and even the weakest government can enforce, when it can enforce nothing else" (Keynes 1923, p. 41). Corden 75 as expectations and financial habits adjust to higher inflation, the demand for real balances falls, and so the rate of inflation rises for a given money-financed fiscal deficit or rate of growth of the money supply. Second, as inflation acceler- ates, eventually stabilization measures-often drastic-have to be taken, and then costs cannot be avoided. Chile has been through that experience; Argentina and Brazil have tried it unsuccessfully; and Mexico is going through it now.9 The Trade-off between Inflation and the Current Account The inflation tax is usually the last resort for bringing about the necessary reduction in real expenditure (absorption) in a balance of payments crisis. Given that ordinary taxes cannot be increased nor government expenditures reduced sufficiently, and that there are limits to domestic financing of a deficit, a govern- ment is faced with the choice between emergency foreign borrowing-hence allowing a continued current account deficit-and monetization of a fiscal defi- cit, which leads to inflation. There is thus a trade-off between inflation and the current account. I call this the "trade-off model." It is very helpful for under- standing what happened in several (mostly Latin American) countries and for comprehending the choices policymakers have had to make. For a given (inflation-adjusted) fiscal balance, a country can improve its noninterest current account (the resource transfer) by increasing inflation. This tended to happen in the debt crisis for a number of countries-Argentina, Brazil, Mexico, and possi- bly Turkey. The debt crisis compelled a switch from financing budget deficits through foreign borrowing to money financing. In addition, the budget deficits themselves had increased because of increased interest payments and poorer terms of trade. In both Brazil and Mexico there was also a shift to domestic debt financing, which (through increasing interest payments) tended to worsen the deficits further. The argument that inflation improves the current account hinges on two assumptions: first, that it leads to higher private savings than otherwise- savings designed to restore, at least partially, real balances-and, second, that the nominal exchange rate is flexible or frequently adjusted. The argument seems counterintuitive when one thinks in terms of a fixed exchange rate regime, because in that case inflation brings about real appreciation, which tends to worsen the current account, even though the inflation tax would still reduce absorption and so have the opposite effect. Furthermore, there are plenty of circumstances, including the immediate onset of the 1981-82 crises, when both current account deficits and inflation rates increased. But this is not incompat- ible with the model, because the model only says that for any given inflation- adjusted fiscal deficit, there is a negative relationship between inflation and the noninterest current account deficit. The 1981-82 crises were actually associated with increased budget deficits. 9. There is an extensive literature on these stabilization efforts and the search for least-cost methods of descending from a high-inflation path. See, for example, Bruno and others (198 8). 76 Macroeconomic Policy and Growth The High-Inflation Countries With regard to inflation, the eighteen countries in our group (this time includ- ing Argentina) can be classified as follows. Eleven were low-inflation countries, with rates of less than 20 percent on average over the period 1980-88 (actually, eight of these had inflation rates of less than 10 percent). Of the seven others, Argentina and Brazil were the only consistently high-inflation countries. Costa Rica, Mexico, and Turkey were low-inflation countries until 1973 and some- times later. Since then they have gone through high-inflation episodes. Brazil is an important outlier. It was a high-growth and high-inflation country over a long period, seeming to challenge the presumption that high inflation is bad for growth. During 1965-73, a period that embraced the "Brazilian mira- cle," the average inflation rate was 28 percent, and the per capita growth rate was 7.2 percent, one of the highest in the developing world. The inflation rate was also high by the standards of those years. But during this period the Bra- zilian inflation rate steadily fell from 61 percent to 13 percent, so that the average figure is really rather misleading. In the actual "miracle" period, 1968- 73, the inflation rate averaged about 20 percent and the per capita growth rate about 9 percent. In the rnext pe.riod, 1973-80, the average inflation rate rose to 45 percent, and the per capita growth rate was still exceptionally high at about 4.5 percent. Recently, however, Brazil's experience seems to support the presumption that inflation is bad for growth, or at least tends to be associated with low growth. Brazil's per capita growth rate was still positive up to 1987 (unlike that of Argentina, Costa Rica, Cote cl'Ivoire, Mexico, and Nigeria). But from 1988 Brazil's growth rate has fallen as the inflation rate has risen. Per capita growth became negative in 1988, as the inflation rate rapidly accelerated, reaching hyperinflationary levels (of about 1,000 percent annually) by 1989. The special-and much discussed-feature of Brazil is that it has been an indexed economy. Indexation has made Brazilians more tolerant of inflation (therefore helping to explain the continuation and acceleration of inflation), and it has reduced the costs of a given inflation. Above all, it has avoided prolonged overvaluations of the real exchange rate. This is surely a reason why high inflation has not prevented high growth until recently. Three other Latin American countries are also interesting with regard to inflation. Colombia had moderately high inflation over long periods, and the effects were not obviously adverse. Colombia's average inflation rate from 1980 to 1988 (it was also fairly stable) was 23 percent. This country was among the first to practice the crawling peg exchange rate system for a prolonged period, and hence the same comment can be made for Colombia as for Brazil. The system has clearly avoided some of the biggest potential costs of inflation. Mexico switched from being a low- to a high-inflation country with the 1982 crisis. Its two bursts of 100 percent annual inflation (1982-83 and 1986) can be readily explained in terms of the trade-off model. It is now going through a Corden 77 stabilization program that appears successful so far. This program has a strong orthodox element (monetary and fiscal restraint). In addition, it involves a wage restraint pact with the trade unions and the regular depreciation of the nominal exchange rate on the basis of a predetermined scale, which is designed to reduce inflationary expectations. Costa Rica was a very-low-inflation country until 1982 but had a short burst of really high inflation as part of its 1982 crisis (this is discussed below). This leaves Turkey, which I noted earlier as a case study of successful adjust- ment. When one takes the increase in inflation into account, especially since 1988, Turkey appears less successful. It became a high- or medium-inflation country with the 1978 crisis. Its inflation is obviously explained by the moneti- zation of fiscal deficits together with a steady reduction in the demand for real balances, an adjustment of expectations to the shift to relatively high inflation since 1978. In earlier days such deficits would have led to balance of payments crises; in fact, they did in 1958, 1968, and 1978. The fixed exchange rate regime ruled out continuous inflation. But, along the lines of the trade-off model, foreign borrowing problems are now avoided by means of the inflation tax. It is noteworthy that a period of quite high inflation has also been associ- ated with a period of high growth rates (as in the case of Brazil earlier), but at least until 1987 the higher inflation rate was fairly stable. The danger is that eventually the need to deal with the inflation problem could put an end to high growth rates, at least temporarily. The Low-Inflation Countries Three observations can be made about inflation in the relatively low-inflation countries (inflation below 20 percent). First, all these countries have had short bursts of high inflation, usually in the 1980-82 period. The bursts were usually caused by an external shock, such as the oil price rise for oil-importing coun- tries, or by a devaluation necessitated by the balance of payments consequences of poorer terms of trade, higher interest rates, or the decline of foreign financing of spending booms. But the inflationary bursts did not last. Even Thailand, the low-inflation country par excellence, illustrates the sus- ceptibility of nations to short-term bursts of high inflation. From 1975 to 1979 its average inflation rate was 7 percent. In 1980 the inflation rate jumped to 20 percent. By 1982 it had fallen back sharply to 5 percent. A more extreme example is Costa Rica. Before 1973 its inflation rate had never been above 5 percent, and from 1975 to 1980 it averaged 12 percent. But, for the kinds of reasons listed above, its inflation rate jumped to 90 percent in 1982. Yet by 1984 it had fallen to 12 percent again. Among the many similar examples that could be given, Nigeria had a big inflation bubble in 1984, at 40 percent, and the next year its inflation rate was down to 5.5 percent. One can always explain these high-inflation episodes. Sometimes they repre- sent no more than the short-term effects of a big devaluation or, more generally, 78 Macroeconomic Policy and Growth of a necessary structural adjustment episode. In Nigeria, a severe decline in imports was necessitated by a balance of payments crisis and implemented through tightened import restrictions. In India droughts have produced inflation bubbles. The episodes were brief because governments had a commitment to low inflation-a commitment derived from historical experience-so that they did not allow these episodes to change their basic low-inflation strategies. The brief episodes did not stimulate inflationary expectations (and hence made it possible to bring inflation down again without great or any cost) because of the governments' low-inflation reputations. A record of a long period with fixed exchange rates, or rates infrequently altered, is helpful here. This must have been a factor in the success of the Costa Rican stabilization and also, currently, in the Mexican success story. A second reason for inflation in the relatively low-inflation countries is that during boom periods some inflation was part of the process that brought about the necessary real appreciation associated with spending booms discussed ear- lier, given that there was a fixed nominal exchange rate or an inadequately adjusted exchange rate. This applies particularly to Colombia, C6te d'Ivoire, Indonesia, and Nigeria, but also to other nations in which there were private or public spending booms. Provicled that inflationary expectations are not created during such a process, and that the government's basic anti-inflation commit- ment (possibly signaled by the exchange rate policy) remains, such a temporary inflation and real appreciation should not lead to continuous inflation. The crucial issue is whether the government can keep control of the process. A third factor in the inflationary episodes of the relatively low-inflation coun- tries is that costs of inflation depend, above all, on the exchange rate regime. The nominal exchange rate may be fixed, or it may not be depreciated suffi- ciently to keep pace with the inflation differential. Given an export boom, or a spending boom that is foreign-financed, a real appreciation might be appropri- ate as part of the process described above. But a common situation has been that the combination of inflation and a fixed exchange rate has led to intensified import restrictions and hence to familiar distortions that have an adverse effect on growth-often a severely adverse effect. Examples of such episodes can be found in many of the seventeen study countries, but an outstanding case is in Nigeria. There, the trade-weighted nominal rate was pretty well fixed from 1981 to 1984, but because of a rate of annual inflation averaging 23 percent over this period, the real rate doubled (appreciated). In addition, the terms of trade deteriorated, mainly because of the falls in oil and cocoa prices, so that a real depreciation was really required. Extremely tight import restrictions were applied, and exports became highly uncompetitive. The unavailability of essential imports for local production as well as the decline in exports must have been a major factor in the negative growth rates of those years. According to the available figures, per capita growth averaged -8 percent during those four bad years. One must blame the combination of high inflation and the fixed exchange rate for these adverse Corden 79 effects. In this case the fixed exchange rate regime did not succeed in anchoring the price level or the money growth rate. An Overall View In cross-country comparisons, can one find a direct correlation between infla- tion rates and growth rates? Clearly other things are not equal between coun- tries, but some econometric results suggest that inflation tends to go with low growth, especially when post-1980 figures are used.10 It is useful to compare four countries that are clearly either low-inflation countries (Thailand and India) or high-inflation ones (Argentina and Brazil). Thailand is a low-inflation country with high growth; Argentina is the opposite. That is, both in the per capita growth tables and the inflation tables they are at opposite ends, so they confirm the presumption that inflation is bad for growth. The matter is not quite so clear for the two biggest economies in our group, India and Brazil, essentially because their relative growth rate positions have changed. From 1965 to 1980, India was a low-growth country (1.5 percent per capita), but from 1980 to 1988 India jumped to the relatively high rate of 3.7 percent. In the latter period, 1980-88, Brazil moved in the opposite direction, from 5.7 percent to 0.6 percent. Thus a comparison based on the recent period again confirms the presumption. A comparison over time for Brazil also con- firms it. As the rate of inflation has climbed, the rate of growth has fallen. Yet one cannot ignore Brazil's episodes of high growth and high inflation merntioned earlier. It may be generally true-even though some cases, such as Brazil, India, and Turkey during certain periods, as well as various African countries, give a contrary result-that low-inflation countries tend also to be high-growth coun- tries, and vice versa. But this does not necessarily mean that high or medium inflation is the principal cause of relatively low growth. Obviously, investment ratios, degrees of openness of economies, and many other factors are relevant. To some extent inflation and low growth may have a common cause. It seems plausible that poor economic management-which may be the result of an inability of a government to resist pressure groups and generally to ensure adequate macroeconomic controls (clearly the case in Argentina)-leads to a variety of policies that produce low growth. And high inflation, which is always unplanned and is a last-resort tax imposed by default, is clearly a symptom or consequence of poor management. Some Lessons An obvious lesson taught by Argentina, Brazil, Chile, and several other Latin American countries is that high inflation, once it keeps going for some years, is 10. A satisfactory statistical relation for 1980-88 between the per capita growth rate and the rate of inflation cannot be found for the seventeen countries. This is essentially because of the inclusion of three African countries with low inflation and low growth, as well as Turkey, which had high inflation and high growth (see table 1). 80 Macroeconomic Policy and Growth hard to reduce without severe cost, both economic and political. It is best to avoid getting started on the road. It is an inefficient tax. It yields immediate benefits to governments that have difficulty cutting expenditures or raising other taxes, but the costs come later, and, as the demand for money falls, the cost- benefit ratio deteriorates. Once on the move, high inflation is difficult to control unless counteraction is taken very quickly, so that there is no more than an inflationary bubble, as in several of the countries in our group. The dislike of inflation even at modest levels that is customary in many of the Asian countries we have been considering can be very healthy. As Brazil and Colombia show, it is possible to live with inflation over long periods-that is, to adapt to inflation and have fairly high growth. A crucial element in such adaptation is exchange rate adjustment to avoid real apprecia- tion. The danger is always that the rate of inflation will accelerate. But it is an interesting question to what extent the long period of high inflation with high growth in Brazil can be blamed for the current hyperinflation. Other countries-even Thailand-show that short bursts of inflation are sometimes difficult to avoid. But they can be made short. Their significance must not be misunderstood. Sornietimes they are the consequences of a necessary devaluation or a by-product of an adjustment to capital inflow, an export boom, or a structural adjustment of some kind. In all cases it is important that infla- tionary expectations do not increase and that a fundamental commitment of the government to low inflation is clearly established. It is also important that the exchange rate is appropriately adjusted so that increasing import restrictions (the Nigerian case) are avoided. IV. EXCHANGE RATE POLICY The Exchange Rate and Growth The central issue is whether keeping a nominal exchange rate fixed when there is domestic inflation or an unfavorable exogenous shock can possibly be favor- able for growth. There are obvious reasons, to which I shall return below, why its effects would be unfavorable. But might there be a favorable aspect? The simple answer is that devaluation or readiness to depreciate may be inflationary, and this, for reasons discussed earlier, could be adverse for growth. The potential problem is not the once-and-for-all rise in the domestic prices of tradables (which may take some time to come about) but the possibility that devaluation would set off an inflationary process. This is only possible if it induces continuous monetary expansion. Devaluation, or the possibility of depreciation, could lead to continuous mon- etary expansion and hence be inflationary in two ways. First, a price-wage spiral may be set off by devaluation, and to avoid deflation caused by a continuous decline in the real money supply the nominal money supply may then be contin- uously increased. Second, a fixed (or reluctantly adjusted) exchange rate may Corden 81 have been an anchor for monetary policy-that is, a restraint on it. The exchange rate commitment may have discouraged inflationary monetary policy and monetization of fiscal deficits. Once the possibility of depreciation is opened up, the constraint on inflationary tendencies disappears. The empirical question here is whether long periods of low inflation in some countries can be explained by the constraint imposed by their fixed rate regimes, or whether, alternatively, the ability to maintain fixed rates was the result of low-inflation monetary policies, the latter explained by more basic anti-inflation attitudes, possibly historically based. This question applies to many low- inflation countries in our group, at least up to 1982, mostly in Asia and Africa, but also Costa Rica and Mexico up to about 1973. Can one argue that increased inflation in many of the countries since 1982 was caused by the increased flexibility of exchange rates, which loosened the constraint? The question is only posed here, because there is no space to attempt an adequate answer. It raises issues of political economy because it concerns the motivation for inflationary, noninflationary, and monetary policies. I have con- cluded provisionally that in several countries (for example, Mexico, Nigeria, and Turkey) an apparent commitment to a fixed exchange rate did not prevent accelerating inflation and that in the five Asian countries (as well as Kenya and Morocco) a switch to more flexible exchange rate regimes left relatively low- inflation policies intact or nearly intact. Hence, in these cases, the exchange rate regime was not the driving force in sustaining low or moderate inflation. Rather, low- or moderate-inflation policies were a direct and more fundamental objec- tive, and the exchange rate followed rather than led monetary policy. Let us now ignore the possible effects of exchange rate flexibility in inducing expansionary monetary policy just discussed. The central concern for exchange rate policy thus becomes its effects on the real exchange rate. There is strong evidence that in the low- or moderate-inflation countries the nominal exchange rate affects the real exchange rate in the same direction and, in the medium run, to a similar extent. The issue is to "get the real exchange rate right," and in particular to avoid or eliminate overvaluation. This is a familiar World Bank and International Monetary Fund (IMF) theme. We have seen that real appreciation may be a necessary element in adjustment to a spending boom financed by an export boom or by deliberate foreign bor- rowing or aid. Exchange rate policy that leads to real appreciation is then just an incidental aspect of the total process, part of the transfer mechanism. In that case, real appreciation cannot be described as overvaluation. I have already discussed the growth implications of such a spending boom. But there remain cases in which there is no export boom or autonomous capital inflow. The problem may be a real appreciation that has resulted from a fixed or reluctantly depreciated nominal rate combined with domestic inflation. Or it may be an overvaluation caused by an adverse external shock that requires a real deprecia- tion that does not take place because of rigidity of the nominal rate. 82 Macroeconomic Policy and Growth There may then be various effects, all potentially adverse for growth. Epi- sodes of real exchange rate overvaluation leading to these effects can be found in most of the countries at some time. It is the desire to avoid these effects that explains World Bank and IMF concern with real exchange rate overvaluation. First, overvaluation causes import restrictions to be imposed or intensified, as a substitute for devaluation, to "switch" demand away from imports. This would have adverse effects on growth through the distortions created, rent- seeking, reduced availability of imported capital goods, and so on, as in the Nigerian example cited above. Second, crisis borrowing takes place at interest rates far in excess of the marginal productivity of domestic investment, or the borrowed funds finance consumption, with fiscal expansion possibly accompanying the real apprecia- tion to maintain demand for domestic resources. This affects growth adversely in the way discussed earlier with regard to the borrowing of the 1970s. Third, the expectation of devaluation induces capital flight. Fourth, as continuous inflation brings about continuous real appreciation and hence tighter import restrictions, devaluation does eventually take place, possibly in a crisis situation. The real exchange rate thus may move in the right direction over time, but it may be very unstable, gradually appreciating, suddenly depreciating in a crisis, then appreciating gradually again, and so on. Such instability is likely to have adverse effects on growth. Country Experiences Every conceivable exchange rate regime can be found among our seventeen countries during the period under consideration. Beiow I look in some detail at Indonesia and Mexico. (The example of Nigeria described earlier should be recalled as well. In that case a fixed nominal rate failed to prevent inflation but led, rather, to increased import restrictions.) Many countries switched their exchange rate regime during the period 1975- 84, having had a fixed rate of some kind until the switch and then having moved to a flexible rate, possibly a crawling peg. Aside from the two franc-zone coun- tries, none have firm exchange rate commitments now, though three- Indonesia, Thailand, and Kenya-have (more or less) fixed-rate regimes with occasional devaluations. The shift to a flexible exchange rate regime of some kind in many Asian and African countries that used to have fixed rates can be explained in part by the shocks of the 1980-82 period and, most importantly, by the reduced ability in all countries to restrict international capital movements even when that is desired (and hence the reduced ability to sustain an exchange rate for any length of time against speculative attacks). Indonesia kept its nominal exchange rate fixed to the dollar from 1971 to 1978, a period in which there was a substantial real appreciation resulting from inflation above world levels but justified by the higher oil income. Since 1978 Indonesia has devalued three times, in each case by about 30 percent. The 1978 Corden 83 devaluation was quite a surprise, because there was no balance of payments problem but rather a concern with the declining profitability of nonoil-export industries. This was a case of exchange rate protection. The 1983 and 1986 devaluations took place very promptly in response to balance of payments prob- lems. The 1986 devaluation was a crucial ingredient in the adjustment program. Indonesia's monetary policy has been fairly conservative, with an average inflation rate in the period 1982-88 of 8.5 percent. The devaluations did have some temporary effects in raising inflation rates at the time, but one cannot really say that the readiness to devalue made monetary policies looser, as is implied by the argument advanced at the beginning of this section. It should be added that capital mobility is high in Indonesia. Thus once a devaluation is expected and hence leads to speculative outflows, it cannot be long delayed. Remaining controls on capital movements have been removed recently, so that in the future there will probably have to be more frequent, and less large, exchange rate adjustments. Mexico's exchange rate policy reflects the continuous tension between two objectives: trying to keep down the rate of inflation through fixing the rate, and restoring competitiveness (as well as stopping or reversing capital flight) by devaluing. Up to 1972 Mexico was characterized by low inflation and fixed exchange rates, but since 1973 a fixed rate has never really stopped high infla- tion. Nevertheless, the memory of the long and successful period of a fixed exchange rate and low-inflation regime, together with the obvious potential benefits of having a stable rate to the U.S. dollar, has led Mexican policymakers to attempt again and again to fix the exchange rate at a new level, even when they regularly fail to attain the low-inflation target. Hence, Mexico has experienced several periods of continuous real apprecia- tion, followed by sharp depreciation (1976, 1982, and 1986). After the big depreciation of 1986, the real rate was kept down until 1988, when it started rising again as the result of the fixing of the exchange rate as part of a stabiliza- tion plan. Currently, the exchange rate is effectively fixed on a predetermined crawl. The periods of real appreciation involve obvious problems-capital flight (recently avoided through very high domestic interest rates) and loss of compet- itiveness for tradable goods producers. One would expect the uncertainties cre- ated by such fluctuations in the real exchange rate to have adverse effects on investment and growth. The contrast with Brazil is striking. Since 1973, Mex- ico's real exchange rate has been much more unstable than Brazil's. The lesson is not necessarily that the nominal exchange rate should have been depreciated more, or more often. After all, experience has shown that in Mexico the real depreciations are eroded by high inflation after a while, and it is quite possible that greater depreciations would simply have led to more inflation. The lesson-which in fact has been learned-is rather that policy packages are required. Reducing inflation requires explicit monetary and fiscal policy deci- sions supported (at least in the Mexican context) by a wage compact. The exchange rate should be only one part of the package. 84 Macroeconomic Policy and Growth Lessons At the risk of oversimplifying, I suggest that this study yields three simple lessons with regard to exchange rate policy. 1. A country has to make a commitment to a noninflationary monetary pol- icy. This does not require a nominal exchange rate commitment, and usually such an exchange rate commitment is not enough to achieve the objective, although it may be helpful. A commitment to fiscal restraint is usually more important. 2. Real exchange rate misalignment and variability should be avoided through appropriate nonlinal exchange rate adjustment, preferably by frequent small changes rather than by large discrete changes. This is an oft-repeated message of the World Bank and the IMF. But in particular cases or episodes, the anti-inflation objective may have to take priority. 3. Devaluation without an appropriate policy of monetary restraint-and firm commitment to such a policy-is undesirable because it will ultimately be ineffective. In general, policy packages-involving monetary, fiscal, and exchange rate policies-are required. REFERENCES Aghevli, Bijan B., and Jorge Marquez-Ruarte. 1985. A Case of Successful Adjustment: Korea's Experience during 1980-84. Washington, D.C.: International Monetary Fund. Bruno, Michael, and others, eds. 1988. Inflation Stabilization. Cambridge, Mass.: MIT Press. Collins, Susan M., and Won-Am Park. 1989. "External Debt and Macroeconomic Performance in South Korea." In Jeffrey D. Sachs and Susan M. Collins, eds., Devel- oping Country Debt and Economic Performance, Vol. 3. Chicago: University of Chicago Press. Cuddington, John. 1989. "Commodity Export Booms in Developing Countries." World Bank Research Observer 4, no. 2 (July): 143-65. Gelb, Alan, and associates. 1988. Oil Windfalls: Blessing or Curse? A World Bank Research Publication. New York: Oxford University Press. Keynes, John Maynard. 1923. A .Tract on Monetary Reform. London: Macmillan. Reynolds, Lloyd. 1985. Economic Growth in the Third World, 1850-1980. New Haven, Conn.: Yale University Press. Sachs, Jeffrey D., ed. 1990. Developing Country Debt and Economic Performance. Vol. 2. Chicago: University of Chicago Press. Sachs, Jeffrey D., and Susan M. Collins, eds. 1989. Developing Country Debt and Economic Performance. Vol. 3. Chicago: University of Chicago Press. PROCE E DING S OF THE W O RLD BANK ANN UAL CO NFE RE NC E ON DE VE LO PM ENT EC ON OM I CS 1 9 9 0 COMMENT ON "MACROECONOMIC POLICY AND GROWTH," BY CORDEN John Williamson Max Corden's paper presents the first results from the massive World Bank comparative study of macroeconomic policy and performance in seventeen countries over almost twenty years. As Corden emphasized, the paper does not have any startling new insights. But it will appeal to those who believe that it is more important to be right than to be novel. The paper interweaves three themes. The first is the history of the seventeen countries, which provides the background. The second is an exposition of stan- dard macroeconomic theory for open, developing economies. Just about all aspects of this theory are expounded at one point or another. The third theme is the lessons that can be drawn from the experiences of these countries. It is these lessons on which I concentrate in my remarks. I happen to agree with all nine of the lessons that Corden draws. My only criticism is that he might have been less cautious in phrasing them. Although they may not be startlingly original, these lessons are important. If we do not bring them into sharp relief, other countries may repeat the mistakes that have been made in such profusion over the last twenty years. Corden's first lesson is that countercyclical policy is feasible and advisable. To paraphrase the poinIt, countries should endeavor to restrict their spending to the level of their permanent income. Colombia is a great example of a Latin Ameri- can country that attempted to do this in the 1970s, and I strongly believe it is not an accident that Colombia had by far the best growth performance of the region in the 1980s. It is true that in the end Colombia overdid it by attempting to sustain demand at a level that did not allow for the "permanent" deterioration in the external environment in the 1980s. Because of that, the nation found itself with unsustainable twin deficits. But Colombia corrected its mistake relatively quickly. It provides a model for emulation. Mexico represents the opposite extreme; not only did it fail to restrain spending but it used its transitory income from high oil prices as collateral to increase spending even more than income. Corden's second lesson is even more elementary, but it has been violated often nevertheless. Investment projects should be carefully vetted for profitability, remembering that real depreciation will occur when the lending boom ends and John Williamson is a senior fellow at the Institute for International Economics in Washington, D.C. (© 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 8i 86 Comment the debt has to be serviced, so that the rate of return in nontradables will need to be higher than that in tradables to justify borrowing. The third lesson is this: beware of euphoria. Corden makes the interesting point that this lesson should be particularly addressed to the sort of governments of which most of us tend to approve-market-oriented governments managed by competent technocrats. A useful dictum that may help successful technocrats avoid overconfidence is that all positive shocks should be treated as though they were transitory and all negative shocks as though they were permanent. The fourth lesson is that countries should be extremely suspicious of inflation. On this point I would strongly back Corden's view as against Rudiger Dorn- busch's suggestion in his paper (this issue) that any rate of inflation below about 50 percent per year should be accepted with equanimity. I agree with Corden that inflation is never part of an optimal tax structure. In fact, the evidence is that it is one of the most regressive of all taxes. Add to that the conclusion of the accelerationist theory that there are no permanent gains in output from accept- ing a higher long-run rate of inflation and the consideration that the occasional corrective inflation can operate more efficiently when the contract structure is based on an assumption that prices will generally be stable, and I see a compel- ling case for accepting Alan Greenspan's criterion of seeking an annual rate of inflation of 2 percent or less. This is not to argue that a minimal rate of inflation must be the overriding priority of economic policy or that it necessarily must be achieved in the short run. But it is to urge that no government should feel relaxed about a rate of inflation that is "just 20 percent." Once that happens, inflation will not stay at 20 percent. How does one reconcile a respect for the dangers of inflation with Corden's fifth lesson, which says that it is possible to sustain high growth with high inflation for a long time, provided that the nominal exchange rate moves to offset the differential inflation? Once again, Colombia provides the outstanding example. Colombia has operated a crawling peg for twenty-three years and has regularly had an annual rate of inflation of around 20 to 30 percent. I nonethe- less wonder whether it might not have been sensible somewhere during those twenty-three years to accept a significant temporary sacrifice in output to elimi- nate inflation. Corden's sixth lesson is that corrective inflation-a one-time rise in the abso- lute price level undertaken to achieve a change in relative prices that is needed to adjust to a real shock-is admissible. As I noted above, I regard the possibility of using corrective inflation as sufficiently important to be a significant argument justifying the attempt to eliminate trend inflation. The seventh lesson is that macroeconomic policy needs a nominal anchor but that this need not be a fixed exchange rate. Indeed, Corden argues explicitly that the attempt to use a fixed nominal exchange rate as the nominal anchor has often proved ineffective, and he could have added Chile and Costa Rica in the early 1980s as further examples of this thesis. The alternative to a fixed Williamson 87 exchange rate is essentially demand management policy. I favor use of a target growth rate of nominal domestic demand as the alternative nominal anchor. The eighth lesson is that exchange rate policy should be used to avoid mis- alignments of the real exchange rate, subject to the same qualification that Dornbusch emphasized earlier. This qualification admits that a temporary freeze of the exchange rate may be a useful element in a stabilization package designed to bring rapid inflation under control. The conventional wisdom now seems to be that within six months of a stabilization, the exchange rate should start to crawl again if this is necessary to prevent an overvaluation from emerg- ing. If this is not done, and inflation has not come to a halt, the danger is that the markets will start to demand a big premium on the domestic nominal interest rate to compensate for the risk of a subsequent jump devaluation. One will get the cost of a crawling devaluation without the benefits in terms of keeping relative prices in line. Hence the freeze of the exchange rate should last for the minimal period needed to use the international structure of relative prices as a guide to the set of internal relative prices that will prevail after stabilization. Corden's ninth lesson is that devaluation needs to be used as a part of a policy package, along with fiscal, monetary, and-if helpful-income policies. The bad academic habit of arguing that devaluation is ineffective in isolation is irrelevant to the policy debate. In addition to the nine lessons Corden drew in his paper, I would suggest two others that seem clearly implied by his stories. The first is that the debt ratio should be maintained at a low level and international reserves maintained at a high level. The purpose is to provide a shock absorber to prevent the need for damaging lurches in policy. The other lesson is that when a country has the chance-which means when it has followed the preceding lesson-it should plan for gradual adjustment but initiate a comprehensive adjustment program promptly. Confusion over the desirability of gradualism has arisen when policymakers have failed to note the distinction between prompt initiation of adjustment, which is always desirable, and immediate achievement of adjustment, which can be enormously costly. Of course, there are circumstances in which adjustment should be achieved imme- diately. It makes no sense to think of a gradual reduction in hyperinflation. But it does make a great deal of sense to achieve balance of payments adjustment over several years, as Thailand did, rather than to try to turn the accounts around in a matter of months, as happened in Latin America in the early 1980s, with disastrous results. I conclude by suggesting topics on which I hope the authors of the study will endeavor to provide some guidance on the basis of the experiences of the sample countries they have investigated. One of these concerns the crucial question raised by Dornbusch: How does one get growth going again after stabilization? Note, incidentally, that this is not the first time economists have confronted this issue. Various phrases were used in the 1930s to indicate the conclusion that an 88 Comment easy monetary policy was not enough for the purpose: "you can't push on a string;" "you can take a horse to water but you can't make it drink." Similarly, the founders of development economics toyed with ideas of big pushes and balanced (or unbalanced) growth strategies to get the growth process going. But gradually interest in this set of issues was displaced by concern to increase savings, to provide adequate infrastructure, to ensure that the real exchange rate was sufficiently competitive to nurture the growth of nontraditional exports, and to ensure that investment passed the market test of profitability. The mes- sage was that if the supply side and the exchange rate are appropriate, then investment and growth will take care of themselves. Can the historical experi- ence of the seventeen countries confirm that this is wiser than immediately looking for drastic policy changes if growth does not return as quickly as one might hope? Is there any evidence that an overactivist policy could prevent reforms from taking root and actually do more harm than good? Or is there indeed some additional ingredient that can promote a resumption of growth? In particular, I wonder whether the historical evidence can shed some light on Dornbusch's own prescription of massive foreign loans as necessary for reinitiat- ing growth. It is not obvious to me that this is correct. For example, in the 1930s countries sought to reignite growth by competitive devaluation in order to secure a negative resource transfer. Regarding Mexico, Dornbusch is quite right to argue that it is a question of turning the corner, and that once confidence revives, the repatriation of flight capital will finance an investment boom. But is a big new foreign loan the right way to revive a lack of confidence that has been caused by a debt overhang? Perhaps the Brady Plan is more relevant, and the real aim of policy should be to reinforce the debt settlement with the commercial banks by doing something about public sector debt. More generally, can the historical experience be used to illuminate the crucial question of the relationship between policy reform and resource transfer? A decade ago, most of us took it for granted that more resource transfer was better. We are sadder now and tend to argue that resource transfer has often been used to perpetuate bad policies that actually worsen a country's long-term prospects. The conclusion we draw is that additional resource transfer should be provided only after policy reforms have been securely emplaced. How compel- ling is the historical evidence that we are not only sadder to take this view but also wiser? PRO CE ED ING S OF T HE W O RLD BANK ANN UAL CONFE RE NCE ON DE V EL O PM ENT E CON OM IC S 1 9 9 0 COMMENT ON "MACROECONOMIC POLICY AND GROWTH," BY CORDEN Susan M. Collins This paper addresses an extremely important topic: the role of macroeconomic policy in promoting growth. It draws on a wealth of experiences, which enables it to cover quite a lot of ground in specific countries. It also is refreshing to find the discussion broadened to include so many different countries. Too often, the focus is on just a few economies, typically the large Latin American ones. These are not representative of experiences of economies in Africa and Asia--or even of the smaller economies in the Western Hemisphere. Corden's paper is built on the skeleton of fiscal expansion that sets the stage for a crisis when lending is interrupted, requiring domestic adjustment. It then focuses on two key aspects of adjustment--inflation and'exchange rate policy- and draws a series of lessons. I agree with many of these lessons. (A major exception is Corden's conclusions on inflation. Here, I found his discussion misleading, and I will explain why below.) Although I agree with many of Corden's lessons, I would have drawn them with a somewhat different brush, and I would have shifted the emphasis. In particular, I do not believe that it is possible to omit the political economy of macroeconomic policymaking in a discussion of how policy can promote growth. Let me revisit the major lessons of the paper, then, from my own perspective. The role of public expenditure booms in the period before a crisis in country after country is really striking. These experiences clearly spotlight fiscal restraint as the cornerstone of sound macroeconomic policy. The paper could have brought this point out even more strongly. The need for adjustment in almost all of these countries was not created just by external shocks such as the reduced access to foreign capital after 1982. It was also a result of unsustainable domes- tic policies. Thus it should come as no surprise that the countries that did better were the ones that adjusted sooner and that had not allowed basic policies (especially the budget and the exchange rate) to get too far out of line in the first place. For example, both the Republic of Korea and Indonesia began to adjust macro policies before they were forced to-and before crisis hit. Corden is exactly right in stressing the need to push good policies during good times. This creates favorable initial conditions that ease the adjustment in bad Susan M. Collins is associate professor of economics at Harvard University. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 89 90 Comment times. In asking how these adjustment policies affect growth, Corden is also exactly right in pointing out that growth during the stabilization period should not be compared with growth during an unsustainable period of poor policy. In a few countries, one very visible element of the aftermath of the post-1982 lending crunch was runaway inflation. After looking at these episodes, Corden reaches the very strange conclusion that "inflation improves the current account." It is certainly not true, however, that countries with higher inflation- or larger increases in inflation-tend to have smaller deficits or larger surpluses. Nor is it true that an increase in inflation in individual countries leads to current account improvement. Perhaps the author meant the causality the other way around: Forced current account improvement without adjustment in macro- economic policy tends to be inflationary. This is the real lesson from countries such as Argentina and Mexico. The countries (for example, Korea) that went from current account deficit to surplus while reducing fiscal deficits and estab- lishing competitive exchange rates were also able to control inflation. Corden then turns to the question of exchange rate management as part of the adjustment. In particular, is there any rationale for keeping the exchange rate fixed in the midst of domestic inflation? Here, I think the answer is simple and clear. It is important to maintain a competitive exchange rate. However, a temporarily fixed rate-at a competitive level-can help to cut inflationary expectations in the early stages of a stabilization program. (We have seen this most recently in the Polish and Yugoslav programs, which began with maxi- devaluation and temporarily fixed rates.) This is the conclusion that Corden suggests-but not strongly enough. Arguments about pegging the value of a high-inflation country's currency to the currency of a low-inflation country as an anti-inflation device are familiar from discussions of the European Monetary System. There, some have argued that pegging the Italian lira to the German deutsche mark "tied the hands" of the Italian monetary authorities, adding to the credibility of their anti-inflation program. But announcing a fixed exchange rate cannot possibly be enough. The problem, of course, is that the public knows that the exchange rate can be devalued-and that in fact it will need to be if it continues to appreciate in real terms. The political will to cut inflation, backed by observable policy changes, must accompany the announcement of a fixed exchange rate for the anti- inflation policy to be credible. An exchange rate that is allowed to become more and more overvalued erodes that credibility quickly. Where does this leave us in terms of macroeconomic policy and growth? I draw a different-though not inconsistent-set of conclusions from Corden's. In my view, it helps to identify four different types of countries. The first group, countries that have managed to keep per capita income growing, reinforces an old lesson: stable, orthodox macroeconomic policies create a fertile environment for economic growth. They also provide room for adjusting to bad luck and to policy mistakes as well as room for addressing such difficult problems as reduc- ing poverty. Collins 91 The second group of countries includes ones that have implemented major economic reforms after a period of poor policy and performance but that for various reasons are not enjoying a revival of economic growth. I would include Jamaica and Mexico here. The difficulty is that it can be politically tough to carry through the reforms because the population gets tired of continually tight- ening its belt. In my view, these experiences present a challenge to the United States and other developed countries that also gain from the economic growth and stability of developing economies. The challenge is to lend a helping hand to support these reform efforts until they begin to pay off. In some cases, this may entail financial assistance for public and private investment. In others, it may mean reducing debt and debt service to free domestic resources for investment. This leaves the issues of appropriate macroeconomic policy for countries that are not growing and have major policy problems, which make up the third and fourth groups. The third group includes the countries that are in the midst of full-blown economic crisis, such as Boiivia in the mid-1980s and Poland at the end of 1989. Here the basics of macroeconomic stabilization must come first, and policies can be designed without dotting all the i's. Furthermore, a crisis tends to generate popular support, particularly for a new government, to do something, even if something begins with a difficult transition period. The fourth group includes the countries that are not in the midst of a full- blown crisis but that are in poor shape nonetheless. This category includes countries with a wide range of difficulties, sometimes primarily macroeconomic and sometimes not. In any case, this group typically faces politically difficult environments for implementing-and carrying through--reform programs. This is because it is not at all clear to domestic residents that an adjustment program, which is likely to entail high transition costs, is preferable to the current "mud- dling through." These countries pose the most important challenges for domes- tic politicians, for policymakers abroad, and for academics. That challenge is how to design and to support economic reform programs that are politically feasible but also present a reasonable chance of raising growth and living stan- dards down the road. To my mind, this is the really tough set of questions, and I hope that it is among the questions that Corden and others focus on next. PRO C E E DING S OF T HE WO R L D BANK ANN U AL CON FE RE NC E ON DE V E LOPM ENT EC ONO MI CS 1 9 9 0 FLOOR DISCUSSION OF THE CORDEN PAPER Discussion centered on the "lessons of experience" Corden raised in his paper and particularly on the specific application of Corden's observations and recom- mendations regarding structural adjustment, inflation, and exchange rate policies. Regarding structural adjustment loans (sALs), a World Bank participant noted that for the most part the World Bank makes such loans to countries in balance of payments difficulties. Along with the SALS, the participant pointed out, come other aid flows, repatriation of flight capital (as in Sri Lanka), and private foreign investment. However, "aid booms" can bring about the same damaging effects as do export booms, including the appreciation of the real exchange rate and all that it entails. The participant felt that sometimes the World Bank is not mindful enough of this dilemma and of the danger of undermining gains from the structural adjustment measures that have just been put in place. Another participant suggested that the concept of "conservative euphoria" that Corden mentioned in his presentation deserves greater attention both as it seems to grip the countries themselves and the international organizations, such as the World Bank, when they are analyzing the prospects for the gains from reforms. The participant suggested that after the 1977-78 political changes in Sri Lanka, the World Bank may have fallen prey to "conservative euphoria" about economic changes there, and that the subsequent heavy capital inflows may have been responsible for destabilizing social effects. The participant felt that international organizations need to be particularly sensitive to the political and social consequences of rapid structural adjustment. A participant noted that Dornbusch's conference paper had stressed the importance of a productive tax structure as part of adjustment, including mod- erate, uniform rates of taxation and the absence of any significant subsidies. This is in line with the International Monetary Fund (IMF) and World Bank approaches and with traditional economic thinking, which distinguish between efficiency and equity on the assumption that once we get the efficiency right, the political process can redistribute income and arrive at equity. But the participant was not sure how income would be redistributed if we rule out subsidies and different tax rates. He also wondered whether there was a contradiction when This session was chaired by 11 Sakong, visiting fellow, Institute for International Economics, Washing- ton, D.C., and former minister of finance, Republic of Korea. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 93 94 Floor Discussion of Corden Paper Dornbusch's paper stated that public sector pricing must be corrected, and any income distribution consequences would be resolved through the general tax structure. He also recalled that Corden's early work stated that a subsidy sys- tem, rather than tariffs, should be used to protect infant industries. Are subsidies now ruled out in current thinking? Turning to Corden's paper, the same participant noted the importance of imported intermediate goods, particularly in Africa, where macroeconomic sta- bility has been achieved through import compression. While Corden had emphasized that there was a trade-off between inflation and the current account deficit, it could work the other way. Letting the current account deficit expand allows greater imports of intermediate goods, which increase domestic produc- tion and have supply effects that reduce the inflation rate. Finally, the partici- pant found the section on inflation ambiguous in its discussion of the relation- ship between inflation and growth, particularly at low rates of inflation. He said that there is some evidence to suggest that the variability of the rate of inflation is what is crucial. He wondered whether there was evidence from the seventeen countries Corden cited to suggest that there is a strong negative correlation between the variance of the rate of inflation and the growth rate. Corden conceded that an element of euphoria, as he defined it, might be overenthusiasm by private and official lenders to jump into promising countries; that might have been true in Sri Lanka. The domestic real exchange rate adjustment-price changes that go with any transfers into a country, whether from aid, loans, or an export boom and the familiar Dutch disease phenomenon-is a quite separate aspect. He argued that it was impossible to avoid such price changes without ruling out loans, investments, and so on. The key, he suggested, was the scope governments gave to countercyclical measures, which would offset the relative price effects to some extent. On the variability of inflation, Corden advised against generalizations. Because all high-inflation countries have variable inflation, there is a clear cor- relation between the average level of inflation and variability. There is, how- ever, no clear answer on the effects on growth from the country studies he had reported on in his paper. Brazil, for years one of the highest-growth countries in the developing world, was also a high-inflation country. But there are other countries, in Asia for example, that have high growth and low inflation. Yet Argentina has high inflation and low growth. A number of African countries have relatively low inflation but also low growth. Corden also cautioned that the choice of the time period under study also had an important effect on the relationship between inflation and growth. Extending his discussion of inflation and growth, Corden posed the following question: if high inflation goes with low growth, does it not automatically follow that high inflation caused the low growth? Incompetent governments find themselves with high inflation because they do not have a decent tax system and end up with many distortions; so there is a common cause for both high inflation and low growth. Even if high inflation is not the cause but a symptom, very high and unstable inflation has to be bad for growth prospects. Floor Discussion of Corden Paper 95 On the question of whether structural adjustment loans generate Dutch dis- ease problems, John Williamson (discussant) noted that he preferred to think about the problem in terms of the growth-maximizing exchange rate. If the exchange rate is overvalued, even assuming there is no foreign exchange con- straint, the incentive to invest is low, resulting in a low growth rate from lack of demand for investment. If, in contrast, the currency is too undervalued, there is presumably a high incentive to invest, but saving is inadequate because it has been used to make foreign investments with a lower yield. That is not very clever either. So what you want to find is the growth-maximizing exchange rate in the middle. The real question then is whether there is a surplus or a deficit at this optimal exchange rate. Structural adjustment loans are very helpful in letting you run a deficit; conversely, with a surplus you want to guard against Dutch disease problems by putting the loans into reserves or paying off commercial debt. The difficulty for policy arises when money is coming into the private sector, because then a fiscal surplus is needed in order to sterilize it. Susan Collins (discussant) suggested that on the tax issue, Dornbusch, if he were still in the audience, would probably cite his comparison case, Argentina, where relatively few people are paying taxes, and tax rates are very high. The point was that if the key objectives in designing a tax structure are compliance and enforcement, a uniform structure makes sense. Clearly, you cannot redis- tribute if you do not have the tax revenues to begin with; then there is some tradeoff between being able to generate revenues and then figuring out how to reallocate some of them. She felt that there was useful middle ground between complete uniformity and exclusive focus on a very small group that pays taxes. On the structural adjustment loan issue, Collins said that countries that had achieved macroeconomic stabilization, readjusted to structural problems, and maintained or revived growth had not done it without receiving capital inflows to help ease the cost of adjustment. It is true that capital inflows can create problems and become part of the problem. The real question is how to use them properly, recognizing that they can always be used poorly. A World Bank participant asked Corden what should be done when a country was ignoring Corden's maxims yet was managing quite well. The participant had India in mind. He also asked whether Corden, from his country studies, had identified the impact of structural adjustment policies recommended by the World Bank on growth, particularly microeconomic effects that would enhance efficiency and speed up the return to growth. On the first point, Corden replied that India had been doing well recently but cannot be considered a great success story over the longer period. And even if a country were managing well, it might be able to do better. The participant then expanded his question to draw attention to the high budget deficit relative to gross national product in India. Corden replied that with high private saving, as in India, the question boiled down to whether the deficits amounted to a good use of the saving; a cost-benefit analysis should always apply, whether funds were borrowed from abroad or come from domestic saving. On the effects of structural adjustment Corden said that growth depended on 96 Floor Discussion of Corden Paper so many factors that it was difficult to measure a direct correlation with any one element. However, he did see a difference between policy regimes that looked only at macroeconomic structural adjustment and those that also focused on microeconomic policies. He added that his studies had not attempted to assess the effects of World Bank and IMF programs but had looked at the totality of the country's policies. A participant contrasted Dornbusch's discussion of the need for indexation with Corden's discussion of the need for a reduction of real wages to make countries competitive through a devaluation. Taken together, these implied that in countries where it was not feasible politically to reduce the large public sector employment, indexation would have an adverse effect on the fiscal balance and at the same time an adverse impact on the real exchange rate. How does one handle this policy dilemma? Regarding exchange rate misalignment, the partici- pant asked Williamson how one determined the growth-maximizing exchange rate in reality, when, for example, countries face conservative euphoria. A participant said that after the two major oil price shocks, the need to recycle petrodollars led to a decline in the quality of investment at the microeconomic level. These past misallocations of investments, he felt, had led to the need for SALS. But the SALS themselves have partially prevented improvement in the qual- ity of investment analysis in the World Bank. For some countries, SALs have swamped World Bank conventional investment loans and analysis for them. The question is: where are we going from here? Another participant asked wlhether the advocacy of conservative monetary policies automatically implies putting a brake on fiscal expenditure. In other words, are foreign lenders and domestic lenders always more conservative than central banks? The participant asked why some countries that had followed anti-inflation policies still had inflation rates of 15 to 20 percent-three to four times the world rate. The participant also expressed the view that exchange rates and overvaluation of currency were sometimes overemphasized. For countries that exported mainly minerals or petroleum, the exchange rate, unless absurdly overvalued, did not really matter that much. Other distortions, such as punitive taxation, were much more important. Corden, referring to the question on public sector wage indexation, said that unlike Dornbusch, he was not very sympathetic to indexation. If there is indexa- tion of wages, it is very difficult to reduce them when a country has had an adverse shock in the form of trade deterioration. If wages have to be reduced, Corden said that it may be easier to reduce them in real terms rather than in nominal terms, and there is a possible role for devaluation. Regarding the World Bank's structural adjustment lending, as compared with its project lending, Corden said that the real question was, is there significant conditionality? Corden felt that it is proper for the World Bank and the IMF to be concerned with general economic policies, not just with projects. In that sense, conditionality inevitably involves criteria such as the level of trade restrictions, which is hard to monitor or even measure, and there is a tendency for these loans Floor Discussion of Corden Paper 97 to slide into becoming ordinary balance of payments support loans without real conditionality. His interpretation of SALs was that the countries did not neces- sarily need the money for balance of payments but rather that the loans were incentives to encourage countries to pursue better policies. On inflation, Corden did not think that 15 to 20 percent inflation rates were necessarily a big problem. He cited Chile as being rather successful with a fairly steady level of 15 to 20 percent, after a long history of erratic high inflation. On whether the real exchange rate was overemphasized in considering growth, Corden pointed out that this conference was concentrating on macro- economic matters. At the same time, it was true that many factors had a bearing on growth. Williamson, commenting on indexation and inflation, said that indexation certainly made it difficult to deal with inflation. But he believed Dornbusch was saying that there are circumstances in which the alternative to indexation is much worse; indexation can be part of a bargain in which labor accepts lower wages as part of a stabilization package but wants a guarantee that the real wage will not be eroded beyond a certain point. In those circumstances, he would agree with Dornbusch that there is a role for indexation. On the question of identifying a growth-maximizing exchange rate, William- son replied that one uses an econometric macroeconomic model if one believes in it. If one doesn't, one has to fall back on some rules of thumb. In his experi- ence, the performance of nontraditional exports is a good indicator. If nontradi- tional exports are booming and perhaps even leading to a balance of payments surplus, the exchange rate is possibly undervalued and certainly not overvalued. If nontraditional exports are stagnant, the exchange rate is probably overvalued, even if one is told that some other source of payments, such as mineral exports, is providing all the foreign exchange needed, and the exchange rate is not perceived to be critical. So a simple short-cut rule is to look at nontraditional exports; if they are moving, that is fine, but if they are not, then the exchange rate is probably overvalued, and that is cause for worry. A World Bank participant observed that the speakers had widely different opinions on the acceptable rate of inflation. The relation between inflation and growth appeared to be very nonlinear, with a point in the relationship after which inflation became much worse much more rapidly. The participant asked about the basis of this abrupt nonlinearity of the relationship and how one determined at what inflation rate it occurred. Another World Bank participant commented that there is misconception both within the World Bank and outside about what the Bank did before SALS were started. Before SALS, the Bank basically financed a subset of a country's public investment budget. A condition of such loans in most cases was a public invest- ment review, and endorsement of the whole public investment budget. There is nothing that says that just because one has SALs now, one cannot have cost- benefit analysis and a review of the public investment budget. On the macroeconomic side, this same participant was struck by the difficulty 98 Floor Discussion of Corden Paper of maintaining stability in real exchange rates even if there are no booms. With the instability of real exchange rates observed even in the countries that belong to the Organisation for Economic Co-operation and Development (OECD), the question was whether one could expect to have exchange rate stability in the developing world. This related to another question: what is the macroeconomic policy role of the OECD countries in assisting the resumption of growth in the developing countries? Another participant noted that on the question of indexation and inflation Dornbusch earlier had flatly stated that indexation does not create an inflation problem as long as the budget is balanced. The problem, of course, was that one did not see a country with a balanced budget and indexation. Furthermore, indexing in the context of public sector prices is only partial indexation. Other sectors as well have major repercussions on inflation. On instability in real exchange rates and the implication of the variability of OECD countries' exchange rates, Corden agreed that there is a problem. One possibility is to fix the rate to a trade-weighted basket. In the case of Asian economies, that means a heavy element of yen and of U.S. dollars, and there are the inevitable problems when there are large yen-dollar rate changes. But those are much simpler problems as compared with the bigger problems with which the paper was concerned: the problems of changing conditions in the developing country itself, which is trying to keep the exchange rate fixed and cannot suc- ceed in getting inflation under control. On the question of deciding on the "correct" level of inflation, Corden voiced the opinion that a 20 percent rate is not harmful if it is steady. But a central problem is that the higher the rate of inflation, the more variable it is, which of course creates uncertainty. And if it is rising, one doesn't know where it will stop. Thinking of the countries in his study, Corden suggested that any country that had managed to have 20 or 25 percent inflation or less did not have a big problem, certainly in Latin America. And countries such as Thailand, with 3 to 5 percent inflation, are to be welcomed and certainly not encouraged to work their way up to 20 percent. On the inflation nonlinearity issue, Collins thought that what was important was that moderate to low inflation was qualitatively different from high infla- tion, and the threshold at which moderate inflation becomes high inflation will vary depending on the country, its history, and other factors, particularly exchange rate and fiscal policies. In a particular setting, looking at the inflation rate, does one expect the rate to remain at that level or come down, or does one believe it is a floor and will rise? If it is the latter, then it will generate a large degree of uncertainty, and people devote resources to trying to hedge against that uncertainty. Collins felt that the difficulty of maintaining real exchange rate stability, given the instability of bilateral exchange rates in the developed world, was really a matter of degree. It is certainly harder if one is trading with countries that are having large exchange rate swings between them. But most of the exchange rate Floor Discussion of Corden Paper 99 variability comes from domestic policies and domestic responses to external shocks. On optimal or reasonable inflation rates, Williamson said that he began with the view that the inflation tax is very regressive. He then embraced the accelera- tionist position that there are no permanent benefits from a faster inflation rate; these are purely transitory; the higher the rate of inflation at which the economy stabilizes, the greater the regressive tax. Williamson added that the ability to use corrective inflation, to adjust to a shock by changing the price level rather than the rate of inflation, is so dependent on inflation expectations; if you wish to get expectations down, the only sensible rate of inflation is 0 to 2 percent. If it is too costly to get inflation down, you wait for an opportune moment. But you do not tell yourself that a 10 percent rate of inflation is just fine, and there is nothing to worry about. Williamson endorsed Corden's position that as long as exchange rates are fluctuating among the industrial countries, any developing country with rea- sonably diversified trade can either stabilize the effective exchange rate, which makes sense from a macroeconomic standpoint, or it can stabilize an individual bilateral exchange rate, which may be more attractive from a microeconomic point of view, but it cannot do both. One has to choose between the two, and if trade is very diversified, on balance he would support stabilizing the effective rate, which is done by pegging to a basket. Finally, on the indexation question, Williamson reiterated that indexation can contribute to making a stabilization program acceptable. However, wage indexation based on a high inherited rate of inflation can be problematic just when the budget is first brought back into balance and one most wants to bring down the inflation rate. Then it becomes impossible to stabilize without a high rate of unemployment. In closing the discussion, Sakong (chair) said that the range among speakers in what was considered moderate inflation was somewhat confusing, and there- fore, that country-specific studies were very useful to complement more general conclusions. The political economy of macroeconomic policy and the resump- tion of growth are very important, and Sakong hoped that the World Bank would further pursue this research issue. PRO C EE DING S OF T HE WORLD BANK ANN UAL CON FE REN C E ON D EVE LO PMENT EC ONOM I CS 1 990 The Environment and Emerging Development Issues Partha Dasgupta and Karl-Goran Maler Development economics for the most part has failed to recognize renewable environ- mental resources as economic goods. Yet the lives of the poor in developing countries are intimately tied to the fate of their local common-property resources-soil, fuel- wood, water, and so on. Ignoring the loss of common-property renewable resources (via degradation, centralization, or privatization) in calculations of net national prod- uct can mask the destruction of resources available to a country's poor behind the false appearance of a growing national economy. This paper describes the difference between the problem of managing local common-property resources in developing countries and the problem of protecting the global commons such as the atmosphere and oceans. It outlines the calculations that would include the local commons in the national income accounting of capital assets and argues that the interests of those most dependent on the increasingly scarce local commons will be best served by placing control in their hands, while the centralgovernment provides intfrastructure and educa- tional support. To control the pollution ojf the global commons, on the other hand, the paper sketches a plan in which most countries would have a high incentive to participate-a global allocation of limited, tradable permits to discharge defined pollutants. Environmental resources are of minor importance to poor countries.... They play an insignificant role in the process of economic development. . . . Such resources are luxury goods, and they loom large in public consciousness only when incomes are high. . . Environmental resources are only a rich country's preoccupations. . . . They are a mere diversion created by economists not sensitive to the true needs of the poor in poor countries.... These sentences will seem at once strange and recognizable. They will seem Partha Dasgupta is professor of economics and philosophy at Stanford University, professor of economics at the University of Cambridge, and a research adviser to the World Institute for Development Economics Research (WIDER), Helsinki. Karl-G6ran Maler is professor of cconomics at the Stockholm School of Economics and a research adviser to WIDER. This study forms part of a larger work currently under way at WIDER. The authors thank Lawrence Goulder, Frank Hahn, James Mirrlees, Mohan Munasinghe, Henry Peskin, Robert Repetto, Shekhar Shah, Robert Solow, Joseph Stiglitz, Martin Weale, and John Whalley for their comments. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 101 102 The Environment and Emerging Development Issues strange because today we all like to believe that the ideas they express are not true. At the same time, they will be recognizable to anyone who has delved into the literature on development economics and looked for environmental issues in it. Even were we to search through the vast body of writings on development planning, not to mention the literature on investment criteria, we would be hard put to discover any sign of environmental resources. Academic development economics does not yet acknowledge their existence, in that one will not find them in any recognized survey article, text, or treatise on the subject. To cite only a few instances, Dreze and Stern (1987) and Stern (1989) are surveys of cost-benefit analysis and development economics, respectively. The former, a ninety-page article, contains precisely one sentence on the subject of nonrenewable and renewable resources (and it is to tell readers where to go if they wish to learn about such matters); the latter, an eighty-eight-page article, also contains a single sentence (and this also tells readers where to go should they wish to learn about such matters). A third example is provided by the two- volume Handbook of Development Economics (Chenery and Srinivasan, eds., 1988), which simply has no discussion of environmental resources and their possible bearing on development processes. The disconnection of academic development economics from a central aspect of the lives of people about whom development economics revolves would appear to be pretty much complete. Environmental resources appear in this literature about as frequently as rain falls on the Sahara.1 Despite this neglect, something like a literature on the environment and eco- nomic development is now emerging.2 In this paper we therefore try to place this small literature in perspective and develop what appears to us to be some of the themes with the greatest research potential-that is, those which also have implications for policy in poor countries. This will enable us also to apply economic theory to obtain new insights into the allocation and management of environmental resources. We vvant to do this by providing a framework of thought which will encourage the reader to view environmental resources as economic goods. This will no doubt seem a banal intention, but it is remarkable how much of current writing on the environment displays an innocence of this point of view. Our approach is based on the conjecture that until environmental 1. Yet, environmental economics as an autonomous subject has developed rapidly in recent years. Much work has been done both at the analytical and empirical levels on the valuation of environmental resources (see, for example, Johansson 1987, 1990) and on the identification of appropriate environmen- tal policy instruments (see, for example, Tietenberg 1990). However (and this is the point we are making in the text), both these sets of questions have been discussed in the context of western industrial democracies. We will confirm later in this article that the context matters very much in the field of environmental economics. 2. The incorporation of environmental resources into social cost-benefit analysis for development planning was attempted in Dasgupta (1982). Earlier treatises on social cost-benefit analysis in poor countries, such as Little and Mirrlees (1969, 1974) and Dasgupta, Marglin, and Sen (1972), contain no discussion of these issues. Readers wishing to study environmental economics should consult Ulph (1989) for a review of a number of texts and treatises written over the past fifteen years or so. Dasgupta and MWler 103 resources become a commonplace furniture of economic thinking and modeling, they will continue to be neglected in the design of policy and in the implementa- tion of policy. Periodic "affirmative actions" on the environment is not the right way of going about things. Now there are several routes we can follow for the purpose of introducing our subject. Here, we pursue one which is perhaps less popular today than it used to be: it is to seek a unifying principle concerning the "technology of production" of environmental goods and services. In other words, we start by viewing the matter from the purely ecological side of things. Later, we view the subject from the institutional end. These are complementary avenues of inquiry. Both need to be covered, but it is pretty much a matter of indifference with which one we begin. I. ENVIRONMENTAL RESOURCES AND THEIR PHYSICAL CHARACTERISTICS Environmental problems are almost always associated with resources that are regenerative or renewable but that nonetheless are in danger of exhaustion from excessive use.3 The earth's atmosphere is a paradigm of such resources. That is, under normal courses of events the atmosphere's composition regenerates itself. But the speed of regeneration depends upon the rate at which pollutants are deposited into it, and it depends upon the nature of the pollutants (smoke discharge is clearly different from the release of radioactive material). Now, in talking of a resource we need, first of all, a way of measuring it. For the case in hand we have to think of an atmospheric quality index. The net rate of regenera- tion of the stock is the rate at which this quality index changes over time. This will depend upon the nature and extent of the pollutants which are discharged, and it will also depend upon the current index of quality; that is, the current level of the stock. These are immensely complex, ill-understood matters. There is a great deal of synergism associated with the interaction of different types of pollutants in the atmospheric sink (see Ehrlich, Ehrlich, and Holdren 1977). But the analytical point we are making here remains valid. Animal, bird, plant, and fish populations are also typical examples of renew- able natural resources, and there are now a number of studies which address the reproductive behavior of different species under a wide variety of environmental conditions, including the presence of parasitic and symbiotic neighbors. Land is also such a commodity, for the quality of arable and grazing land can be main- tained by careful use. Overuse, however, impoverishes the soil and eventually produces a wasteland. (The symbiotic relation between soil quality and vegeta- tion cover is, of course, at the heart of the current anxiety over Sub-Saharan erosion.) 3. Except for some remarks in section 11, we are ignoring nonrenewable resources here for reasons of space. For an account of what resource allocation theory looks like when we include exhaustible resources in the production process, see Dasgupta and Heal (1979). 104 The Environment and Emerging Development Issues Underground basins of water often have a similar characteristic. The required analysis, however, is a bit more problematic in that we are concerned both about its quality and quantity. Under normal circumstances an aquifer under- goes a self-cleansing process as pollutants are deposited into it. (Here, the sym- biotic role of different forms of bacteria, as in the case of soil and the atmo- sphere, is important.) But the effectiveness of the process depends, as always, on the nature of pollutants (for example, the type of chemicals) and the rate at which they are discharged. Furthermore, many aquifers are recharged over the annual cycle. But if the rate of extraction exceeds the rate of recharge, the water table drops, thereby raising extraction costs. In fact, for coastal aquifers the issue is not only one of depositing pollutants; excessive extraction alone may cause the groundwater table to drop too far, and saltwater intrusion can then destroy the basin. (For instances of these occurrences, see Dasgupta 1982 and Center for Science and Environment 1982, 1985.) These examples suggest that a number of issues in environmental economics are within the subject matter of capital theory (this is very much the spirit of the exposition in Clark 1976). But there are added complications, among them the fact that the impact on the rate of regeneration of environmental resources of a wide variety of investment decisions is not fully reversible, and in some cases quite irreversible. (As always, one should not be literal. A very slow rate of regeneration produces a strong flavor of irreversibility.) In this limited sense, issues concerning what is usually labeled "pollution" can be studied in the same general sort of way as those concerning animal, bird, plant, and fish popula- tions, aquifers, forests, and soil quality. This provides us with a unified way of thinking about matters. It allows us to use insights we draw from a study of one class of environmental resources when studying another. It forces us to pay attention to the intertemporal structure of economic policies. If this were all, life would be relatively simple. But it is not all. We argue presently that admitting environmental resources into economic modeling ushers in a number of additional, potent complications for development policy. They arise out of the fact that fcor poor people in poor countries, such resources are often complementary to other goods and services. So an erosion of the environmental resource base can make certain categories of people destitute even while an economy on average is seen to be growing. There is thus an intellectual tension between aggregate concerns (for example, the greenhouse effect, or the appropriate mix of natural resources and manufac- tured capital in aggregate production) that sweep across regions, nations, and continents, and concerns (for example, the decline in firewood or water avail- ability) that are specific to the needs and concerns of poor people from as small a group as a village community. This tension should be borne in mind. Environ- mental problems present themselves in the form of different images to different people. At this stage of our understanding it would be a wrong research strategy to try to put them all together into one large basket. In the following section we ignore the detailed, institutional features which surround the use of local envi- Dasgupta and Mider 105 ronmental resources in favor of studying the intertemporal nature of environ- mental policy in terms of aggregate considerations. This allows us to discuss in a simple manner the link between measures of net national product (NNP) and rules for social cost-benefit analysis of investment activity. As a by-product, it enables us to talk briefly about a matter much discussed today-sustainable development-that a decade and a half ago was shown to be the implication of a wider, richer inquiry going under a different name-optimal development. We will pick up the institutional trail in section III. II. DETERMINANTS AND CONSTITUENTS OF WELL-BEING AND NET NATIONAL PRODUCT It is useful to remember that the kinds of resources we are thinking of here are, on occasion, of direct use in consumption (as with fisheries), on occasion in production (as with plankton, which serves as food for fish species), and some- times in both (as with drinking and irrigation water). Their stocks, as we noted earlier, are measured in different ways, depending on the resource: in mass units (such as biomass units for forests, cow dung, and crop residues); in quality indexes (such as water and air quality indexes); in volume units (such as acre- feet for aquifers); and so on. When we express concern about environmental matters we in effect point to a decline in their stock. But a decline in their stock on its own is not a reason for concern. This is seen most clearly in the context of exhaustible resources, such as fossil fuels. To not reduce their stocks is to not use them at all, and this is unlikely to be the right thing to do. To be sure, this is not the case with renewable natural resources, since we could in principle limit their extraction and use to their natural regeneration rate and thus not allow their stocks to decline. But this too may well be the wrong thing to do, in that there is nothing sacrosanct about the stock levels we have inherited from the past. Whether or not policy should be directed at expanding environmental resource bases is something we should try to deduce from consid- erations of population change, intergenerational well-being, technological possi- bilities, environmental regeneration rates, and the existing resource base. The answer cannot be pulled from the air. That it can is a common misconception. Here is a recent example of this: "We can summarize the necessary conditions for sustainable development as con- stancy of the natural capital stock; more strictly, the requirement for nonnega- tive changes in the stock of natural resources, such as soil and soil quality, ground and surface water and their quality, land biomass, water biomass, and the waste-assimilation capacity of the receiving environments" (Pearce, Barbier, and Markandya 1988, p. 6). The passage involves a "category mistake," the mistake being to confuse the determinants of well-being (for example, the means of production) with the constituents of well-being (for example, health, welfare, and freedoms). But leaving that aside, the point is not that sustainable development, even as it is 106 The Environment and Emerging Development Issues defined by these authors, is an undesirable goal. It is that thus defined it has negligible information content. (We are not told, for example, what stock levels we ought to aim at.) This is the price that has to be paid for talking in terms of grand strategies. The hard work comes when one is forced to do the ecology and the economics of the matter. So then, information about stocks on their own is not a sufficient statistic for well-being. What we are after are present and future well-being and methods of determining how well-being is affected by policy. And it is not an accident that the index which, when properly computed, can be used toward this end is net national product.4 Defining Net National Product Net national product is held in ill repute today. (In what follows, we will be concerned only with the real value of NNP.) It is often thought that it is even in principle incapable of reflecting aggregate intergenerational well-being. This belief is not correct. Subject to certain technical restrictions, for any conception of the social good, and for any set of technological, transactional, informa- tional, and ecological constraints, there exists a set of shadow (or accounting) prices of goods and services. These prices, when used in the estimation of NNP, ensure that small projects relative to the size of the economy that increase this index are at once also those that increase the social good.5 A great many of these shadow prices will be person- or household-specific. Put more generally, they will be agent-relative, to reflect the fact that needs and incomes differ along gender, age, caste, or ethnic lines, and that many of the exchanges which take place in poor agrarian and pastoral societies are not anonymous; that is, such exchanges take place in personalized, or named markets.6 These considerations bring out the intellectual tension we spoke of earlier. The point is that in practice it is impossible to take all this into account in one 4. The theory of optimal development in economies with environmental resources was developed over fifteen years ago, and it produced a series of sharp prescriptions, much sharper than the recent literature on sustainable development. Implied in the theory of optimal development were rules for social cost- benefit analysis of investment projects and for estimating net national product. See, for example, Maler (1974), Solow (1974), Dasgupta and Heal (1979), Dasgupta (1982), and Lind (1982). 5. The technical restrictions amount to the requirement that both the set of feasible allocations and the social ordering reflecting the social good are convex. The assumption of convexity is dubious for pollu- tion problems, as was illustrated in Starrett's classic article (see Starrett 1972). Nevertheless, in a wide range of circumstances it is possible to separate out the "nonconvex" sector, estimate net product for the "convex" sector, and present an estimate of the desired index as a combination of the net product of the convex sector and estimates of stocks and their changes in the nonconvex sectors. This is a simple inference from the work of Weitzman (1970) and Portes (1971). 6. We are borrowing the latter terminology from Hahn (1971). As an example of a named market, consider the Hindu jajmani system, which supports a set of mutual economic obligations between landowners (peasant masters) and low-caste families. (See, for example, Wiser 1936, Epstein 1967, and Bowes 1978.) We are not suggesting that such personalized markets are efficient, even when we take account of existing transaction and informational costs. Nor are we suggesting that they lead to equitable allocations. The outcomes are often the result of implicit bargaining among people with vastly different bargaining strengths. See below in the text. See also Dasgupta (1990b) for further discussion. Dasgupta and Miler 107 grand measure of real NNP. The thing to do is to improve on our current ways of doing things and capture these concerns in a piecemeal fashion. This is the approach we will take here. Current estimates of NNP are based on biased sets of prices. They are biased in that the prices that are imputed to environmental resources in situ are usually zero, even though we know that their accounting prices are positive. Thus the social profits of projects which degrade the environment are lower than the profits that are actually imputed to them. This means, in turn, that wrong sets of projects are chosen. In particular, resource-intensive projects look better than they in fact are. One can go further. The bias would be expected to extend itself to the prior stage of research and development. When environmental resources are under- priced, there is little incentive for agencies to develop technologies which econo- mize on their use. The extent of the distortion created by this underpricing will vary from country to country. Quite obviously, poor countries cannot compete with rich ones in the field of basic research. They have to rely on the flow of new knowledge produced in advanced industrial economies. Despite this, poor coun- tries need to have the capability for basic research. This is because the structure of their shadow prices is likely to be quite different, most especially for non- traded goods and services. Even when it is publicly available, basic knowledge is not necessarily usable by scientists and technicians unless they themselves have a feel for basic research. Basic knowledge should not simply be taken off the shelf. Ideas developed in foreign lands ought not merely be transplanted to the local economy. They need to be adapted to suit local ecological conditions. This is where estimating shadow prices (and more generally, creating incen- tives which correspond to these shadow prices) can be of considerable help. Adaptation is itself a creative exercise. Unhappily, it is often bypassed. The foreign technology is simply purchased and installed with little modification. There is great loss in this. This leads to the question of how we should value environmental resources in situ; or in other words, what their shadow prices are in situ. It also leads to the question of how we should measure NNP once we have estimated shadow prices. As it happens, this latter question has a precise answer. How we should define NNP is therefore not a matter of opinion; it is a matter of fact. The theory of intertemporal planning tells us to choose current controls (for example, current consumptions and the mix of current investments) in such a way as to maximize the current value Hamiltonian of the planning problem. As is now well-known, the current value Hamiltonian is the sum of the flow of current well-being and the shadow value of all the net investments currently being undertaken.7 (Thus, the planning exercise generates the entire set of inter- temporal shadow prices.) It is possible to show that the current value Hamilto- 7. The current value Hamiltonian will in general also contain terms reflecting the social cost of breaking any additional constraint that happens to characterize the optimization model. But we are avoiding such technicalities in the text. 108 The Environment and Emerging Development Issues nian measures the return on the wealth of an economy, measured in terms of well-being (see the appendix). The theory also tells us that if the objective of planning (that is, the intergenerational index of well-being) is stationary, then along the optimal program the Hamiltonian remains constant through time. (In saying this we are suitably normalizing the relevant variables. The best eco- nomic treatment of all this is still Arrow and Kurz 1970.) This provides us with the necessary connection between the current value Hamiltonian and NNP. NNP is merely a linearized version of the current value Hamiltonian, the linearization amounting to representing the current flow of well-being by the shadow value of all the determinants of current well-being. In the simplest of cases, where current well-being depends solely on current con- sumption, NNP is reduced to the sum of the social (or shadow) value of an economy's consumption and the social (or shadow) value of the changes in its stocks of real capital assets. The Hamiltonian calculus in fact implies something more. It implies also that along an optimal program NNP at any date is the maximum consumption the economy can enjoy at that date, were it made a condition that consumption be never allowed to fall. In short, NNP at any date along an optimal program is the maximum sustainable consumption at that date. This was not seen immediately as an implication of the mathematical theory of programming, although it should have been transparent from the work of Arrow and Kurz (1970) and Solow (1974). So a proof was provided by Weitzman (1976) in the context of economies which are capable of sustaining a steady economic state, and it was extended to the context of economies with exhaustible resources in Dasgupta and Heal (1979, pp. 244-46; see also Solow 1986). Now "changes in capital stocks" means changes in all capital stocks- manufactured capital as well as natural resources. As of now, we know of no country which tries to account for environmental resource changes in its national income accounting. As we noted earlier, the effect on estimated growth rates could be significant were countries to do this, since it is entirely possible that growth in per capita GNP is currently being achieved in a number of poor countries via a deterioration of the environmental resource base. Growth in NNP over the recent past may well have been negative in a number of countries which are judged to have enjoyed positive growth.8 To understand how this can be, consider as an extreme hypothetical case a 8. See the calculations in Repetto and others (1989). Suggestions that conventional measures of NNP should be augmnented by measures of net nature product (see the interesting work of Agarwal and Narain 1989) are prompted by this concern and are perfectly consistent with what we are saying in the text. By net nature product, Agarwal and Narain mean a measure of net changes in the environmental resource base. The correct measure of NNP is, of course, the sum of conventional NNP and net nature product. It is this aggregate which we are elaborating upon in the text. For tactical reasons it may well be desirable to present valuations of changes in the environmental resource base separately from conventional NNP figures. But we are concerned in the text with analytical issues. For this reason wve are discussing an overall measure of NNP. Dasgupta and Miler 109 country which lives ("optimally") simply by exporting its exhaustible resources and importing consumption goods with the export revenue. Assume, too, that resources are costless to extract and that other than extraction there is no domestic production. NNP in this country, when correctly measured, would be nil. No matter that its current consumption may be high, the social value of the rate of disinvestment would exactly match the shadow value of its consumption rate. Furthermore, being exhaustible, exports must eventually go to zero, and so consumption must eventually go to zero. Such an economy cannot manage a positive, sustainable consumption path. It is this fact which is reflected in NNP, when correctly measured. We would not, however, know any of this from conventional national income accounts. NNP in this economy, as conventionally measured, could be large, and it would even be seen to be rising if the rate of disinvestment were stepped up (see Dasgupta and Heal 1979, pp. 244-46). An important practical question (much discussed in the recent literature; see El Serafy and Lutz 1989) concerns the appropriate treatment in NNP of expendi- ture for protecting the environment. The answer is that it all depends upon whether such expenditures are for the purposes of redressing a flow of environ- mental damage or for enhancing stocks of environmental resources. To take an example, expenditures for liming lakes to counter the flow of acid rains main- tain environmental quality. Were this not to be included in final demand, increased liming (which increases well-being) would be recorded as a decline in NNP. Similarly, capital expenditures (for example, the construction of stack-gas scrubbers, sewage treatment plants, and solar energy equipment) should also not be deducted from estimates of NNP, for they go to augment the durable- capital base. On the other hand, those expenditures which go to enhance resource bases, such as forests, get reflected in the value of changes in resource stocks. In order to avoid double-counting, such expenditures therefore should be excluded from NNP computations. The general moral is that before writing down an expression for NNP it is best to express in a formal model the characteristics of the activities being undertaken in the economy under study. What passes for intermediate expenditure and what does not will be reflected in the Hamiltonian, and one can then readily construct the formula for the country's NNP, and one can simultaneously obtain the system of shadow prices. The procedure is a shade pedantic. But it provides a safeguard against mistakes. In the appendix we provide a formal account of this, and for completeness we sketch the manner in which future uncertainty can be taken into account. Difficulties associated with the estimation of shadow prices and real NNP are compounded by the fact that unlike computers and tractors, environmental resources often affect current well-being directly as stocks, and not merely as service flows (an exception is noise pollution). Fisheries and aquifers are useful not only for the harvest they provide (this is the flow) but also as stocks, because harvesting and extraction costs are low if stocks are large. Tropical forests are beneficial not only because they may supply timber (this is the flow of service); 110 The Environment and Emerging Development Issues they are beneficial also as stocks, because they prevent soil erosion and, in the case of large tropical forests, help maintain a varied genetic pool, and contribute substantially to the recycling of carbon dioxide. To take another example, transportation costs (in particular caloric costs) for women and children would be less were the sources of firewood not far away and receding. Likewise, air and water quality have direct effects on well-being; it is, let us remember, the concentration of pollutants which is relevant here. This has an implication for the current value Hamiltonian, and therefore for real NNP (see the appendix). Estimating Shadow Prices The prior question, of how we should estimate shadow prices for environ- mental resources, is a complex one. But it is not uniformly complex. For com- modities such as irrigation water, fisheries, and agricultural soil, there are now standard techniques of evaluation. These techniques rely on the fact that such resources are inputs in the production of tradable goods (see, for example, Brown and McGuire 1967 for irrigation water; Cooper 1975, Clark 1976, and Dasgupta 1982 for fisheries; and Repetto and others 1989 for soil fertility). For others, such as firewood, and drinking and cooking water, the matter is more complex. The fact remains, though, that even these are inputs in production; specifically, inputs in household production. And this provides us with a method for estimating their shadow prices. To obtain them we will need to have an estimate of household production functions. In some cases (as with fuelwood) the resource input is a substitute for a tradable input (for example, kerosene); in others (as with cooking water) it is a complement (sometimes a weak comple- ment) to tradable inputs (for example, food grain). Such information is in princi- ple obtainable. It allows the analyst to estimate shadow prices. (See Schecter and others 1989 and Maler 1990c. 'We should note, however, that applied work in this area has shied away from analyzing production in poor households in poor countries. There is an urgent need for research in this field. See section IV below.) The approach we have outlined above (the "production function approach") allows one to capture only the "use value" of a resource, and its shadow price may well exceed this. Why? The reason is that there are additional values embodied in a resource stock. One additional value, applicable to living resources, is their "intrinsic worth" as living resources. (We plainly do not think that the value of a blue whale is embodied entirely in its flesh and oil, or that the value of the game in Kenyan game parks is simply the present value of tourists' willingness to pay!) It is pretty much impossible to get a quantitative handle on intrinsic worth, and so the right thing to do is to take note of it, keep an eye on it, and call attention to it whenever the stock is threatened. Another source of value is more amenable to quantification. It arises from a combination of two things common to environmental resources: uncertainty in their future use values and irreversibility in their use. (Genetic materials in tropical forests provide a prime example.) The twin presence of uncertainty and Dasgupta and Miler 111 irreversibility implies that even were the aggregate well-being function neutral to risk, it would not do to estimate the shadow price of an environmental resource solely on the basis of the expected benefit from its future use. Irreversibility in its use implies that preservation of the stock has an additional value-the value of extending one's set of future options. Future options have an additional worth precisely because with the passage of time more information is expected to be forthcoming about the resource's use value, (This additional worth is often called an option value.) The shadow price of a resource is the sum of its use value and its option value (see Arrow and Fisher 1974; Henry 1974; Dasgupta 1982, Fisher and Hanemann 1986; and Maler 1989). As we noted earlier, environmental resources in situ are mostly regarded as free in current economic exercises even while their shadow prices are positive. As we have also noted above, this leads to biases in policy and to biases in the design and installation of new technology. Recall also that even at the prior level of research and development they lead to biases, for there is little incentive for agencies to develop technologies which economize on the use of environmental resources. Indeed, an entire debate on whether economic and environmental considerations are in contraposition is a misplaced one; they are when economic calculations are biased. They would be consonant with each other if environ- mental resources were put on par with the kinds of goods and services which occupy the attention of economists most of the time. III. MARKETS AND THEIR FAILURE All this has been from what one might call the operations-research side of things. It is an essential viewpoint, but it is a limited viewpoint. Its complement is the institutional side, with all its attendant difficulties. Indeed, it is the institu- tional side which has most often been the starting gate for environmental eco- nomics. Particular emphasis is placed upon the fact that markets for these goods either do not exist or are prone to malfunctioning when they do exist (see Maler 1974; Baumol and Oates 1975; and Dasgupta 1982). By "markets" we do not necessarily mean price-guided institutions. Rather, we mean institutions which make available to interested parties the opportunity to negotiate courses of actions. And by "malfunctioning markets" we mean circumstances in which such opportunities are not present (because, say, property rights are unspecified); in which they are exploited at best partially (because, say, the bargainers do not know each other well; see Farrell 1987); or in which they are somewhat one-sided. (This often-one-sidedness of opportunities means that we are thinking of distributional issues as well and not merely those bearing on efficiency.) Nowhere is this generalized form of market failure more common than in those hidden interactions which are unidirectional-for example, deforestation in the uplands, which often inflicts damages on the lowlands. As always, it pays to concentrate first on the assignment of property rights before seeking remedies. The common law, if one is permitted to use this expression in a 112 The Environment and Emerging Development Issues universal context, usually recognizes the rights of polluters, not those of pol- lutees. Translated into our present example, this means that the timber mer- chant who has obtained a concession in the upland forests is under no obligation to compensate farmers in the lowlands. If the farmers wish to reduce the risk of heightened floods, it is they who have to compensate the timber merchant for reducing the rate of deforestation. Stated this way, the matter does look morally bizarre, but it is how things are. Had property rights been the other way around-recognizing pollutees' rights-the boots would have been on the other set of feet, and it would have been the timber merchant who would have had to pay compensation to farmers for the right to inflict the damages which go with deforestation. When the cause of damages is hundreds of miles away, however, and when the victims are thousands of impoverished farmers, the issue of a bargained outcome usually does not arise.9 Thus, judged even from the view- point of efficiency, a system of polluters' rights in such an example would be disastrous.10 We would expect excessive deforestation. Stated in another way, the private cost of logging would in this case be lower than its social cost. Now when the social costs of production of environmental goods are higher than their private costs, resource-based goods may be presumed to be under- priced in the market. Quite obviously, the less roundabout, or less distant, is the production of the final good from its resource base, the greater is this underpric- ing, in percentage terms. Put another way, the lower is the value added to the resource, the larger is the extent of this underpricing of the final product. We may then conclude that countries which export primary products do so by subsidizing them, possibly on a massive scale. Moreover, the subsidy is paid not by the general public via taxation but by some of the most disadvantaged mem- bers of society: the sharecropper, small landholder, or tenant farmer; the forest dweller; and so on. The subsidy is hidden from public scrutiny; that is why nobody talks of it. But it is there. It is real. We should be in a position to estimate it. As of now, we have no such estimates. Matters are usually quite different for economic and ecological interactions which are reciprocal, such as the use by several parties of a piece of grazing land. Many of the production and exchange contracts we see in poor agrarian and pastoral societies over the use of such resources are explicit ones, and compli- 9. Even under such conditions, bargained outcomes do arise sometimes, since community leaders, nongovernment organizations, and a free press have been known to galvanize activity on behalf of the relatively powerless. In recent years this has happened on a number of occasions in India in different sets of contexts. The most publicized one has been the Chipko movement, which involved the threatened disenfranchisement of historical users of forest products. This was occasioned by the state claiming its rights over what was stated to be "public property" and then embarking on a logging program. The connection between environmental protection and civil rights is a close one. There is absolutely no question that political and civil liberties are instrumentally useful for environmental protection. 10. The classic on this subject (Coase 1960) had an argument proving the neutrality of the assignment of property rights on allocative efficiency. Coase's theorem requires stringent assumptions, including the little-noticed one that there are only two parties involved. With more than two parties matters are different, for Shapley and Shubik (1969) and Starrett (1973) have shown that an economy can fail to possess a core allocation if there are polluters' rights over private "bads," such as household garbage. In their examples a core allocation, however, does exist with pollutees' rights. Dasgupta and Mailer 113 ance with them is enforced by means of elaborate rules, regulations, and fines. (In the current literature, see in particular Feder and Noronha 1987; Wade 1987, 1988; Hecht, Anderson, and May 1988; Agarwal and Narain 1989; Chopra, Kadekodi, and Murty 1989; and Ensminger, forthcoming.) However, many contracts are merely implicit, the obligations they entail having often been codified over the years in the form of social norms. In such cases, associated social sanctions sometimes are imposed on violators of such norms, occasionally on those who fail to impose sanctions on violators, even more rarely on those who fail to impose sanctions on those who fail to impose sanctions on violators, and so on. 1 Thus, the fact that a piece of environmental property is not in private (or government) hands does not mean at all that there is institutional failure. The anthropological literature is replete with studies of communities that have developed elaborate patterns of monitoring and control over the use of what are today called common property resources. Economic analysis is thought by some to have implied that common property resources can only be managed through centralized coordination and control (by "centralized" we mean the government or some agency external to the commu- nity of users). Referring to the problem of the commons in the theoretical literature, Wade (1987, p. 220), in a much-cited article, writes that "the prevail- ing answer runs as follows: when people are in a situation where they could mutually benefit if all of them restrained their use of a common-pool resource, they will not do so unless an external agency enforces a suitable rule." And he proceeds to describe enforcement mechanisms in his sample of villages which do not rely on external agencies. But the literature has not implied the necessity of centralized or external control. The theory of games has unraveled the variety of institutional mecha- nisms (ranging from taxes to quantity controls) which can in principle support desirable allocations of common property resources. The theory makes clear, and has made clear for quite some time, that enforcement of the agreed-upon allocation can be carried out by the users themselves. In many cases this may well be the most desirable option. As always, monitoring, enforcement, infor- mation, and transaction costs play a critical role in the relative efficacy of these mechanisms, and we will have something to say about this in what follows. (For a formal, mathematical account, and an informal discussion of the possibilities implied by the formal analysis, see Dasgupta and Heal 1979, chap. 3, sections 11. By a social norm we mean a generally accepted injunction to follow a specified behavioral strategy. Social norms are internalized by people up to a point, so that the infinite chain of meta-norms we have just mentioned in the text is not required for sustaining reciprocity and cooperation. See in particular Wiser (1936), Polanyi (1944, 1977), Goody (1973), Scott (1976), Chambers, Longhurst, and Pacey (1981), Cashdan (1989), and a fine analytical discussion by Elster (1989). We are not suggesting that social norms are efficient, nor that they are necessarily equitable. In fact, inefficiencies and inequities abound. (See, for example, Popkin 1979, Beteille 1983, Iliffe 1987, and Elster 1989.) We are merely asserting that they exist, and that they support outcomes which would not prevail in their absence. Aumann (1981), Fudenberg and Maskin (1986), and Abreu (1988) provide a formal basis for seeing how norms can be sustained even when people have not internalized them, and how they can be both inefficient and inequitable. The required analysis makes use of the theory of repeated games. See Dasgupta (1990, forthcoming) for an elementary exposition. 114 The Environment and EmergGng Development Tssues 4-5.) The confirmation of theory by evidence on the fate of different categories of common property resources is a pleasing success of modern economic analysis. 12 IV. PUBLIC FAILURE AND THE EROSION OF LOCAL COMMONS There is a vast difference between global and local commons. The open seas are common property resources, as are usually village ponds. As economic analysis makes clear, what are problems for the former are by no means prob- lems for the latter. However, it is the global commons, and popular writings on them (for example, the influential article by Hardin 1968), which have shaped popular images of all common property resources. This has been most unfortu- nate because, unlike global commons, the source of the problems associated with the management of local commons is often not the users but other agencies. The images invoked by "the tragedy of the commons" are mostly not the right ones when applied to local commons. The point is that local commons (village ponds and tanks; pastures and threshing grounds; watershed drainage and river beds; and sources of fuelwood, medicinal herbs, bamboo, palm products, resin, gum, and so on) are in no society open for use to all. They are open only to those having historical rights, through kinship ties, community membership, and so on. Those having historical rights of use tend, not surprisingly, to be very protective of these resources. Local commons are easy enough to monitor, and so their use is often regulated in great detail by the community, as we noted earlier, either through the practice and enforcement of norms or through delib- erate allocation of use (see, for example, Wade 1987). The extent of common property resources as a proportion of total assets in a community varies greatly across ecological zones. In India they appear to be most prominent in arid regions, mountain regions, and unirrigated areas. They are least prominent in humid regions and the river valleys (see Agarwal and Narain 1989; and Chopra, Kadekodi, and Murty 1989). An almost immediate empirical corollary of this is that income inequalities are less where common property resources are more prominent. Aggregate income, however, is a differ- ent matter altogether, and it is the arid and mountain regions and unirrigated areas which are the poorest.'3 This needs very much to be borne in mind when policy is devised. In an important and interesting article, Jodha (1986) used data from eighty villages in twenty-one dry districts from seven states in India to estimate that among poor families the proportion of income based directly on common prop- 12. Game-theoretic analyses of common property resources have almost invariably concentrated on the case of large numbers of users, where each user contributes a tiny amount to environmental degrada- tion, but where the total effect, by virtue of the large numbers involved, is substantial. Repetto (1988) confirms that it is this class of cases, and not instances of large investment projects, which provides most of the bases of environmental degradation in poor countries. 13. As might be expected, even within dry regions dependence on common property resources falls with rising wealth across households. The interrelationship between destitution and the erosion of the rural environmental resource base is developed in a wider analytical context in Dasgupta (1989). Dasgupta and Miler 115 erty resources is for the most part in the range of 15 to 25 percent. This is a nontrivial proportion. Moreover, these resources are very much complementary to the sources of income from private property resources, which are mainly labor; milch and draft animals; land for cultivation and crops, but often not the stubble in the postharvest period; common agricultural tools, such as ploughs, harrows, levelers, and hoes; fodder-cutting and ropemaking maclhines; and seeds. Common property resources also provide the rural poor with partial protection in times of unusual economic stress. For landless people they may be the only nonhuman asset at their disposal. As it also happens, a number of such resources-such as fuelwood and water for home use, medicinal herbs, resin, and gum-are the responsibility of women and children. A similar picture emerges from Hecht, Anderson, and May (1988), which describes in rich detail the importance of the extraction of babassu (a Brazilian palm tree) products among the landless in the Brazilian state of Maranhao. The complementarity between this extraction activity and agricultural work is strik- ing, most especially for women. These extractive products are, as it happens, a particularly important source of cash income in the period between agricultural food crop harvests. It is not difficult to see why common property resources matter greatly to the poorest of the rural poor in a society, nor, therefore, is it difficult to understand the mechanisms through which such people may well get disenfranchised from the economy even while in the aggregate it enjoys economic growth (a formal account of the processes through which this can occur is developed in Dasgupta 1989). If one is steeped in social norms of behavior and understands community contractual obligations, one does not calculate every five minutes how one should behave. One follows the norms. This saves on costs all around, not only for the individual as an "actor," but also as a policeman and judge. It is also the natural thing for one to do if one has internalized the norms. But this is sustain- able so long as the background environment remains pretty much constant. It will not be sustainable if the social environment changes suddenly. One might even be destroyed. It is this heightened vulnerability, often more real than per- ceived, which is the cause of some of the greatest tragedies in contemporary society. For they descend upon people who are, in the best of circumstances, acutely vulnerable. The sources which trigger destitution by this general means vary. The erosion of common property resource bases can come about in the wake of shifting populations (accompanying the growth process itself), rising populations, tech- nological progress, unreflective public policies, predatory governments, and thieving aristocracies. There is now an accumulation of evidence on this range of questions, and in what follows we present an outline of the findings in three sets of studies. In his work on the drylands of India, Jodha (1986) noted a decline in the geographical area covering common property resources ranging from 26 percent to 63 percent over a twenty-year period. This was in part due to the privatiza- tion of land, well over half of which in his sample had been awarded to the rural 116 The Environment and Emerging Development Issues nonpoor. He also noted a decline in the productivity of common property resources on account of population growth among the using community. In an earlier work, Jodha (1980) identified an increase in subsistence requirements of the farming community and a rise in the profitability of land exploitation from cropping and grazing as a central reason for increased desertification in the state of Rajasthan in India. Jodha argued that, ironically, it was government land reform programs in this area, unaccompanied by investment in improving the productive base, which had triggered the process (for a formalization of the dynamics of such a process, see Dasgupta 1982, chap. 6). Ensminger (forthcoming), in a study of the privatization of common grazing lands among the Orma in northeastern Kenya, indicates that the transformation took place with the consent of the elders of the tribe, and she attributes this willingness to changing transaction costs brought about by cheaper transporta- tion and widening markets. The elders were, quite naturally, from the stronger families, and it does not go unnoted by Ensminger that privatization has accen- tuated inequalities. However, she provides no data to tell whether the process has increased the prevalence of destitution among the economically weak. In an earlier, much-neglected work on the Amazon basin, E. Feder (1977, 1979) described how massive private investment in the expansion of beef-cattle production in fragile ecological conditions has been supported by domestic gov- ernments in the form of tax concessions and provision of infrastructure, and by loans from international agencies such as the World Bank. The degradation of vast tracts of valuable environmental resources was, not surprisingly, accom- panied by the disenfranchisement of large numbers of small farmers and agri- cultural laborers from the economy, and it made traditional forest dwellers destitute, at best, and simply eliminated them, at worst.14 The sources of the transformation of common property resources into private resources described in these three sets of studies are, of course, quite different. Consequently, the ways in which they affected those with historical rights have been quite different. But each is understandable and believable. Because they are confirmed by economic theory, the findings of these case studies are almost certainly not unrepresentative. They suggest that privatization of village com- mons and forest lands, although hallowed at the altar of efficiency, can have disastrous distributional consequences, disenfranchising entire classes of people from economic citizenship (for alternative demonstrations of this theorem, see Cohen and Weitzman 1975 and Dasgupta and Heal 1979, chap. 3). They also 14. See also Dasgupta (1982, chap. 2) and Hecht (1985). The data suggest that during the decades of the 1960s and 1970s protein intake by the rural poor declined even while the production of beef protein increased dramatically. Much of the beef was destined for exports, for use by fast-food chains. These matters, which are an instance of the intricate link between economic, social, and financial institutions, have been taken up anew by Repetto (1988), Mahar (1988), and Binswanger (1989). The latter in particular has shown how in Brazil the exemption from taxation of virtually all agricultural income (allied to the fact that logging is regarded as proof of land occupancy) has provided strong incentives for the acquisition of agricultural lands by the higher-income groups and a general incentive for the acquisition of forest lands for the purposes of deforesting them. Dasgupta and Miler 117 show that public ownership of such resources as forest lands is by no means necessarily a good basis for a resource allocation mechanism. Decisionmakers are in these cases usually far removed from the site (living as they do in imperial capitals); they have little knowledge of the ecology of such matters; their time horizons are often short; and they are in many instances overly influenced by interest groups far removed from the resource in question. All this is not at all to suggest that rural development is to be avoided. It is to say that resource allocation mechanisms which do not take advantage of dis- persed information; which are insensitive to hidden (and often not so hidden) economic and ecological interactions (that is, general equilibrium effects); which do not take the long view; and which do not give a sufficiently large weight to the claims of the poorest within rural populations (particularly the women and children in these populations) are going to prove environmentally disastrous. It appears, then, that during the process of economic development there is a close link between environmental preservation and the well-being of the poor, most especially the most vulnerable among the poor. Elaboration of this link has been one of the most compelling achievements at the interface of anthropology, eco- nomics, and nutrition science. The links between environmental degradation and an accentuation of depri- vation and hardship can take forms which are even today not always appreci- ated. The responsibilities for gathering fuelwood and fetching water for domes- tic use in most rural communities fall upon women and children. When allied to household chores and their farming obligations, the workload of women in South Asia in terms of time is often one-and-a-half to two times that of men (see, for example, Fernandes and Menon 1987; Kumar and Hotchkiss 1988; B. Agarwal 1989). This workload has over the years increased directly as a conse- quence of receding resources. It is very much worth reminding ourselves that we are speaking of a category of people of whom more than 50 percent suffer from iron deficiency, of whom only a little below 50 percent suffer from wastage, and who in some parts of the world work fifteen to sixteen hours a day during the busy agricultural season. Thus, communities in the drylands of the Indian sub- continent and in Sub-Saharan Africa today often live miles away from fuel and fodder sources and permanent water sources. Surveys in East Africa have shown, for example, that women and children spend up to five hours a day collecting water during the dry season (see Food and Agricultural Organization 1987). The consequence is that anything between 10 and 25 percent of daily daytime energy expenditure is required for collecting water.15 15. See Chen (1983) for a review of the link between improved water supply and health benefits among the rural poor. We should note that a similar problem is associated with fuelwood collection. In northern India, for example, it is thought that some 75 percent of firewood for domestic use comes from twigs and fallen branches. From data that are now available from the drylands of India on time allocation on the part of women in fuelwood collection, the energy costs in this activity would seem to be also in the range of 10 percent to 25 percent. It should be noted that estimates of the energy-cost of collection are essential ingredients in tlre calculation of the shadow prices of fuelwood and water. 118 The Environment and Emerging Development Issues All this cannot but be related to the fact of high fertility and low literacy in rural areas of most poor countries. Poverty and the thinness of markets make it essential for households to engage in a number of complementary production activities: cultivation, cattle grazing, fetching water, collecting fuelwood, cook- ing food, and producing simple marketable products. Each is time-consuming. (Labor productivity is low not only because capital is scarce but also because, as we have just noted, environmental resources are scarce, too.) If it is to survive, a household simply has to accomplish these tasks each day, and a small household cannot do them all. Each household needs many hands. Children are needed as workers by their parents, even when parents are in their prime. Children are not merely a consumption good (as in Nerlove, Razin, and Sadka 1987), nor are they only a means to old-age security (as in Cain 1983). They are also of current use to parents. But a high rate of fertility and population growth further dam- ages the environmental resource base, which in turn can provide further (pri- vate) incentives for large families, which in turn further damages the resource base, and so on-until some countervailing factors (whether public policy or some form of Malthusian check) stop the spiraling process. But by the time this happens millions of lives have suffered (for further development of these issues, see Dasgupta, forthcoming, and Nerlove and Meyer 1990). Information concerning the ecology of local commons is often dispersed, and is usually in the hands of the historical users. There are exceptions, of course, but as a general rule this makes it desirable that the local commons be protected as commons and that decisions regarding local commons be left in the hands of the users themselves. This is because the local commons will almost certainly remain the single source of essential complementary goods for poor people for a long while yet. To be sure, it is essential not only that governments provide infrastructural and credit and insurance facilities, but that they make new infor- mation concerning technology, ecology, and widening markets available to the users. But there is little case for centralized control. Quite the contrary, there is a case for facilitating the growth of local, community decisionmaking-in particu- lar, decisionmaking by women, who are for the most part the main users of such resources. The large, often fragmented literature on local common property resources is beginning to offer us an unequivocal picture that during the process of economic development the protection and promotion of environmental resources would be best served were a constant public eye kept on the conditions of the poorest of the poor in society. Environmental economics and the eco- nomics of destitution are tied to each other in an intricate web. We should not have expected it otherwise. V. INTERNATIONAL FAILURE AND THE EROSION OF GLOBAL COMMONS Global commons pose a different type of problem. The impossibility of estab- lishing adequate property rights to the atmosphere, to watersheds, and to large Dasgupta and Maler 119 bodies of water, such as the oceans, are a cause of inefficiencies in the allocation of resources. In the case of the atmosphere (for example, over the matter of global warming), there is not even the option of "voting with one's feet." Fur- thermore, future generations are not directly represented in today's forum for decisionmaking. Their interests are included only indirectly through transac- tions between different coexisting generations. Thus the inefficiencies and ineq- uities involved are not merely static ones but are intergenerational ones as well (see Maler 1990b for a more detailed discussion of these issues). From this it follows that the international community needs consciously to design systems which improve upon existing resource allocation mechanisms. The most complicated international environmental problems are, like the local commons, characterized by reciprocal externalities; that is, most countries that contribute to environmental damages also suffer from them. Emissions of greenhouse gases are an instance of this. A central problem is that "reciprocal" countries do not contribute to the damages in equal amounts. Thus, for a cooperative outcome to be achievable, some financial transfers will be neces- sary, if only in an implicit manner. Several such systems suggest themselves- debt relief for the preservation of the Amazon being among the most frequently talked about. This is not to say that agreements cannot be reached without side payments; it is only to say that they will tend to be less efficient. Barrett (1990) has argued, for example, that one should not expect all countries to sign the Montreal protocol on emissions of chlorofluorocarbons (CFCS). (The protocol involves no side payments.) If an equilibrium exists, it can only involve some countries signing the protocol. The reason is that were only a few countries to sign the protocol, national benefits from further reduction in CFC emission would be high. This would induce more countries to sign. However, were many countries to sign the protocol, national benefits from further reduction would be small, and it would then not be worth a country's while to sign the agreement. Direct (side) payments among countries for solving environmental problems have not been so common. When made, side payments have tended to be non- pecuniary; for example, trade and military concessions (see Krutilla 1966; Kneese 1988). Very recently, an agreement has been reached on reducing the production and use of CFCS in developing countries. This has involved the cre- ation of an international fund for technological transfers to these countries. It is a most promising development, and it needs to be studied carefully for the purposes of further development. One broad category of allocation mechanisms well worth exploring in the international context involves making the global commons quasi-private. The basic idea, which originated in Dales (1968), is similar to the principle currentiy being experimented with in the United States. The idea, if extended to the international sphere, would have the community of nations set bounds on the total annual use of the global commons, such as the atmosphere, have it allocate 120 The Environment and Emerging Development Issues an initial distribution of transferable national rights which add up to the aggre- gate bound, and allow the final allocation among different users to be deter- mined by competitive markets.16 For example, consider greenhouse gases. Suppose the community of nations desires to reduce emissions to a prescribed global level. Units of the various gases would then be chosen so that all gases have the same (expected) effect on global climate. In other words, at the margin the emission of one unit of any one gas would have the same (expected) climatic effect as the emission of one unit of any other gas. The scheme would allow countries to exchange permits for one gas for permits for any other. Countries would receive an initial assignment of marketable permits. As is well known, this scheme has informational advan- tages over both taxes and quantity controls on individual emissions. (See Dasgupta, Hammond, and Maskin 1981 for a formal analysis of optimal incen- tive schemes for pollution control.) Furthermore, were the permits to refer to net emissions (that is, net of absorption of carbon dioxide by green plants), the scheme would provide an incentive for countries with fast-growing tropical rain forests to earn export revenue by encouraging forest growth and then selling permits to other countries. The scheme also has the advantage that the necessary side payments required to induce all (or most) countries to participate in the agreement can be made through the initial distribution of emission permits. Countries which do not expect severe damages from global warming would also wish to participate were they to be provided initially with a sufficient number of permits (or rights). The sticking point will clearly be over reaching an agreement on the initial distribution of permits among nations. (How a national government allocates the nation's rights among agencies within the country is a different matter.) But the point here is that if the bound set on annual aggregate greenhouse emissions is approximately optimal, it is always possible to distribute the initial set of rights in such a way that all countries have an incentive to join the scheme. For this reason one cannot overemphasize the fact that there are large potential gains to be enjoyed from international cooperation; a scheme involving the issue of marketable permits in principle offers a way in which all naticns can enjoy these gains. The argument that "national sovereignty" would be endangered is in fact no argument, for the point about global commons is precisely that they are beyond the realm of national sovereignty. VI. SUMMARY In this article we have tried to present a perspective on what we take to be the central emerging issues at the interface of environmental and development con- 16. See Tietenberg (1980, 1990) for reviews of the experience that has been accumulated with such schemes in the United States. See also Dasgupta (1982) and Maler (1990a) for mathematical formaliza- tions of these schemes under varying environmental circumstances. The motivation behind these formal- izations is that they enable us to calculate the efficiency gains realizable by such resource allocation mechanisms. Dasgupta and Maler 121 cerns. The fact that for such a long while environmental and development economics have had little to say to each other is a reflection only of these academic disciplines; it does not at all reflect the world as we should know it. Poor countries are for the most part agrarian and pastoral, and it is but natural that the bulk of society in these lands depends crucially on renewable natural resources. One of the sobering lessons of the international development experi- ence has been that the magnitudes of poverty and destitution have proved singu- larly resistant to reduction in many parts of the globe. It is beyond our compe- tence to try to explain this, but there is growing evidence that acute poverty and environmental degradation are closely linked in most poor countries. We have argued that the poor in resource-exporting countries are very likely subsidizing these exports. A reasonable rule of thumb for the "environmentalist" would therefore be to keep a constant eye on the poorest of the poor. Their activities (for example, migration patterns of communities or time-use patterns of poor women) are often a good signal of the state of the environment. By the same token, it is fatuous to talk and write about poverty and development unless we simultaneously study the fate of environmental resources under alternative resource allocation mechanisms. The separation of environmental and develop- ment economics has proved to be enormously costly in terms of lost hopes and wasted lives. We have argued that local and global commons pose quite different problems and that environmental damages at the local level have often been inflicted upon such communities (possibly unwittingly) by outside agencies-very often by their own governments. There are countries where information about the envi- ronmental resource base is almost wholly absent. If environmental resources are to be brought into line with other capital assets, they must as a minimum enter national income accounting. In this paper we have presented an outline of how to go about doing this in a meaningful way. We would imagine that it is the global commons which will occupy the international stage in the immediate future, as evidence accumulates on the mechanisms underlying their degrada- tion. We have presented the bare bones of a resource allocation mechanism in which countries receive marketable permits for the use of the global commons and in which all nations have an incentive to participate. APPENDIX Consumption and Accumulation with Environmental Effects: A Definition of NNP In this part of the appendix we present what we hope is a canonical model involving consumption and accumulation with environmental effects. Our aim is to display the connection between shadow prices, rules for project evaluation, and national income accounting in a context which is simple but which has at the same time sufficient structure to allow us to obtain a number of the prescrip- 122 The Environment and Enierging Development Issues tions we alluded to in the text, especially in section II. Keeping to essential matters here, we ignore the kinds of "second-best" constraints (including dis- equilibrium phenomena) which have been the center of attention in the literature on project evaluation, as in Dasgupta, Marglin, and Sen (1972) and Little and Mirrlees (1974). We consider an economy with a multipurpose, manmade capital good whose stock is denoted by K1. If L1 is the labor effort combined with this, the flow of output is taken to be Y = F(K1, Li), where F(.) is an aggregate production function. (In what follows we assume that all functions satisfy conditions which ensure that the planning problem defined below is a concave program. For example, we assume that F(.) is concave.) The economy enjoys in addition two sorts of environmental resource stocks: clean air, K2, and forests, K3. Clean air is valued directly, whereas forests have two derived values: they help keep the atmosphere clean, and they provide fuelwood, which, too, is valued directly. Finally, we take it that there is a flow of environmental amenities, Z, which directly affects aggregate well-being. Forests enjoy a natural regeneration rate, but labor effort can increase it. Thus we denote by H(L2) the rate of regeneration of forests, where L2 is labor input for this task, and where H(.) is, for low values of L2 at least, an increasing function. Let X denote the rate of consumption of fuelwood. Collecting this involves labor effort; let this be L3. Presumably, the larger the forest stock the less is the effort required (in calorie requirements, say). We thus assume that X = N(K3, L3), where N(.) is an increasing, concave function of its two arguments. Output Y is a basic consumption good, and this consumption is also valued directly. However, we take it t-hat the production of Y involves pollution as a by- product. This reduces the quality of the atmosphere both as a stock and as a flow of amenities. We assume, however, that it is possible to take defensive measure against both these ill effects. First, society can invest in technologies that reduce the emission of pollutants, and we denote the stock of this defensive capital by K4. If P denotes the emission of pollutants, we have P = A(K4, Y), where A is a convex function, decreasing in K4 and increasing in Y. Second, society can mitigate damages to the flow of amenities by expending a portion of final output, at a rate R. We assume that the resulting flow of amenities has the functional form, Z = J(R, P), where J is increasing in R and decreasing in P. There are thus four things that can be done with output Y: it can be consumed (we denote the rate of consumption by C); it can be reinvested to increase the stock of K1; it can be invested in the accumulation of K4; and it can be used, at rate R, to counter the damages to the flow of environmental amenities. Let Q denote the expenditure on the accumulation of K4. Now, the environment as a stock tries to regenerate itself at a rate which is an increasing function of the stock of forests, G(K3). The net rate of regeneration is the difference between this and the emission of pollutants from production of Y. Dasgupta and Mdler 123 We can therefore express the dynamics of the economy in terms of the following equations: (1) dK1/dt =F(K1, L1)-C-Q-R (2) dK2/dt = G(K3) - A[K4, F(K1, L1)] (3) dK3/dt = H(L2) - X (4) dK4/ dt = Q (5) X = N(K3, L3) (6) Z = J{R, A[K4, F(K1, L1)]. The current flow of aggregate well-being, W, depends positively upon aggre- gate consumption, C; the output of fuelwood, X; the flow of environmental amenities, Z; and the quality of the atmospheric stock, K2. However, it depends negatively upon total labor effort, L = L1 + L2 + L3. (As noted above, labor effort could be measured in caloric terms.) We thus have W(C, X, Z, K2, L, + L2 + L3). Stocks of the four types of assets are given at the initial date; the instantaneous control variables are C, Q, R, X, LI, L2, and L3; and the objective is to maxi- mize the (discounted) sum of the flow of aggregate well-being over the indefinite future. We take well-being to be the numeraire. Letting p, q, r, and s denote the (spot) shadow prices of the four capital goods, K1, K2, K3, and K4 respectively, and letting v be the imputed marginal value of the flow of environmental ameni- ties, we can use equations 1 through 6 to express the current value Hamiltonian, V, of the optimization problem as V = W[C, N(K3, L3), Z, K2, L1 + L2 + L3] (7) + p[F(K1, L1) - C - Q - R] + q{[G(K3) - A[K4, F(K1, L1)]} + r[H(L2) - N(K3, L3)] + sQ + v[JtR, A[K4, F(K1, L1)]} - Z]. Recall that the theory of optimum control instructs us to choose the control variables at each date so as to maximize expression 7. (Notice that we have used equations 5 and 6, and so the controls are now C, Z, Q, R, L1, L2, and L3.) Writing by Wc the partial derivative of W with respect to C, and so forth, it is then immediate that along an optimal program the control variables and the shadow prices must satisfy the conditions: (i) Wc = p; (ii) WXN2 + WL = rN2; (iii) WZ = v; liv) WL = [(q - VJ2)A2 - p]F2; (v) WL = -rdH(L2)/dL2; (vi) p = vJ1; and (vii) p s. F2 stands for the partial derivative of F with respect to its second argument, 124 The Environment and Emerging Development Issues L1; and as mentioned earlier, L = L1 + L2 + L3. We have used this same notation for the derivatives of N(.), J(.), and A(.). Interpreting these conditions is instructive, for they tell us what kinds of information we need to estimate shadow prices. (We do not write down the intertemporal "arbitrage conditions" which these shadow prices must also satisfy, since we do not use them.) But for our purposes, the point to note is that we can immediately derive the correct expression for NNP from equation 7. It is the linear support of the Hamiltonian along the optimal program. In order to keep the expression for NNP from becoming overly cumbersome, let us denote by 0* the vector of all the nonprice arguments in the Hamiltonian function along the optimal program at any given date. Thus, Q* = (C*, Z*, Q*, R*, K*, K*, K*, K*, L*, L, L*). It follows from taking the Taylor expansion around 0* that the linear support of the Hamiltonian is (8) V(O*) + WcC*- + W' V + W Z* + WK,K + WL(L* + L* + L*) pdK,*/dt + qdK*/dt + rdK* /dt + sdK4 /dt where Z- is equal toJtR*, A[K , F(K*, L )]}. We may divide the whole expression by Wc to express NNP in aggregate consumption numeraire. It should also be recalled that by assumption WL is negative. Now, for evaluating a marginal project V(O*) is irrelevant, as it is a constant term. We may as well then drop it. In this case, NNP in the optimizing economy, measured in well-being numeraire, is the remaining term of equation 8, namely: (9) NNP = WCC + W,xX + WzJIR, A[K4, F(K1, L.)]} + WK K2 + WL(L1 + L2 + L3) + pdK1/dt + qdK2/dt + rdK3/dt + sdK4/dt. However, for international comparisons V(O*) is of interest, since countries differ in the stocks of capital they hold. In this case NNP should be measured on the basis of equation 8 (see Weale 1990). It is possible to show that there is a permissible representation of the well-being function for which (10) V(O*) = 6(pK* + qK* + rK* + sK*) where 6 (> 0) is the rate at which the flow of aggregate well-being is discounted. This means that the Hamiltonian measures the sum of the social returns on all of society's capital assets. In short, it is a measure of society's wealth (see Solow 1986). Notice that all resources and outputs are valued at the prices which sustain the optimum program.17 To highlight the points we want to make here, we have 17. One could altemnatively think of a sequence of policy reforms and use shadow prices defined at the existing structure of production. Given that the planning program is by hypothesis concave, a sequence of such moves would take the economy ultimately to the optimum. For a simplified exposition of the connection between these two modes of analysis (reforms and optimization), see Dasgupta (1982, chap. 5). Dasgupta and Maler 125 chosen to work with a most aggregate model. Ideally (income) distributional issues will find reflection in the aggregate well-being function. These considera- tions can readily be translated into the estimates of shadow prices (see Dasgupta, Marglin, and Sen 1972). Let us suppose that we are involved in the choice of projects. A marginal project is a perturbation of the optimal program. It can be expressed as a ten- vector (dC, dX, dR, dL1, dL2, dL3, dI1, d12, dI3, dI4), where I, = dK,/dt, (i = 1, 2, 3, 4); and dC, and so on, are small changes in C, and so on. A marginal project has no effect on the fourth term on the right-hand side of equation 9. (For the purposes of social cost-benefit analysis we could, therefore, simply ignore this term when estimating NNP.) Along an optimal program, the social profitability of the last project is nil; that is, its contribution to NNP is nil. This follows from the fact that the controls are chosen so as to maximize expression 7. All this is well-known, and our purpose here is to obtain some additional insights. Scrutiny of equation 9 tells us the following: * Were wages to equal the marginal ill-being of work effort, wages would not be part of NNP. Put in other words, the shadow wage bill ought to be deducted from gross output when we estimate NNP. However, if labor is supplied inelastically, it is a matter of indifference whether the wage bill in this optimiz- ing economy is deducted from NNP. Conversely, were we to recognize a part of the wage bill as a return on the accumulation of human capital, that part would be included in NNP. C Current defensive expenditure, R, against damages to the flow of environmen- tal amenities should be included in the estimation of final demand (see the third term in equation 9). Moreover, investments in the stock of environmen- tal defensive capital should also be included in NNP (see the final term of equation 9). Now, expenditures which go toward enhancing the environment find expression in the value that is imputed to changes in the environmental resource stock. We may conclude, therefore, that such expenditures should not be included in estimates of NNP. (Notice the absence of sQ in equation 9.) • The value of changes in the environmental resource base (K2 and K3) should be included in NNP. However, anticipated capital gains (or losses) are not part of NNP. Uncertainty In this part of the appendix we present a sketch of the arguments involved in the definition of NNP when there are random events in the future that are expected to have an effect on the value of the then-existing capital assets. As an example, we could imagine the discovery and installation of cleaner production technologies which make existing abatement technologies less valuable. For simplicity of exposition, we will assume that such discoveries are uninfluenced by policy; for example, research and development policy, although such policy 126 The Environment and Emerging Development Issues can easily be incorporated. The analysis builds on Dasgupta and Heal (1974) and Dasgupta and Stiglitz (1981). However, these earlier studies aid not address the measurement of NNP, which is our present concern. It is most striking to consider discrete events. Thus, imagine that at some random future date, T, an event occurs which is expected to affect the value of the then-existing stocks of capital. We consider the problem from the vantage point of the present, which we denote by t = 0. Let us assume that there is a (possibly subjective) probability density function, 7rt, over the date of occur- rence. (We are thus supposing for expositional ease that the event will occur at some future date.) From this we may define the cumulative function 4b. We take it that the social good is reflected by the expected value of the sum of the discounted flow of future well-being. Were the event in question to occur at date T, the economy in question would enter a new production and ecological regime. In what follows we continue to rely on the notation which was devel- oped in the first part of the appendix. As is proper, we now proceed to work backwards. Thus, let KT(with i = 1, 2, 3, 4) denote the stocks of the four assets at date T. Following an optimal economic policy subsequent to the occurrence of the event would yield an expected flow of aggregate well-being. This flow we discount back to T. This capitalized value of the flow of well-being will clearly be a function of KT Let us denote this by B(KT, KT, KT, Kr). It is now possible to show that until the event occurs, the optimal policy is to pretend that the event will never occur, and to assume that the flow of aggregate well-being is given, not by W(.), as in the first part of the appendix, but by (1 - bD)W(.) + 7r,B(.). We may therefore conclude from the analysis of the first part of the appendix that NNP at any date prior to the occurrence of the event is given by the equation NNP = (1 - F)LWCC + WXX + WZZ + WK K2 (11) + WL(L1 + L2 + L3) + pdK1/dt + qdK2/dt + rdK3/dt + sdK4/dt] + 7r[BK1K2 + BK2K2 + BK3K3 + BK4 K4]- (As in the first part of the appendix, we are suppressing the time index for clarity.) Notice that if the event is not ever expected to occur, then 7rt = 0 for all t, and consequently (1 - 4't) = 1 for all t. In this case, equation 11 reduces to equation 9. Note also that no marginal economic activity can affect the terms within the second pair of square brackets at date t. It follows that we could as well drop these terms from estimates of NNP. Now, were we to do this, equation 11 would look the same as equation 9. But their values would not be the same. This is because the shadow prices (for example, p, q, r, and s) appearing in equation 11 assume quite different values from those in equation 9: future possibilities hypothesized in this model economy are quite different from those assumed in the model economy of the previous section of the appendix. Finally, we should note that the shadow prices appearing in equation 11 are Arrow- Debreu contingent commodity prices. Dasgupta and Miler 127 REFERENCES Abreu, Dilip. 1988. "On the Theory of Infinitely Repeated Games with Discounting." Econometrica 56, no. 2: 383-96. Agarwal, Anil, and Sunita Narain. 1989. Towards Green Villages (New Delhi, India: Center for Science and Environment). Agarwal, Bina. 1989. "Rural Women, Poverty and Natural Resources: Sustenance, Sustainability and Struggle for Change." Economic and Political Weekly 24, no. 43: WS46-66. Arrow, Kenneth J., and Anthony Fisher. 1974. "Preservation, Uncertainty and Irrever- sibility." Quarterly Journal of Economics 88, no. 2: 312-19. Arrow, Kenneth J., and Mordecai Kurz. 1970. Public Investment, the Rate of Return and Optimal Fiscal Policy. (Baltimore, Md.: Johns Hopkins University Press). Aumann, Robert. 1981. "Survey of Repeated Games." In M. Shubik, ed., Essays in Game Theory and Mathematical Economics. Mannheim, Germany: Wissenschafts- verlag, Bibliographisches Institut. Barrett, Scott. 1990. "The Problem of Global Environmental Protection." Oxford Review of Economic Policy 6, no. 1: 68-79. Baumol, William, and Wallace Oates. 1975. The Theory of Environmental Policy. Englewood Cliffs, N.J.: Prentice Hall. Beteille, Andre, ed. 1983. Equaiity and Inequality: Theory and Practice. New Delhi, India: Oxford University Press. Binswanger, Hans. 1989. "Brazilian Policies that Encourage Deforestation in the Ama- zon." World Bank Environment Department Paper 16. Washington, D. C. Processed. Bowes, Pratima. 1978. Hindu Intellectual Tradition. New Delhi, India: Allied Publishers. Brown, Gardner, and C. Bart McGuire. 1967. "A Socially Optimum Pricing Policy for a Public Water Agency." Water Resources Research 3, no. 1: 33-44. Cain, Meade. 1983. "Fertility as an Adjustment to Risk." Population and Development Review 9, no. 4: 688-702. Cashdan, Elizabeth, ed. 1989. Risk and Uncertainty in Tribal and Peasant Economies. Boulder, Colo.: Westview Press. Center for Science and Environment. 1982. The State of India's Environment. New Delhi, India: Center for Science and Environment. . 1985. The State of India's Environment. New Delhi, India: Center for Science and Environment. Chambers, Robert, R. Longhurst, and A. Pacey, eds. 1981. Seasonal Dimensions to Rural Poverty. London: Francis Pinter. Chen, Lincoln C. 1983. "Evaluating the Health Benefits of Improved Water Supply through Assessment of Nutritional Status in Developing Countries." In B. Under- wood, ed., Nutrition Intervention Strategies in National Development. New York: Academic Press. Chenery, Hollis, and T. N. Srinivasan, eds. 1988. Handbook of Development Eco- nomics, vols. 1 and 2. Amsterdam: North-Holland. 128 The Environment and Emerging Development Issues Chopra, Kanchan G., K. Kadekodi, and M. N. Murty. 1989. Participatory Develop- ment: People and Common Property Resources. New Delhi, India: Sage. Clark, Colin W. 1976. Mathematical Bioeconomics: The Optimal Management of Renewable Resources. New York: Wiley. Coase, Ronald. 1960. "The Problem of Social Cost." Journal of Law and Economics 3, no. 1: 1-44. Cohen, John S., and M. L. Weitzman. 1975. "A Marxian View of Enclosures." Journal of Development Economics 1, no. 4: 287-336. Cooper, Richard. 1975. "An Economist's View of the Oceans." Journal of World Trade Law 9: 347-77. Dales, J. H. 1968. Pollution, Property and Prices. Toronto: University of Toronto Press. Dasgupta, Partha. 1982. The Control of Resources. Oxford, England: Basil Blackwell; Cambridge, Mass.: Harvard University Press; New Delhi, India: Oxford University Press. . 1989. "The Allocation of Hunger." Walras-Bowley Lecture, Econometric Society. .1990. "The Environment as a Commodity." Oxford Review of Economic Policy 6, no. 1: 51-67. (To be republished in 1991 in A. Stevenson and D. Vines, eds., Information Strategy and Public Policy. Oxford, England: Basil Blackwell.) *- Forthcoming. An Inquiry into Well-Being and Destitution. Dasgupta, Partha, Peter Hammond, and Eric Maskin. 1981. "On Imperfect Information and Optimal Pollution Control." Review of Economic Studies 47, no. 4: 857-60. Dasgupta, Partha, and Geoffrey Heal. 1979. Economic Theory and Exhaustible Resources. Cambridge, England: Cambridge University Press. Dasgupta, Partha, S. Marglin, and Amartya Sen. 1972. Guidelines for Project Evalua- tion. New York: United Nations. Dasgupta, Partha, and Joseph E. Stiglitz. 1981. "Resource Depletion under Technologi- cal Uncertainty." Econometrica 49, no. 1 (January): 85-104. Dreze, Jean, and Nicholas H. Stern. 1987. "The Theory of Cost-Benefit Analysis." In A. J. Auerbach and M. Feldstein, eds., Handbook of Public Economics, vol. 2. Amster- dam: North-Holland. Elster, Jon. 1989. The Cement of Society: A Study of Social Order. Cambridge, England: Cambridge University Press. Ehrlich, Paul, A. Ehrlich, and J. Holdren. 1977. Ecoscience: Population, Resources and the Environment. San Francisco: Freeman. El Serafy, Salah, and Ernst Lutz. 1989: "Introduction." In Y. J. Ahmad, S. El Serafy, and E. Lutz, eds., Environmental Accounting for Sustainable Development. Washing- ton, D. C.: World Bank. Ensminger, Jean. Forthcoming. "Co-opting the Elders: The Political Economy of State Incorporation in Africa." American Anthropologist. Epstein, T. Scarlett. 1967. "Productive Efficiency and Customary Systems of Rewards in Rural South India." In R. Firth, ed., Themes in Economic Anthropology. London: Tavistock Publications. Farrell, Joseph. 1987. "Information and the Coase Theorem." Journal of Economic Perspective 1, no. 2: 113-30. Dasgupta and Maler 129 Feder, Ernest. 1977. "Agribusiness and the Elimination of Latin America's Rural Prole- tariat." World Development 5: 559-71. . 1979. "Agricultural Resources in Underdeveloped Countries: Competition between Man and Animal." Economic and Political Weekly 14, nos. 30-32: 1345-66. Feder, Gershon, and Raymond Noronha. 1987. "Land Rights Systems and Agricultural Development in Sub-Saharan Africa." World Bank Research Observer 2, no. 2: 143- 70. Fernandes, W., and G. Menon. 1987. Tribal Women and the Forest Economy. New Delhi, India: Indian Social Institute. Fisher, Anthony C., and W. Michael Hanemann. 1986. "Option Value and the Extinc- tion of Species." Advances in Applied Micro-economics 4. Food and Agricultural Oiganization. 1987. Fifth World Food Survey. Rome. Fudenberg, Drew, and Eric Maskin. 1986. "The Folk Theorem in Repeated Games with Discounting and with Incomplete Information." Econometrica 54, no. 3: 533-54. Goody, Jack, ed. 1973. The Character of Kinship. Cambridge, England: Cambridge UJniversity Press. Hahn, Frank H. 1971. "Equilibrium with Transaction Costs." Econometrica 39, no. 3: 417-40. Hardin, Garrett. 1968. "The Tragedy of the Commons." Science 162: 1243-48. Hecht, Suzanna. 1985. "Environment, Development and Politics: Capital Accumulation and the Livestock Sector in Eastern Amazonia." World Development 13, no. 6: 663- 84. Hecht, Suzanna, A. B. Anderson, and P. May. 1988. "The Subsidy from Nature: Shift- ing Cultivation, Successional Palm Forests, and Rural Development." Human Organi- zation 47, no. 1: 25-35. Henry, Claude. 1974. "Investment Decisions under Uncertainty: The Irreversibility Effect." American Economic Review 64, no. 4: 1006-12. Iliffe, John. 1987. The African Poor. Cambridge, England: Cambridge University Press. Jodha, N. S. 1980. "The Process of Desertification and the Choice of Interventions." Economic and Political Weekly 15. -. 1986. "Common Property Resources and Rural Poor in Dry Regions of India." Economic and Political Weekly (India) 21 (July): 1169-81. Johansson, Per. 1987. The Economic Theory and Measurement of Environmental Bene- fits. Cambridge, England: Cambridge University Press. . 1990. "Valuing Environmental Damage." Oxford Review of Economic Policy 6, no.1: 34-50. Kneese, Alan V. 1988. "Environmental Stress and Political Conflicts: Salinity in the Colorado River." Washington, D. C. Resources for the Future. Processed. Krutilla, John. 1966. "The International Columbia River Treaty: An Economic Evalua- tion." In A. V. Kneese and S. Smith, eds., Water Research. Baltimore, Md.: Johns Hopkins University Press. Kumar, Shiv K., and D. Hotchkiss. 1988. "Consequences of Deforestation for Women's Time Allocation, Agricultural Production, and Nutrition in Hill Areas of Nepal." International Food Research Institute Research Report 69. Lind, Robert, ed. 1982. Discounting for Time and Risk in Energy Policy. Baltimore, Md.: Johns Hopkins University Press. 130 The Environment and Emerging Development Issues Little, Ian M. D., and James A. Mirrlees. 1969. Manual of Industrial Project Analysis. Paris: Organisation for Economic Co-operation and Development. - . 1974. Project Appraisal and Planning for Developing Countries. London: Heinemann. Mahar, Dennis. 1988. Government Policies and Deforestation in Brazil's Amazon Region." World Bank Environment Department Working Paper 7. Washington, D. C. Processed. Maler, Karl-G6ran. 1974. Environmental Economics: A Theoretical Inquiry. Baltimore, Md.: Johns Hopkins University Press. . 1989. "Environmental Resources, Risk and Bayesian Decision Rules." Stock- holm, Sweden: Stockholm School of Economics. Processed. . 1990a. "The Acid Rain Game, 2." Stockholm, Sweden: Stockholm School of Economics. Processed. . 1990b. "International Environmental Problems." Oxford Review of Economic Policy 6, no. 1: 80-108. . 1990c. "The Valuation of Environmental Resources." Stockholm, Sweden: Stockholm School of Economics. Processed. Nerlove, Mark, and Anke Meyer. 1990. "Endogenous Fertility and the Environment: A Parable of Firewood." Philadelphia: University of Pennsylvania. Processed. Nerlove, Mark, Assaf Razin, and Efraim Sadka. 1987. Household and Economy: Wel- fare Economics of Endogenous Fertility. New York: Academic Press. Pearce, D., E. Barbier, and A. Markandya. 1988. "Sustainable Development and Cost- Benefit Analysis." Paper presented at the Canadian Environment Assessment Work- shop on Integrating Economic and Environment Assessment. Polanyi, Karl. 1944. The Great Transformation. New York: Holt, Rinehart and Winston. - . 1977. The Livelihood of Man. New York: Academic Press. Popkin, S. L. 1979. The Rational Peasants: The Political Economy of Rural Society in Vietnam. Berkeley: University of California Press. Portes, Richard. 1971. "Decentralized Planning Procedures and Centrally Planned Econ- omies." American Economic Review 61 (Papers and Proceedings): 422-29. Repetto, Robert. 1988. "Economic Policy Reform for Natural Resource Conservation." World Bank Environment Department Paper 4. Washington, D. C. Processed. Repetto, Robert, and others. 1989. Wasting Assets: Natural Resources in the National Income Account. Washington, D. C.: World Resources Institute. Shapley, Lloyd, and Martin Shubik. 1969. "On the Core of an Economic System under Externalities." American Economic Review 59, no. 4: 678-84. Schecter, M., M. Kim, and L. Golan. 1989. "Valuing a Public Good: Direct and Indirect Valuation Approaches to the Measurement of the Benefits from Pollution Abate- ment." In H. Folmer and E. van lerland, eds., Valuation Methods and Policy Making in Environmental Economics. Amsterdam: Elsevier. Scott, J. 1976. The Moral Economy of the Peasant. New Haven, Conn.: Yale University Press. Solow, Robert M. 1974. "Intergenerational Equity and Exhaustible Resources." Review of Economic Studies 41 (Symposium on the Economics of Exhaustible Resources): 29-45. Dasgupta and Miler 131 1986. "On the Intergenerational Allocation of Exhaustible Resources." Scan- dinavian Journal of Economics 88, no. 2: 141-56. Starrett, David A. 1972. "Fundamental Non-Convexities in the Theory of Externalities." Journal of Economic Theory 4, no. 1:180-99. -. 1973. "A Note on Externalities and the Core." Econometrica 41, no. 1: 179- 84. Stern, Nicholas H. 1989. "The Economics of Development: A Survey." Economic Jour- nal 99, no. 3: 597-685. Tietenberg, T. H. 1980. "Transferable Discharge Permits and the Control of Stationary Source Air Pollution: A Survey and Synthesis." Land Economics 56: 391-416. . 1990. "Economic Instruments for Environmental Regulation." Oxford Review of Economic Policy 6, no. 1: 17-33. Ulph, Alister. 1989. "A Review of Books on Resource and Environmental Economics." Bulletin of Economic Research 42, no. 3: 219-28. Wade, Robert. 1987. "The Management of Common Property Resources: Finding a Cooperative Solution." World Bank Research Observer 2, no. 2: 219-34. . 1988. Village Republics: Economic Conditions for Collective Action in South India. Cambridge, England: Cambridge University Press. Weale, Martin. 1990. "Environmental Statistics and the National Accounts." Cam- bridge, England: University of Cambridge. Processed. Weitzman, Martin L. 1970. "Optimal Growth with Scale-Economies in the Creation of Overhead Capital." Review of Economic Studies 37, no. 4: 555-70. -. 1976. "Welfare Significance of National Product in a Dynamic Economy." Quarterly Journal of Economics 90, no. 1: 156-62. Wiser, W. H. 1936. The Hindu Jaimani System. Lucknow, India: Lucknow Publishing House. PROCE ED ING S OF T HE W O RLD BANK ANN UAL CON FE RENC E ON DE V EL OPM ENT EC ONO MI CS 1 9 9 0 COMMENT ON "THE ENVIRONMENT AND EMERGING DEVELOPMENT ISSUES," BY DASGUPTA AND MALER Mohan Munasinghe This lucid and concise paper by two leading researchers neatly encapsulates and explains the essential relationships between the environment and development from an economic viewpoint. The broad scope of the subject matter has, how- ever, forced the authors to skim over several vital areas that need further explo- ration. Because conferences put limits on discussants as well as authors, I will attempt only to expand on a few key implications arising mainly from the practical implementation of the elegant analytical framework Dasgupta and Maler present. From a broad philosophical viewpoint, I believe that treating environmental issues within an economic framework is the correct approach. Efficient eco- nomic management of all resources (including environmental or natural resources) helps to identify sustainable development strategies that can also comprehensively address other pressing socioeconomic problems, such as pov- erty. Nevertheless, let me pose to the authors a basic question that some non- economists might raise-that is, whether disciplines other than economics might provide useful insights into environmental problems. As we know, the neoclassi- cal economic approach depends on a number of underlying assumptions, such as the self-interest of economic agents and their rationality (Hirschleifer 1985). Researchers in other disciplines, however, have identified nonselfish patterns of individual and group behavior characterized by altruism and cooperation. Sim- ilarly, evidence of preference reversal, status quo bias, and so on demonstrates that human behavior is not always rational. Thus environmental analysis must deal not only with physical resources and human-made capital but also with biological and social behavior systems. Biology, for example, has produced broad underlying theories, such as the "selfish gene," to explain diverse aspects of behavior. Is there then some fruitful synthesis possible of these other disciplin- ary approaches with the standard neoclassical economic worldview? Moving to the specific subject matter, a central feature of the paper is the current value Hamiltonian (CVH). If we accept the economic approach, the CVH is indeed a useful metaphor and organizing principle to identify how environ- mental resources could be identified and valued-thereby improving the quality Mohan Munasinghe is chief of the Environmental Policy and Research Division in the Environment Department of the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 133 Figure 1. Conceptual Framework for National Economic Decisionmaking and Sustainable Environmental Management MACROECONOMY National level hidustry Transport Energy Water Health Agriculture Other |nsecto and theeretwar of the economy Links *Resource -1/ Links requirements | , ., r * ~~~~~~~~~~~~~~~~~~~Resource availability *Water sector s .Waterdegand outputs (disaggregate) *Water sector * National objectives tr constraints - , WATER RESOURCES SECTOR L and constraints Flood and aito Hydro- Potable |risatton Marine sectoral/intermediate level 8 Floodh aNd vgtinSewerage and 4_ _ _ _ control apower e water '|| andother |Water subsector interactions Z POTABLE WATER SUBSECTOR AND PROJECT | Supply Demand 4 Local/micro letel I> _ \ Subsector planning and management * Investment planning * Pricing policy * Operations . Physical controls * Loss optimization . Technological methods * Supply quality optimization * Education and propaganda * And so forth * And so forth Munasinghe 135 of economic decisionmaking. In brief, if one had a set of optimal shadow prices, perfect foreknowledge, and so on, one could arrive at optimal decisions. But we live in an imperfect world, and this raises formidable problems in practically implementing the conceptual model described by the authors. For example, many countries are unable to calculate even conventional net national product (NNP); they basically rely on gross product figures that do not reflect the appro- priate depreciation of physical infrastructure and capital goods (Lutz and Munasinghe, forthcoming). Clearly, the priority in such cases is to estimate conventional NNP, and ascertain whether economic performance is satisfactory on this basis, before seeking more "refined" corrections of national product to account for degradation of "environmental capital." Although Dasgupta and Maler provide a useful conceptual framework for incorporating environmental resources into economic decisionmaking, their concluding assertion that they have done this in a meaningful way is not justified because they do not adequately explore the practical implementation of this conceptual framework. In section II they do recognize the difficulty of applying their methodology broadly, and they prudently advocate a piecemeal approach. Nevertheless, a discussion of specific steps for improving the current way of doing things would have been helpful. For example, the authors state that "the planning exercise generates the entire set of intertemporal accounting prices." Many countries do not follow this process in a meaningful sense, and it is not clear whether the authors are suggesting a move in the direction of more central- ized economic planning. If they are, we should recall the very severe environ- mental degradation now being revealed in Eastern Europe. An appeal to real NNP that incorporates depletion of environmental resources is pedagogically attractive and correct. The paper could have helped the practi- tioner, however, by making clearer how one gets to this measure. There is some ambiguity about whether we should go straight to this measure using accounting (shadow) prices or whether we should correct conventional NNP by incorporat- ing the net "nature" product. An intermediate step would be to start with the latter on the way to the former. Another issue is whether to estimate these measures in currency units or as indexes. The authors seem to agree with the consensus in the U.N. System of National Accounts on the use of satellite environmental accounts to supplement the usual national income accounts. However, even delineating meaningful physical indicators of basic environmen- tal resources, such as forests and water resources, remains elusive but is required before we address the more difficult problem of valuing such assets. Dasgupta and Maler demonstrate the elegance of the CVH approach well in their paper, but the degree of aggregation in the presentation tends to understate the formidable difficulties at more detailed levels-when we deal with disaggre- gated vectors of consumption goods, C, and asset stocks, S. To see the point more clearly, consider figure 1, a broad conceptual schema of the decisionmak- ing framework and structure of a typical economy. The vertical bar on the right reminds us that the environment is a continuum that cuts across these often 136 Comment arbitrary institutional structures. At all levels of the hierarchy in the figure, the World Bank is seeking ways of practically incorporating environmental effects into its mainstream macroeconomic, sectoral, and project analyses, but the move from concepts to real-world applications is far from straightforward. There is also a global, transnational level above the national one, which I discuss below. The application of the authors' approach at the highest level in the national hierarchy (the multisectored macroeconomy) has been explored earlier in the discussion of corrected NNP. But in the World Bank's own research and policy work, we have to start with much more mundane problems. It would be helpful to have Dasgupta and Maler's views on which issues of environmental resource degradation could be effectively incorporated into, for example, the intersec- toral resource allocation process or public expenditure reviews that we often carry out at the macroeconomic level in collaboration with finance or planning ministries. Another pressing set of questions focuses on how natural resource management issues interact with conventional macroeconomic policy concerns such as structural adjustment, stabilization measures, and trade liberalization. If we move down to the intermediate (sectoral) level, much of the problem lies in the contrast between the holistic nature of environmental issues and the fragmented way in which human societies and institutions are structured. A typical case is the water resource sector depicted in figure 1. Policy analysis of water resources should look at all sources and uses of water within a country (and if necessary, across national borders as well). Nevertheless, the data are invariably fragmented among different sectors dealing with irrigation, pipeborne water supply and sewerage, hydropower, navigation, and so on. The cross-cutting nature of environmental analysis will help to develop a comprehen- sive water resources strategy, but even then policy application often becomes less effective because analysts and decisionmakers must work with many unco- ordinated implementing agencies. Focusing finally on the lowest (subsectoral-project-microeconomic) level, the ubiquitous valuation problem has become particularly pressing in the World Bank's country and lending work. The World Bank's recent environmental assessment policy has elevated environmental impact analysis (EIA) to the same level of importance as the more traditional economic, financial, and technical analyses of World Bank projects. In this context, would the authors advise us to seek the "holy grail" of valuing all environmental impacts and incorporating them into a single economic cost-benefit analysis (cBA)? Or would they recog- nize that some impacts cannot be valued meaningfully and instead use (for example) multicriteria objective functions to trade off economically quantifiable and nonquantifiable consequences of projects? It is likely that future World Bank project evaluations will uncover difficult cases in which the EIA-oriented analysis and conventional CBA suggest opposite conclusions regarding the proj- ect decision. Thus our own ongoing research program seeks to push the use of CBAs as far as possible before falling back on multicriteria objective functions. Munasinghe 137 Even if a fully satisfactory methodology is not developed for trading off high economic returns against significant nonquantifiable environmental exter- nalities, the exploration of these issues will help the World Bank redesign poli- cies and projects more effectively in order to reconcile such conflicting goals. Let us now turn to some of the stimulating issues concerning the conceptual treatment of resource stocks raised in this paper. The authors' emphasis on the sad neglect of environmental issues in the development economics literature should be applauded. One could make a more general case that until recently, economics has tended to ignore the fundamental fact that economic systems are integrally embedded in a broader ecological system whose carrying capacity is not necessarily infinite. Another issue of interest is raised toward the end of the authors' remarks on environmental resources and their physical characteristics. Here the authors make a tantalizing reference to the complementarity of environmental and other productive inputs for the poor in developing countries, which may lead to more inequitable impacts of environmental degradation. It is not clear, however, whether Dasgupta and Maler are implying that there is much greater substi- tutability between (for example) human-made capital and environmental resources in the industrialized countries. There are many counterexamples. Fur- thermore, in many instances there may be a threshold of irreversibility (in both developed and developing countries) beyond which it would not be wise to deplete some environmental resources. The life support services of the environ- ment (for example, those provided by the stratospheric ozone layer) clearly fall into this category. In this context, the paper offers an uncharacteristically ungenerous interpretation of Pearce, Barbier, and Markandya (1988). In my own understanding of their writings, the necessary condition for sustainable development and constancy of natural capital stock does not imply nonuse. Although some substitution is feasible, it is the depletion of nonsubstitutable or indispensable environmental resources below sustainable levels that should be avoided. Another point I would note is that in their discussion of estimating a broader measure of NNP, the authors make an abrupt switch from resource depletion to the somewhat different issue of environmental protection expenditures. I should clarify that El Serafy and Lutz (1989), whom the authors cite, refer to an approach that proposes to treat current environmental restoration costs as inter- mediate expenditures (that do not contribute to sustainable NNP). El Serafy and Lutz did not recommend the mixing of capital and current expenditures for environmental protection. The paper also draws an important distinction between problems of local and global commons. The authors make the valuable point that traditional societies have evolved effective means of using local shared resources in a sustainable manner, but these arrangements frequently break down in the process of eco- nomic change because of modernization and externally imposed changes, such as government patronage, distorted pricing or incentives, and population pres- 138 Comment sures. The distributional impact of local environmental degradation also is likely to be severe. The foregoing discussion inevitably raises the question of whose current value Hamiltonian is to be maximized-a point the authors could have discussed further. This can be illustrated with a simplified static analysis at the national, global level that has practical policy implications (Munasinghe 1990). In figure 2, Y represents the net output of productive economic activity, as a function of some resource input (say, energy), without accounting for environmental impacts. Because of policy distortions (for example, subsidized prices), the point of operation in many developing countries might be A, where the resource is being used wastefully. Therefore, without invoking any environmental consid- erations, both economic efficiency and resource intensity (that is, energy effi- ciency) could be improved by moving from A to B. One example of such a move might be to improve energy end-use efficiency or to reduce energy supply system losses. Now consider the curve ECNQ, which represents easily quantifiable national environmental economic costs associated with resource use. These might include air-pollution-related health costs of a coal power plant or the costs of environ- mental protection equipment (such as scrubbers and electrostatic precipitators to reduce noxious gas and particulate emissions) installed at such a plant, or the costs of resettling populations displaced from a dam site. A new net output curve Figure 2. Net Output, Resource Use, and Environmental Cost ITB 0 EN z Resource use (energy) Note: YENQ = Y- ECNMTQ. YEN7 = Y-ECNT+. Y net output (uncorrected); FE= net output (environmentally corrected); T = net output (technologically advanced); EC = environmental costs. Subscripts are defined as follows: NQ national easily quantifiable environmental costs; AT = national total environmental costs; G = global. Munasinghe 139 is defined by YENQ = Y - ECNQ. The maximum of this curve at C lies to the left of B, implying less use of (now more costly) energy. Next, the curve ECNT shows total national environmental costs (both easily quantifiable and difficult-to-quantify negative impacts). With this further cor- rection, the corresponding net output curve is YENT = Y - ECNT. The maxi- mum has shifted to D, and the optimal energy consumption level has also declined. Finally, ECG represents global environmental costs of energy use (including national impacts), and YEG = Y- ECG is the correspondingly corrected net output, which implies an even lower level of optimal energy use. The curve YT shows net output for a technologically advanced future society that has achieved a much lower resource intensity of production. The foregoing discussion illustrates the crucial policy dilemma for already poor developing countries: how to reconcile responsible stewardship of the environment with development goals and the elimination of poverty (which will require increased use of energy and raw materials). Thus in figure 2 all countries (including the poorest) would readily adopt measures that will lead to shift 1 (shifts are marked by a circled numeral in the figure), which simultaneously and unambiguously provides both economic efficiency and environmental gains. Most developing countries are indicating increasing willingness to undertake shift 2. Implementing shift 3, however, will definitely mean that many develop- ing countries will have to cross a "pain threshold" as other pressing socio- economic needs compete against the costs of mitigating adverse environmental impacts that are poorly defined or difficult to value. Shift 4 implies optimization of a global CVH. Because benefits would accrue mainly to other countries, this would hardly appeal to a poor developing nation unless concessional external financing was made available. Let us extend our inquiry by exploring some issues relating to better manage- ment of the global commons, which the authors single out in their concluding paragraphs as central to the policy process in the immediate future. Although scientific analysis has provided only broad and rather uncertain predictions about potential global warming, it would be prudent for humankind to buy an insurance policy in the form of actions to mitigate greenhouse gas emissions. An attractive insurance premium for the developing world would be a set of inex- pensive measures that could address a range of national and global environmen- tal issues without hampering development efforts. Several proposals have been made recently for setting up a global environmental fund to help developing countries in their efforts, and some industrial countries already have indicated their willingness to contribute. Currently discussions are going on among world bodies and governments to define effective criteria and mechanisms for both generating and disbursing funds from such a facility. Although a broad, work- able agreement will not be easy to reach, global financing issues might be analyzed and resolved through a trade-off involving several criteria: afford- ability, additionality, and concessionality; fairness and equity; and economic efficiency. 140 Comment Consider, in the context of the affordability criterion, the energy sector, which is the major contributor to potential global warming. The developing world cannot afford to finance its normal power development needs of about $100 billion per year for the next ten years (billion equals 1,000 million). Even though economically viable energy management options that lower overall costs could reduce this burden significantly, some growth in the use of energy by developing countries is inevitable. The adoption of pollution abatement policies that further increase energy costs-thereby crossing the pain threshold-will not be feasible without external funding on concessional terms. Furthermore, such financial assistance should be additional to existing conventional aid received by developing countries. The fairness criterion arises because of sharp disparities in the global distribu- tion of income. The per capita gross national product (GNP) of low-income economies (with half of the world's population) averaged US$290 in 1987, or less than one-sixtieth of the U.S. value ($18,530). In the two largest developing countries, China and India, per capita GNP was $290 and $300, respectively. Correspondingly, the U.S. per capita energy consumption of 7,265 kilograms of oil equivalent in 1987 was fifteen and thirty-five times greater than that of China and India, respectively. A more historically linked dimension of equity is noted in the recent Brundt- land Commission report (World Commission on Environment and Develop- ment 1987), which argues that past growth in the industrial countries empha- sized needs rather than resource limitations. Such growth exhausted a relatively high share of global resources (including physical resources consumed in pro- ductive activity as well as the waste-absorbing capability of the global ecosys- tem). For example, the developed countries accounted for more than 80 percent of fossil-fuel-related cumulative carbon dioxide emissions worldwide during 1950-86 alone, and on a per capita basis, these countries were responsible for more than eleven times as much total cumulative carbon dioxide emissions as the developing countries. Clearly, any reasonable growth scenario for develop- ing nations that followed the same material-intensive path as the industrialized world would result in unacceptably high levels of greenhouse gas accumulation as well as more general depletion of natural resources. Indeed, this resource- intensive historical growth path suggests that the developed countries owe an environmental debt to the larger global community. This approach could help to determine how the remaining finite global resources may be equitably shared and used sustainably. The authors touch on this point in their summary (when they mention the initial allocation of national rights to use the remaining global commons), but they do not pursue the international equity implications far enough. The final consideration in any global agreement on the environment is eco- nomic efficiency. If global environmental costs of human activity can be quan- tified, the "polluter pays" principle may be applied to generate revenues, which will open up a rich array of economic options. For example, if total emission Munasinghe 141 limits are established for carbon dioxide, then market mechanisms such as trad- ing in emission permits among (and within) nations could be used to increase efficiency, as Dasgupta and Maler note. We conclude that pressures to address environmental issues (especially global ones) place a severe burden on developing countries. Even with additional exter- nal assistance, the feasible near-term response of the developing countries can- not extend much beyond sound economic management of natural resources that is consistent with their domestic developmental and environmental goals. But the developed countries could facilitate this process by providing concessional financial and technical assistance that the developing countries need today. They could also show leadership now, by trading off some growth for improved environmental quality and by pioneering the use of advanced technologies that will usher in the less material-intensive econormies of the future (curve YT in figure 2). Such a successful demonstration would help convince the developing countries to undertake more costly abatement measures and cross this pain threshold early in the twenty-first century. REFERENCES El Serafy, Salah, and Ernst Lutz. 1989. "Environmental and Resource Accounting: An Overview." In Yusuf J. Ahmad, Salah El Serafy, and Ernst Lutz, eds., Environmental Accounting for Sustainable Development. A United Nations Environment Programme-World Bank Symposium. Washington, D.C.: World Bank. Hirschleifer, Jack. 1985. "The Expanding Domain of Economics." American Economic Review 75 (December): 53-68. Lutz, Ernst, and Mohan Munasinghe. Forthcoming. "Accounting for the Environment." Finance and Development. Munasinghe, Mohan. 1990. Energy Analysis and Policy. London: Butterworths Press. Pearce, David, Edward Barbier, and Anil Markandya. 1988. "Sustainable Development and Cost-Benefit Analysis." Paper presented at the Canadian Environment Assessment Workshop on Integrating Economic and Environment Assessment. World Commission on Environment and Development. 1987. Our Common Future. London: Oxford University Press. PR O CE E DING S OF THE W ORL D BANK ANNUA L C ONFE RE N CE ON D E VE LOPM E NT E C O N OM ICS 1 9 9 0 COMMENT ON "THE ENVIRONMENT AND EMERGING DEVELOPMENT ISSUES," BY DASGUPTA AND MALER John Whalley Partha Dasgupta and Karl-Goran Maler have produced a useful and compact paper, for which they are to be complimented. They point out that the conven- tional development economics literature makes little explicit reference to envi- ronmental issues, and they summarize what literature there is. They survey the distinguishing characteristics of environmental resources, and they discuss deter- minants and measurement of well-being, taking into account environmental factors, market failures and the environment, and global and local "commons" issues. They conclude with some comments on what they see as needed direc- tions for future research. Useful though the paper is, it might have been better if it had begun with a statement of issues and moved on to discuss them, drawing on the academic literature as needed, rather than constraining itself to the literature and trying to extract from it what might be relevant to environmental issues. I stress putting the issues foremost because I think it is important to recognize that many policy- makers in developing countries are now struggling with two sources of pressure to take environmental considerations into account. The first is the pressure from the developed world to cut emissions, slow deforestation, and so on, because of concerns over global warming. The second is the growing pressure from their own domestic groups to pursue industrialization and development, as the devel- oped world has done, and yet avoid the relative neglect of the environment. Researchers, academic literature, and even the World Bank do not seem at present to have much to say about these matters. The problem of global warming needs to be looked at carefully. Although there is a general scientific consensus on the likelihood of global warming, research and opinion remain ambiguous on its consequences. Opinion on conse- quences seems to point to a "2 x CO2 scenario," in which carbon dioxide concentrations are predicted to double from their current levels of about 350 parts per million (ppm) to 700 ppm by the years 2030 to 2075. This doubling probably will increase global mean temperatures by between 1.5 and 4.0 degrees Celsius; it will raise sea levels by 20 centimeters to 1 meter; it will increase the John Whalley is professor of economics and director of the Center for the Study of International Economic Relations at the University of Western Ontario, London, Ontario, Canada. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 143 144 Comment incidence of flooding, typhoons, hurricanes, and other extreme climatic events; and it will have microclimatic effects (such as local desertification or increased rainfall) that are hard to pinpoint accurately today. Pressures are growing in the developed world for some form of policy response to global warming, perhaps through a carbon tax or tradable permits scheme. Tax rates currently being discussed are large (100 percent), and devel- oped countries want developing countries included. Developed countries are expressing similar concerns over deforestation, especially in Brazil, but also more and more in a larger number of developing countries. Developing countries have argued for special treatment on policy responses to global warming on the grounds that they have not been the primary source of greenhouse emissions and that their full inclusion in schemes for carbon taxes or tradable emission permits will truncate their development. Questions of causation and responsibility aside, recent research is pointing to at least four strong conclusions on the problems and effects of these global environmental policy approaches. First, a carbon tax (or tradable permit scheme) will not prevent a 2 x CO2 scenario but will only delay it, perhaps by only ten to fifteen years. Second, what happens to the revenues from such a tax is of enormous relevance for developing countries. Current estimates suggest that the difference to developing countries between a national-based tax scheme and a tax collected by an international agency with revenues redistributed among countries on an equal per capita basis could be approximately $150 billion per year or, say, four times current aid flows (billion equals 1,000 million). A third research finding is that exemption schemes such as an annual emission level per capita may cause energy-intensive manufacturing activity to relocate to the developing world. Fourth, the free-rider and enforcement problems with such schemes are extremely large, pointing to a long-standing concern among developing countries that other policy measures (such as threats to use trade measures or conditionality in lending programs) will be employed to force devel- oping countries to reduce emissions. The other major issue that deserves greater recognition is the sketchiness of our knowledge about the interplay between developmental objectives on the one hand and environmental concerns on the other. Developing-country policy- makers and citizens, enlightened by information now becoming available on environmental degradation in Eastern Europe and the Soviet Union, are asking whether they want to (or indeed can) develop in the same way as the developed countries, and what the options are. Here the guidance of experience and theory is even more hazy than it is on the consequences of global warming. But the concerns are, if anything, stronger. The economies that have experienced strong growth in recent years (the Republic of Korea, Taiwan, the countries that belong to the Association of Southeast Asian Nations, and now Turkey and Mexico) have all gone through an energy-intensive heavy industrialization phase. To my knowledge, no prece- Whalley 14S dent yet shows clearly how this stage can be circumvented. Moreover, when environmental concerns have come to the fore in developing countries-such as proposals for automobile emission controls in Mexico City, or for preventing soil erosion at the village level in India-they have not usually won out over developmental concerns. In these instances developing countries have judged the costs of environmental action and the constraining effects on growth too severe. What we know about this question is limited, but the dominant hypothesis for now, at least, seems to be that "green development" is both unproven and likely very costly. To conclude, let me return to Dasgupta and Mdler's paper and reiterate my main point. It is helpful to the environmental policy process in developing countries to survey the issues as framed by the development economics litera- ture. But my sense is that it would be even more helpful for us to directly address the issues that development policymakers in these countries see as crucial. PROCE ED ING S OF THE W O RLD BANK ANN U AL CON FE RE NC E ON D EVE LO PM ENT EC ONO MI CS 1 9 9 0 FLOOR DISCUSSION OF THE DASGUPTA-MALER PAPER The discussion started with a participant agreeing with Dasgupta and Maler's focus on a correctly defined and calculated net national product (NNP) that includes environmental resource use. The participant argued that other comple- mentary measures of welfare may be justified in arriving at such a comprehen- sive NNP concept, but they should not take the focus away from the correct NNP measure. The participant then commented on the authors' discussion of defen- sive expenditures in the conference paper and suggested that they should be included in NNP only to the extent that there is an investment component to them. Another participant observed that a fair degree of equality is necessary within a decisionmaking group for the management of local common property resources to work well. He wondered whether the authors agreed that there would be considerable pressure for inequality to develop as one moves to market-oriented private property resources. Responding to the comments by John Whalley (discussant), Partha Dasgupta said that Karl-Goran Maler and he had emphasized the literature in their paper precisely because they wanted to highlight the fact that environmental resources have unfortunately not been a part of standard economic thinking until recently. One of the first steps in tackling these issues is the valuation of environmental resources. This is easier to do in some cases (for example, productivity losses from soil erosion), and much harder in others (for example, the value of a lost species). The obvious measurement difficulties should not deter us from pursu- ing measurement questions, nor for that reason should we separate out and treat environmental resources differently from other goods and services. Keeping environmental resources separate is really a bad tactic, because then the onus is always on the preserver to show that it is worth taking a closer look at such resources. And the fact that intertemporal concerns are important is not some- thing that is special to the environment; it applies to anything in an economy that survives for any length of time. In further response to Whalley, Dasgupta said that the paper had deliberately gone beyond issues such as global warming and international negotiations, important though they are, to highlight the badly neglected links between desti- tution and environmental degradation in developing countries. Humdrum This session was chaired by Francis Colaco, principal economic adviser, Office of the Vice President for Sector Policy and Research, the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 147 148 Floor Discussion of Dasgupta-Miler Paper resources such as firewood and water that children and women collect on a day- to-day basis do not get the same attention as global warming. Answering Mohan Munasinghe's (discussant) question about whose Hamilto- nian the authors would seek to optimize in their framework, Dasgupta clarified that the use of a Hamiltonian in their paper was no different from the standard valuation of an investment project or the shadow wage rate, as, for example, in project evaluation. Munasinghe was quite right that there is a difficult valuation problem for depreciation of even standard industrial infrastructure, let alone natural resources-mostly we see gross national product (GNP) figures, not NNP figures, for many developing countries. Nevertheless, there is the view that the NNP growth rate should reflect changes in the well-being of a society. The paper sought to emphasize that if it is a welfare index one is trying to measure, then depreciation of all of society's capital assets must be incorporated. Whether it is an exhaustible resource or a renewable resource, it is still a resource, and the rate of change in the stock, however measured, should be valued and deducted from the conventional NNP growth rate. Referring to the adjusted NNP calculation embodied in the authors' Hamilto- nian approach, a participant raised the danger of double-counting the value of environmental resources. The question was whether one should rely on the idea that natural resource use is another element of investment, possibly negative investment, that should enter into the measurement of the NNP at appropriate shadow prices, or whether the resource depletion should also go directly into the measure of the NNP adjustment for environmental effects. The participant thought that one could not use both approaches simultaneously. He felt it was much better to focus on the goods and services that are produced, among which there are some negative quantities, as, for example, deforestation and erosion. What really matters then is to find some way of shadow pricing the goods appropriately. Although also difficult, the shadow pricing approach is likely to be still somewhat easier than the concept of having to create a quantitative notion of natural environmental stocks and then finding suitable prices to apply to them to calculate their depreciation. The same participant also pointed out that adjusting for environmental resource use may not necessarily reduce the growth rate, as the paper main- tained. If one begins from the point when environmental degradation starts, then the following year there will indeed be a reduction in the environmental stock, bringing the growth rate down compared with the conventional measure. But that is probably not going to be the case in many countries. It is likely that degradation has been taking place for a long time and that one has to change the value of the stock for previous years as well, and the growth rate may not then change. A World Bank participant asked the authors if, in incorporating natural resources into welfare measures, they thought of them basically as substitutes for man-made capital in the traditional neoclassical sense, or as complementary factors. Floor Discussion of Dasgupta-Mdler Paper 149 Another participant asked whether the authors had any empirical experience of the priority structuring of production functions referred to by Maler, particu- larly in a democratic political system. The participant thought that setting prior- ities was a complex and highly ambitious task, and the problem lay in who determines the priorities and who implements the policies arising from them. He cited a European point of view that held gasoline pricing in the United States to be irrational, with social costs exceeding private costs. Another participant noted that he found the paper's treatment of economic development puzzling. He associated development with expanding specializa- tion and market opportunities for production, trade, and use of resources in ways in which they are worth more. It also meant destruction of local, tradi- tional institutions that the paper found indispensable for the management of local commons and the environmental resources that are inputs in local produc- tion. So, if one is to take the paper's approach seriously, what is one to make of development? Responding to the questions, Dasgupta replied that he hoped it was clear from the paper that Maler and he had taken pains to avoid talking about grand strategies of development in the context of the environment. This was done deliberately; the more one thinks of environmental resources as humdrum com- modities, the more likely it is that they will be preserved, because we will then treat them just as other goods and services and factor them into our calculations of future costs and benefits. Clearly, some environmental goods will be ignored because of the difficulty of calculating their accounting prices. In response to the specific question about the two approaches to incorporating environmental resources in NNP and double-counting in the current value Ham- iltonian, Dasgupta clarified that their discussion of NNP notwithstanding, they were also emphasizing the accounting prices of goods and services that should be incorporated into production and investment decisions. However, Dasgupta did not think that the current value Hamiltonian involved double-counting: the Hamiltonian measures the flow of well-being in a given year plus the social value of the change in the vector of natural assets that year, the latter picking up the fact that the assets are a potential source of well-being in the future. In measur- ing the current flow of well-being, one needs to capture the notion that existing stocks may also provide current well-being directly. If the stocks do not provide any current flow of well-being, then this component of current well-being would, of course, disappear. Dasgupta said that he realized that in standard capital theory, capital stocks are valued for their indirect effects. There is not much delight in looking at a factory; the delight comes from one's expectations of the factory's future production. But environmental stocks may also directly yield current well-being. On Munasinghe's question about establishing thresholds to limit resource exploitation, Miler suggested that one could make a case for them on the basis of uncertainty about the underlying ecological, hydrological, and geochemical processes, and uncertainty about the value of these resources and their account- 150 Floor Discussion of Dasgupta-Maler Paper ing prices. Responding to Whalley's comments about the importance of global warming and the potential policy choices facing developing countries, Maler said that the main problem is to create a system that will benefit all member countries for being part of the system. The acid rain problem in Europe is similar. Maler felt that a system with tradable emission permits could create the right kind of incentives, if the initial rights are distributed in an appropriate way. Munasinghe and another participant had raised the question of substi- tutability and complementarity of environmental capital with man-made capi- tal. Maler responded to this by noting that the only thing the paper mentioned was that at the household level, environmental resources are quite often comple- ments to other inputs in the household production function. However, on a larger scale, it was an empirical question. Regarding the production function approach and setting priorities, Maler answered that the point of using production functions was to enable the analyst to come up with estimates of the accounting prices for environmental resources. Then these accounting prices could be used for setting priorities. Maler cited the example of Sweden, where this approach had been used to preserve certain natural areas and for designing environmental policy instruments. On the question of the treatment of defensive expenditure, a World Bank participant adduced that if one is undertaking defensive expenditure that is entirely used up in the year and has no investment component in it, then one should net it out, as maintained by El Serafy and Lutz (cited by Dasgupta and Maler). However, a double-entry procedure might be more correct-that is, to include the defensive expenditure and also to factor in the value of environmen- tal damages. This procedure may be more correct generally, because if the defensive expenditure has some investment component and therefore longer- term effects, then the damage value will not be identically equal to it. The same World Bank participant echoed the earlier question about dealing with changes in stocks arising out of changes in the state of knowledge. How does one measure changes in GNP between two years when there was a discovery of a gigantic oil field in a country? Presumably, the answer is that 1987's GNP as seen from the viewpoint of 1988 is much higher than 1987's GNP as seen from the viewpoint of 1987. But he was not sure what that said about welfare or how usable these concepts are. He thought there was a problem here. As a parallel to the discussion of the current flow of well-being directly attrib- utable to a resource stock, Munasinghe asked whether the level of current consumption could also affect the valuation of environmental resources and changes in them. In other words, if one was starving in the Sahara, one would put a somewhat different value on, say, global warming than if one was living in an affluent suburb of Washington or New York. So there is an interaction also between the current level of consumption and the value one might put on envi- ronmental stocks. In concluding, Dasgupta returned to the question raised earlier about the treatment of development in the paper and its focus on traditional mechanisms Floor Discussion of Dasgupta-Miler Paper 151 to manage common property resources. He did not want to give the impression that the paper espoused a hands-off approach and the preservation of such traditional societies at the cost of integrating them into the market and denying them the benefits of increased economic specialization. Instead, the paper had tried to emphasize that it is precisely during development that traditional, desti- tute groups need protection to avert some of the greatest social tragedies. If common property resources are sharply eroded, it is often the poorest of the poor who bear the brunt of the resulting destitution. If fuelwood is harder to gather, it is the poor who are not able to buy it in the market, and the additional burden of more distant collection falls on the women and the children. Thus, the paper sought to raise the issue of how a possibly long and painful transition from destitution can be eased. Francis Colaco (chair) thanked the speakers and the participants at the confer- ence and closed the session. PROCE ED ING O OF T HE W OR L D BANK ANNU AL CON FE RENC E ON DE V E LO PM ENT EC ONO M ICS 1 9 9 0 Regional Sustainable Development and Natural Resource Use Peter Nijkamp, C. J. M. van den Bergh, and Frits J. Soeteman The concept of sustainable development has been discussed intensively at a global level in the past few years. This paper investigates sustainable development in a practical planning context by introducing and outlining the notion of regional sustainable devel- opment (RSD)-a translation and operationalization of sustainable development on a regional scale. Implicit in RSD is that it should always be compatible with global sustainability and that RSD of all regions of a spatial system implies sustainable develop- ment for the system as a whole. From a planning viewpoint, an identification of critical success factors (csFs) is of crucial importance for RSD. A CsFis a necessary condition for balanced regional development that can be guided by policy intervention. In most cases, the notion of sustainable resource use (SRu) appears to provide a practicalframework for identifying a csF, because renewable stocks of natural resources are a key factor for RSD in most countries. CSFs may usually be found by investigating the regional supply of natural resources and using their features (exhaustible, renewable, accessible, multi- functional, and so on) to identify measurable indicators for RSD. The paper discusses and critically evaluates RSD models with regard to their design, specification, and use. In addition, it presents three case studies that may help clarify the notions of RSD, CSFs, and SRU and demonstrate their operational character in different regions: the Peel area in the Netherlands, the Sporades Islands in Greece, and rural land in Botswana. The paper concludes with retrospective review and a prospective exploration of important research questions. 1. RECENT ENVIRONMENTAL ISSUES The 1970s marked the beginning of the broad awareness of actual and poten- tial conflicts between economic progress in production, consumption, and tech- nology and the environment. Since then, the environment has become the sub- ject of intensive research in both developed and developing nations. In various countries, "standard" types of environmental pollution-notably air pollution, water pollution, solid waste, and noise-have been tackled fairly successfully, at least on a local scale. Most abatement policies have been oriented toward envi- Peter Nijkamp is professor of economics in the Faculty of Economic Sciences and Econometrics at the Free University of Amsterdam. Jeroen C. J. M. van den Bergh and Frits J. Soeteman are also with the Faculty of Economic Sciences and Econometrics at the Free University of Amsterdam. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 153 1S4 Regional Sustainable Development and Natural Resource Use ronmental issues of a tangible-often local or regional-nature. Moreover, a large number of regulations have been introduced in the field of industrial pollution, sewage treatment, protection of landscapes, conservation of monu- ments, and so on. As a reading of the literature makes evident, the policy and research focus on urgent environmental issues has been marked by fluctuating patterns over the past few decades. The first wave of massive environmental concern, in the 1970s, mainly called attention to negative externalities and their social costs in the form of environmental pollution, and to resource depletion (notably in the fields of energy resources, minerals, and fisheries) connected with the industrial structure of our economies. This interest was further intensified by the height- ened public awareness of the deleterious consequences of various specific chemi- cal processes, technologies, and forms of household consumption (for example, smoking and automobile use). In the past decade, however, a shift has taken place from partial environmen- tal analysis to a focus on, first, global-regional interactions and, second, economic-ecological interactions of environmental problems. There is an increasing recognition that a more coherent approach is needed in dealing with environmental issues, including land use, urban development, common property resources, spatial inequality, and intergenerational equity. Global-level environmental interactions have been marked by two features. One is the globalization of environmental impacts. The other is the regionaliza- tion of an often hardly visible but quite substantial decline in the quality of global environmental resources. The global effects of environmental decay-reflected among other things in alarming phenomena such as ozonization, desertification, and acid rain-came in most cases as scientific surprises and were hardly addressed in actual poli- cymaking until recently. But, especially since the publication of the report of the Brundtland Commission (WCED [World Commission on Environment and Development] 1987), we have witnessed a significant increase in interest in global environmental problems. In addition to global issues, a large number of small-scale and marginal changes at the local or regional level have clear global dimensions. A wide variety of incremental pollution phenomena (for example, persistent micro- pollutants), seemingly hardly important by themselves, have severe accumula- tive and synergetic environmental impacts. All such impacts call for more coher- ent local environmental planning. In this respect, local land use is becoming one focal point of concern for policymakers and researchers. The spatial issue can thus be examined from the viewpoint of local trends causing global effects and global trends leading to local effects. An illustration of the first type of problem is poor natural resource management in some coun- tries, which threatens both the physical basis of these countries and also destroys the vulnerable ecosystem of the planet to an unprecedented extent. Other illus- trations of local-to-global influence are cases of overgrazing and deforestation Nijkamp, van den Bergh, and Soeteman 155 that may lead to wider soil erosion, sedimentation, flooding, and salinization. The second type of problem concerns the local-scale environmental effects that emerge from global trends. Acid rain, erosion, desertification, destruction of the ozone layer, eutrophication, ocean pollution, and resource extraction are taking place on a worldwide scale, but their effects can clearly be observed at a local or regional scale. Thus, the global-regional interdependence of environmental problems confronts us not only with quantitative changes but also with qualita- tive (or structural) long-term changes at all places on earth (see Bartelmus 1986). This calls for a new view of spatial economic-environmental problems and policy issues. The economic-ecological interactions arise in general because concerted socio- economic development requires a compromise between material growth and environmental constraints, including environmental quality and vital natural resources. Although in different countries the conditions under which a bal- anced development may come about will show much variation, the conflicting nature of the above objectives is evident in all countries. Especially in the short run, the conflict between material growth and environmental quality may be rather severe; in the long run, a mitigating effect may emerge, because continued economic growth needs a sound resource base, and structural protection and upgrading of environmental quality in turn presuppose economic growth (see also Nijkamp and Soeteman 1989 and Shefer 1974). Such a coevolutionary development (Norgaard 1984), in both the developed and the developing world, takes for granted that economy and ecology ultimately do not conflict with one another. But such a coevolution-even one that is also based on equitable devel- opment options for present and future generations-does not necessarily require mutual positive spillover effects between the economy and the ecology. It is this latter idea that is more recently echoed in the notion of ecologically sustainable economic development. Given the growing prominence of environmental problems, it is no surprise that sustainable development has become a key catchphrase in economic plan- ning and resource management. However, the interpretation of this concept is still ambiguous, and definitions of it are abundant-ranging from continued economic growth to steady state economies where economic growth mainly serves to compensate for environmental depreciation (a more extensive overview of definitional problems can be found in Archibugi and Nijkamp 1989; van den Bergh and Soeteman 1990; and Pezzey 1989). As a result, even though the idea of sustainable development has dominated most recent discussions on develop- ment and environment, the actual effects of this idea are not yet clearly visible in terms of a drastic reorientation or new departures for socioeconomic develop- ment policy (see also Collard, Pearce, and Ulph 1989). Clearly, balanced strate- gic planning based on merging the principles of economics and ecology appears to be difficult. Only a few examples so far convincingly demonstrate that a concerted reconciliation of seemingly conflicting options-at least in the long run-is not impossible. 156 Regional Sustainable Development and Natural Resource Use According to the Brundtland Report (WCED 1987), the idea of sustainable development reaches far beyond environmental protection, as it means a process of change in which exploitation of resources, direction of investments, orienta- tion of technological development, and institutional changes are made consis- tent with future as well as present needs. Consequently, sustainable develop- ment is not a fixed state of harmony but rather a balanced and adaptive process of change. This would then be characterized by a dynamic Pareto-optimal tra- jectory in which progress in one system-that is, either the economic or the ecological-would not be to the detriment of the other system. Sustainability takes for granted a balance between economic development-all quantitative and qualitative changes in the economy that offer positive contributions to welfare-and ecological sustainability-all quantitative and qualitative environ- mental strategies that serve to improve the quaiity of an ecosystem and hence also have a positive impact on welfare. It is noteworthy that the concept of welfare has to be understood here in a broad sense as the individual or collective utility derived from the availability or use of scarce commodities, including environmental goods, whether such utility attributes can be measured in monetary terms or not (see Nijkamp and Soete- man 1988). Consequently, toxic materials, ionizing radiation, the beauty of the landscape, well-preserved monuments, traffic safety, wholesome food, and availability of shelter all may be regarded in principle as arguments in a welfare function. For example, in the framework of agricultural land use, the welfare gains from agriculture should be measured by income or production generated in the agricultural sector, but they should also incorporate negative effects on the landscape, species diversity, or ecostability (see also section VI). Clearly, various changes in land use patterns may be caused by factors outside the agricultural system itself, such as the climate. The fact that both conventional economic factors and environmental goods may contribute to welfare, and must also be traded off against each other, does not of course imply that in an extreme case one of the two systems might be completely depleted. Both economic and environmental systems need a certain minimum threshold value to survive. Ciriacy-Wantrup (1952) advocated the use of a minimum bequest value in strategic environmental policies, in particular calling for the establishment of safe minimum standards for conservation by avoiding overexploitation of critical zones of the environment by limiting human activities that make it uneconomical to halt or reverse depletion. Thus, the idea of sustainable development requires a careful consideration of sustain- able threshold levels for both economic and environmental systems. It is clear from the above remarks that sustainable development issues are manifesting themselves in various forms. An extremely important form is land use and uses of land-related resources. For example, deforestation in Brazil may be necessary for agriculture or energy supply in a regional economy, but it is extremely detrimental to global ecological stability. Housing construction in Nijkamp, van den Bergh, and Soeteman 157 densely populated areas may be necessary from the viewpoint of a growing population and a decline in family size, but at the same time it may impair the visual beauty of an ecologically vulnerable area. Thus, to a large extent, land use may be regarded as a focal point of sustainable development policies in a spatial setting. This leads to the necessity to specify more precisely the interactions between different resource and land use options in a given area and the spillover effects from, and to, other areas. Such a more local and regional orientation is mandated not only by the character of the economic and environmental interac- tions but also by the spatial orientation of policies concerned with land use. Our study focuses attention on the spatial dimensions of ecologically sustain- able economic development in the context of regional resource use affecting land use. In other words, we discuss the issue of regional sustainable development in relation to land use. Our paper clarifies the intricate relationships between ecology and the economy in land use by bringing together in a structured way relevant recent literature and by elaborating, clarifying, and operationalizing the notion of regional sustainability with particular emphasis on the identification of critical success factors for RSD. II. LAND USE, REGIONAL SUSTAINABLE DEVELOPMENT, AND SUSTAINABLE RESOURCE USE Conflicts between economic progress and environmental sustainability are not exclusively a phenomenon of recent decades. Even in ancient times, Plato was concerned about landscape changes in Attica, as witnessed by his complaint in the Critias that recent developments had turned the environment into "bones of a wasted body . .. richer and softer parts of the soil having fallen away, and the mere skeleton being left" (cited in Clark 1986, p. 8). Changes in land and resource use associated with agriculture, industry, or urban development in many European countries also have affected landscapes (and related environ- mental conditions) in all periods between nomadic cultures and modern indus- trialization. Until World War II, the land use implications of a mixed agri- cultural and industrial society were relatively modest. Postwar land use, however, has been influenced not only by industrial activity in many countries but also by urbanization and recreation. Consequently, land use is rapidly becoming a focal point of environmental-economic research. The broad recognition of resource and land use as an issue for scientific research is quite new in economics. Apart from the Physiocrats, who regarded the productive capacity of the natural environment (mainly land) as the major source of welfare, economic thinkers until recently have rarely paid due atten- tion to land as an important production factor. For example, in classical eco- nomics, capital and labor, not land, were regarded as the main welfare genera- tors. It is interesting to note, however, that the classical economists were aware of the possibility that an economy might stagnate as a result of lack of natural- mainly agricultural-resources. 158 Regional Sustainable Developmentand Natural Resource Use Largely as a result of neoclassical thinking in the post-World War II period, nature was not considered the source of welfare but merely the supplier of raw material subsequently used by labor and capital. Clearly, land did not become irrelevant, but in the words of Randall and Castle (1985), "there seemed no reason to accord land any special treatment that would suggest its role is quite distinct from that of the other factors. Land could safely be subsumed under the broader aggregate of capital" (p. 573). From the viewpoint of land use, there is now a clear need to pay more explicit attention to the spatial scale of environmental phenomena. A main question is how to avoid a "tragedy of the commons" in view of the long-term threats exerted by seemingly inevitable and persistent changes in land use at both a local and a global scale. In principle, in a giobal system, all processes are endo- genously determined, whereas in a mesosystem (for example, a single region), a considerable part of the relevant economic and environmental processes is exog- enously determined. Such external influences do not have a uniform effect on all regions of a global system. For example, global warming of the atmosphere may lead to a rise in total organic production in the world as a whole, but no doubt will lead to different socioeconomic and ecological impacts in various regions of the world. Furthermore, regions of a global system form an open set of mutually interacting areas, so that certain changes in one area have consequences for other areas. For example, Arntzen (1989) has shown that spatial mobility of human activities in an open system of regions-as a result of periods of drought-may induce resource pressure in other areas in a developing country. Thus, at a regional level of analysis, there is much scope for various spatial tradeoffs: interregional, regional-global, and intertemporal. This leads us to the notion of regional sustainable development as a particular type of sustainable development in an open spatial system (see also Kairiukstis 1989). In a spatial setting, RSD refers to both the potential offered by the environment as a resource base for economic growth and to the constraints imposed by the environment on an unlimited growth potential, while taking into consideration the mitigating or reinforcing effects of environmental dangers caused by an open spatial system. If a given region's RSD is independent of the rest of the world, then it is self- sufficient; it may be regarded as a closed system. More interesting and relevant, however, are open regions. In an open multiregional system, the final result for RSD depends, among other things, on the type and volume of goods and services the region offers to the rest of the world. Interregional trade, which in the standard literature is supposed to be determined by comparative cost advantages and scale economies, may have a distorting impact on sustainability in an open spatial system, of course, unless all negative environmental externalities are included as imputed social costs in spatial trade flows. Thus internalization of external costs is a prerequisite for RSD. Aside from assessing regional variation and regional interaction, there is another reason for focusing attention on a regional scale-that is, the fact that a Nijkamp, van den Bergh, and Soeteman 159 regional system is more amenable to policy management, examination, and control than a global system. In this context, planning regions may allow for the attainment of certain planning objectives in the most efficient way (see Paelinck and Nijkamp 1976). A region is a geographical area that meaningfully may be regarded as a coher- ent entity from the viewpoint of description, analysis, administration, planning, or policy. Various types of regions are distinguished in the literature-for exam- ple, homogeneous or functional regions. Functional environmental regions are often based on intensive interactions within particular regions between environ- mental resources such as groundwater, river, and meteorological systems. Homogeneous environmental regions are usually based on the existence of com- mon resources in the area concerned. Another typology distinguishes natural (ecological or environmental) regions, economic regions, and administrative (or political) regions (Paelinck and Nijkamp 1976). Regional development is often critically dependent on the regional supply of resources. Many types of regional economic dependence on the resource base can be distinguished (see van den Bergh and Nijkamp 1990). Some examples follow: * A regional economic system may be directly dependent on the resources in the region. Usually in such cases the dependence is on resources that may serve as essential and cheap productive inputs to economic activities-for example, energy resources. Concomitantly, the environmental capacity as a sink for waste materials and pollution can restrict economic activity. * Some regions may be dependent on export of resources as a main source of income-for example, energy-exporting boom towns in the province of Alberta, Canada. * A regional resource sector may have many effects on other activities in a region, among other things as a result of an increased demand for public services, utilities, and infrastructure; demand for labor, capital, and space; and spinoffs to other private sectors. Thus, the development of a resource sector may generate shifts in sector allocation, income levels and distribution, and exchange rates. From a welfare viewpoint, it makes sense to define RSD as development that ensures that the regional population can attain an acceptable level of welfare- both at present and in the future-and that this regional development is compat- ible with ecological circumstances in the long run while it also tries to accom- plish a globally sustainable development. Clearly, specification of minimum bequest values, implementation of shadow projects, or allowance for limited but given substitutability between productive and environmental capital are special cases of RSD. Also, knowledge transfer or technology transfer for preventive environmental policies may be relevant in this framework. Consequently, RSD has to fulfill two goals (see van den Bergh and Nijkamp 1990). First, it should ensure for the regional population an acceptable level of welfare that can be 160 Regional Sustainable Development and Natural Resource Use sustained in the future. Second, it should not conflict with sustainable develop- ment at a supraregional level. The latter goal implies that RSD for a single region as a spatial decisionmaking unit should be compatible with global sustainable development. Consequently, if all regions of a global system are marked by RSD, the development of the global system will be sustainable as well. Clearly, the RSD paths of specific regions may have different characteristics because of specific regional circum- stances, for example, availability and use of natural resources and socio- economic capital, environmental vulnerability and resilience, and socio- economic distribution of income and employment. Thus, it is not easy to typify RSD in general terms. Given the existence of trade, transport, dispersion of species, or other socio- economic and ecological linkages between regions, it may be possible to attain sustainable development at a global level without having RSD at the regional level. In extreme cases it might even be possible that global sustainable develop- ment demands that regions "sacrifice" some degree of their own development, welfare, or environmental sustainability. Such detrimental effects may be acceptable from a supraregional (or global human need) perspective. This situa- tion of a retreat to "weak sustainability" may happen, for example, when certain regions are used for specific environmental or economic purposes (such as con- servation, concentration of industrial activity, or dumping of waste). According to the Brundtland Report (WCED 1987, p. 45), it is not even realistic to suppose that every ecosystem (which may encompass more than one region) can be preserved and kept intact everywhere. In these cases we have to look for regional implications of sustainable development, which may call for solutions to the difficult problem of regional compensation for welfare losses. Compensatory measures can be used for both spatial and time phenomena. The idea of such compensation is appealing because it allows alternative policy choices over space and time. However, in various practical situations, many questions still must be answered to make this option really operational. One question is how compensation is allocated (for example, what are the environ- mental and socioeconomic consequences of the compensating project?). Other issues involving regional compensation may relate to the possible time lags between destruction of environmental amenities and their reconstruction or to the costs to be borne by various groups. Recently, various interesting proposals have been made in this context. One is to replant forests in Central America as a compensation for the carbon dioxide produced by a new power station in New England. Of course, the idea of compensation is easier to apply to large individ- ual projects with a single decisionmaker. For numerous small daily decisions by individual persons (such as painting a house, driving a car, or smoking a ciga- rette), direct compensating measures are hard to carry out because of the so- called large number case (see Baumol and Oates 1988). To provide a more concrete and operational approach to RSD, we may focus attention on a more limited issue: sustainable resource use (SRU). This applies Nijkamp, van den Bergh, and Soeteman 161 mainly to renewable stocks of natural resources (for example, forests) and reflects the idea that the use of goods and services provided by such a resource base can be regulated to maintain an optimal stock level. We define a resource base here as the complex of resources and their regenerative support systems- resources that are critically important for regional welfare. Sustainable use of a region's stock of resources may be regarded as an important, though not suffi- cient or even necessary, condition for RSD. Such a conservation strategy is a kind of risk-avoiding strategy. If a stock of renewable resources is used wisely, it may generate a flow of materials, services, or both for an unlimited period. If this flow is sufficient for generating an acceptable welfare level for the regional population ("sustainable welfare"), it is clear that it is of advantage to aim at conservation of the resource. Thus, resource management is a critical variable, because overuse or extinction of a resource is in most cases irreversible. More- over, SRU guarantees regional income at least for a considerable time. RSD may then provide an appropriate bridge between SRU and sustainable development. An important concept in this framework is the carrving capacity of a region, which refers to the existence of ecological upper limits to a further increase in the economic system, often reflected in a logistic growth path of a region. When applied to human population in a region, carrying capacity may indicate the number of people that a region can support. In other cases it may be used as an indication of the number of tourists the region can accommodate. Three major aspects play a role in the determination of carrying capacity: biophysical ele- ments, behavioral elements, and economic amenities. The first aspect includes spatial densities, stress factors related to ecosystems, sensitivity of animal and vegetation species, soil, climate, and groundwater and surface water characteris- tics. In the second category fall spatial interaction, congestion, and crowding, which can be related to preferences of the population, tourists, or both. The third category includes the supply of goods and services, housing facilities, and employment. Carrying capacity is essentially a dynamic concept; it presupposes a proper and flexible management of natural environments over time, so that, for example, some biophysical aspects may become less constraining. Espe- cially, economic amenities can be influenced to improve the carrying capacity. Carrying capacity can be used to determine the boundaries of a feasible RSD or SRU. Some of the other features of a resource base that are useful in this context are exhaustibility, potential depletability, multifunctionality, stability, private versus public ownership, and whether the resource base is a closed or open system. The question of which variables are of critical importance for the attain- ment of RSD is discussed in the next section. III. CRITICAL SUCCESS FACTORS FOR RSD RSD and SRU are not just theoretical concepts; they also serve as a practical guideline for regional development policy. As suggested above, RSD serves to 162 Regional Sustainable Development and Natural Resource Use find a balance between the following objectives (compare Siebert 1987 and World Bank 1987): * Economic progress and/or potential of the area concerned * Ecological values and/or constraints of the area concerned * The economic and environmental interest of parties not directly involved (other regions; different generations). These three objectives are certainly mutually conflicting, at least in the short run, because a maximization of any one objective will affect the two remaining ones. In the long run, however, these three objectives are less conflicting; instead of substitution a situation of mutual complementarity (or coevolution) may emerge. A critical success factor then may be interpreted as a particular crucial variable whose availability or presence is able to enhance-in a qualitative or quantitative sense-the performance of at least one objective without reducing the performance of the remaining ones (a Pareto-optimal condition). Critical success factors for RSD can also be seen as the necessary conditions that act as crucial variables for the attainment of a balanced development in terms of RSD or SRU in a region. Examples of such critical success factors are the availability of water in arid or semiarid areas in view of the needs of agriculture, industry, and households; the existence of accepted institutional rules for the management of a common resource pool (for example, in the fishery sector); and the presence of limits to growth caused by the carrying capacity of a region. Clearly, a critical success factor presupposes that it can-directly or indirectly- be influenced by public policy measures oriented toward the attainment of RSD. To identify critical success factors in RSD, a stepwise approach may be fol- lowed. In view of the great many different types of potential uses of a resource base, as well as human preferences for and the characteristics of the resource base, one should identify which factors are most relevant for a specific regional development plan. A first and obvious step in this analysis of RSD is the delineation and character- ization of the region. Given the multidimensional nature of RSD, a wide variety of different types of regions can be distinguished-for example, developed regions, densely populated areas, urban regions, industrial areas, environmen- tally protected areas, regions lagging in many aspects, islands, and recreational areas. It is impossible to typify a set of regions a priori in an unambiguous manner according to such fairly general criteria for RSD. Only when regional characteristics are specified in more detail, and regional development objectives are formulated, is it possible to typify RSD. For this reason, research on RSD should be devoted partly to case studies of various types of regions, an approach adopted below. A meaningful regioDal classification from the viewpoint of RSD may be based on the following characteristics of the regions under consideration: socio- economic circumstances, economic development phase, level of pollution or congestion, natural resource base, and regeneration and depletion rates of resources. Nijkamp, van den Bergh, and Soeteman 163 A second step in the analysis of critical success factors in RSD is a stocktaking of the characteristics of the internal structure of a region, its interactions with other regions, and other relevant global impacts. Using this information, one may arrive at a clearer view of the potentials for and constraints on the region's future development. One may also extrapolate from past and present situations to determine feasible future states or scenarios of the regional system concerned. A third step in identifying critical success factors involves assessing both prob- able and uncertain developments and those that deviate from the extrapolated path. Probable developments include expected policy actions, changing prefer- ences because of a scarcity of amenities, expected technological developments, investment programs, conservation strategies, or social welfare programs. Uncertain and more drastic developments may involve the disappearance of old economic sectors, the emergence of new sectors, drastic changes in public invest- ments, and changes in institutional arrangements. The assessment of future developments relevant to RSD also should entail extraregional developments, especially with respect to resource availability, demand for exports, world mar- ket prices, international competition, and external pollution emissions. A fourth step in critical success factor analysis is identifying and evaluating different development paths for the regional system concerned. As a first step, a set of performance indicators and an evaluation method must be decided upon. Two types of indicators are important in the evaluation of RSD. The first type indicates the desirability of a state of the system at a point in time. The static welfare indicators may include income distribution, regional income per capita, sectoral diversity, unemployment, public services, infrastructure, land use, level of congestion, amount of vegetation, species diversity, and water quality. The second set of indicators that must be developed for dynamic evaluation encom- passes information about the potentials and constraints for future development of a regional system. In evaluating the time paths of indicators and in case of multiple indicators, use of multicriteria techniques for the evaluation of alternative development options with respect to RSD seems logical. But an explicit policy evaluation of specific indicator outcomes is not easy, and important issues such as stability and fluctuations cannot always be included directly. Then, the use of constraints for demarcating acceptable development paths seems more useful. This also comports with the fact that RSD does not give rise to simple judgment criteria and it certainly will not result from optimizing a single criterion. IV. INSTITUTIONAL POLICY ASPECTS OF RSD An integrated regional policy may be centrally organized, delegated to lower levels (regional, local) or organized from the viewpoint of functional specializa- tion. This assignment problem of competencies and operational tasks is a focus of this section, which seeks to make a first step in designing a framework for integrated regional policy organization oriented toward the achievement of RSD or SRU. 164 Regional Sustainable Development and Natural Resource Use Clearly, in the framework of RSD an integrated economic and environmental approach to policymaking is a necessity. However, its implementation will face a variety of difficulties. One of these difficulties is related to the institutional structure and competence of regional planning (see Zimmermann 1982). In this context, the institutional structure refers to the whole system of rules and regula- tions by which competencies, tasks, and responsibilities are divided among actors. Present planning structures have evolved over a long time. In general, new governmental tasks and responsibilities emerge as a result of the perceived imperfect working of the market economy. Consequently, rooted in their own specific historical backgrounds, separate planning cultures have evolved-for example, housing programs arose as a result of bad hygienic circumstances at the end of the last century; employment planning as a result of the depression in the period before World War II. Physical planning became popular in the 1920s as a result of an unbalanced development between industrialization, housing, and transport. Environmental and social planning are the most recent develop- ments in this respect. A general characteristic of many of these planning efforts is that in early phases of planning the emphasis was mainly on prohibitions and regulations, whereas in later phases more emphasis has been laid on conditions for develop- ment. This is certainly true for environmental planning. It started with measures coping with the negative environmental externalities of production and con- sumption processes. Nowadays there is a tendency toward more source-oriented policies. It will be clear that a development policy in which environmental policy will be a key factor for welfare development is impossible without a simultaneous design of coordination channels between the various functional areas of policy concern. Many coordination problems are the result of the above-mentioned historical roots of segmented planning. The coordination problems may also be the result of a different educational background of specialists, separate planning languages, separate planning horizons, use of different data bases, lack of inte- grated formal and evaluation models, and use of different performance indicators. Another issue in the context of coordinated and integrated planning is hier- archical structuring. A hierarchical planning design may be seen either from the viewpoint of responsibilities and tasks delegated to various levels or from the viewpoint of a hierarchical division of goals. An advocate of a more economic approach to institutional assignment prob- lems is Zimmermann (1982), who formulated some guidelines with respect to the choice of the level at which an environmental policy should be carried out: * Degree of uniformity or diversity of regional preferences with regard to certain environmental and socioeconomic goods and services (especially with regard to equity aspects) Nijkamp, van den Bergh, and Soeteman 165 * Degree of economies of scale regarding the supply and provision of environ- mental goods (especially with regard to efficiency aspects) * Extent of spillover effects (both equity and efficiency aspects are relevant here). Economics seems to provide at least some meaningful guidelines for a demar- cation of planning responsibilities, although ecological considerations also should be taken into account. The National Physical Planning Agency in the Netherlands defines environmental policy as "the stimulation of certain spatial and ecological conditions in such a way that the real aspirations of individuals and groups in society can be realized to a full extent" and "the diversity, coher- ence and sustainability of the physical environment can be guaranteed as much as possible." Administrative and legal boundaries seldom coincide with those of the envi- ronmental problem region concerned. This has resulted from the fact that envi- ronmental problems are often being carried out on the basis of assigned respon- sibilities without special emphasis on the spatial dimensions of the environmen- tal problem. To carry out a proper policy-oriented toward RSD-the assign- ment should also be based on spatial ecological characteristics of the problem (Klijn 1988). To clarify the latter observation, table 1 illustrates some environmental prob- lems and the spatial range of their impact. For example, climatic change is a process that takes hold in the atmosphere and works through the whole ecosys- tem. Therefore, an efficient policy to influence this process necessitates a global planning approach. It is thus plausible that from an ecological point of view that an institutional assignment of responsibility should be derived from the spatial impact of the environmental problem (see Clark and Munn 1986). A proper environmental policy also should take account of interdependencies Table 1. Environmental Problems and the Spatial Range of Their Impact Environmental problem Spatial range of Climatic Acidifi- Nutrifi- Aridifi- Destruc- Disper- Distur- impact change cation Pollution cation cation tion sion bance Atmosphere o o o Geomorphological structure o 0 Relief o Groundwater o o o Surface water o 0 0 0 Soil o o o o Vegetation o o Animal life l I I 1 o 1 o o: intervention point. I: impact of environmental problem on ecosystem. Source: Kliin (1988). 166 Regional Sustainable Development and Natural Resource Use between regions. According to Odum (1971), the main goal of environmental policymakers should be to minimize the impact of human activities on the environment from the perspective of ecosystems. Clearly, because environmental and socioeconomic variables influence each other, a coherent policy is necessary to find a balance between these variables. But a coherent policy does not mean that it should automatically be imple- mented at a centralized level. Decentralizing national policy, however, raises some difficulties. For example, regional or local agencies may strive for maxi- mization of welfare of their own region, thereby neglecting the interregional effects of some of their activities or measures (spillover problems; see Siebert 1985). In addition, environmental media may differ as far as their spatial cover- age is concerned. This implies that the planning regions will overlap, which in turn will create coordination problems (Ewringmann and Hansmeyer 1980). Decentralization of environmental policy may be justified because regional authorities are in a better position to identify regional preferences and to imple- ment regional targets, and they will normally be better informed than national authorities regarding the implementation or successes of environmental policy instruments (Siebert 1985). Horizontal coordination between economic and environmental policy also may lead to greater efficiency. In this context it might be useful to distinguish between institutional policy aspects and executive policy aspects. Institutional policy aspects are related to the distribution of responsibilities over various policy levels (constitutional power). Executive policy aspects are related to decisions taken during the policy period to ensure that the policy goals are achieved (executive power). It is evident from the previous remarks that the existence of a proper institu- tional assignment is a critical success factor for RSD, especially in view of the differences in spatial coverage of economic and environmental policies. V. MODELS FOR SUSTAINABLE DEVELOPMENT Sustainable development has induced many scientific discussions on the intri- cate relationships between economy and ecology, but by no means has it led to a proper specification analysis of suitable models. The issue of RSD-and also the identification of critical success factors-inevitably evokes the question of appropriate model design and structure. An important problem is which specific type of model is relevant for gaining insight into RSD conditions or for tracing RSD paths. This question may be dealt with by focusing on model structure and specification aspects on the one hand and model use and evaluation aspects on the other (see van den Bergh and others 1990). The most significant features of models for sustainable development that distinguish them from other models used for environmental problems are the following: * A module that describes the dynamics of resource bases and ecosystems; the Nijkamp, van den Bergh, and Soeteman 167 indirect effects and consequences of specific economic developments for natu- ral environments can then be traced. * Ecological feedback effects of economic activity on the economic system, which are how the ecosystem provides the economic system with dynamic physical constraints and potentials. * Inclusion of qualitative development and change, calling for a detailed descrip- tion of sectoral interactions and of decision and behavioral processes. (Clearly, uncertainty about social and technological processes makes it very difficult to include such qualitative elements.) * Inclusion of production and welfare elements, which means both a modeling of the economic structure and an evaluation of the output (including also an evaluation of welfare derived from the nonproductive use of the natural envi- ronment); in general, both potential and actual, productive and nonproduc- tive uses of the environment are to be modeled. Sustainable development is concerned with evolution over a long time period, focusing on stability issues and especially structural changes-that is, changes that result in qualitatively different characteristics of states or behavior of the system under consideration. Clearly, models for sustainable development should be dynamic in any case (see Smith 1977). Some structural changes may be generated endogenously by a model. Mathematical theories of chaos and bifur- cation, for example, have shown that simple dynamic models can generate behavior with changes in qualitative characteristics of states (see Baumol and Benhabib 1989; Nijkamp and Reggiani, forthcoming). But most structural changes result from forces from outside a model, because they are uncertain in their nature and characteristics. In modeling terminology, structural changes may emerge in the following ways: * Time paths may include nonsmooth or discontinuous parts as a result of reaching boundaries imposed by model constraints. * The parameters of a system of dynamic equations may change and give rise to different qualitative behavior (for example, the number or character of equi- libria may change). * Relationships and variables may be deleted from or added to the model at hand, or the functional form of a relationship may change. * Stochastic specifications may generate time paths with sudden changes in the value of variables; in the case of small disturbances, this also shows a high sensitivity of model behavior to parameter or initial stock values. Clearly, it is possible to use dynamic models for RSD in a way that anticipates various-sometimes less probable-changes, for example, by combining several scenarios with simulation models. But real uncertainty about processes leading to structural changes and about changes in qualitative characteristics in the system means that it is hardly possible to overcome these problems, as this is essentially a consequence of limited knowledge about real-world processes. 168 Regional Sustainable Development and Natural Resource Use Especially the complicated pattern of interactions within and between economic and ecological processes calls for a detailed description. Indirect, feedback, nonlinear, time-delayed, and other kinds of relationships can be dealt with most appropriately in a formal logical framework. Simulation models are especially suitable for incorporating many theoretically and empirically obtained results of partial studies. In that sense a model provides a tool for obtaining many valuable insights, which can be tested and improved. Moreover, with inclusion of uncer- tainty in specification and use of models, one may obtain more quantitative, testable, and reliable estimations than an intuitive reflection on relationships between uncertainty and indicator values (for example, with simulation meta- regression techniques; see Law and Kelton 1982). Most dynamic (multi)regional models are based on programming, simulation, or analytical techniques. Most modeling approaches use only a single technique. The shortcomings of each technique can be compensated for by merging multi- ple techniques in a coherent combination. For example, Lonergan (1981) pro- posed combining programming models with simulation models. The simulation model was used in his framework to describe the ecosystem being controlled for economic purposes. An optimizing module was used to include normative aspects (that is, maximizing total added value of regional activities), while it received data from the simulation model to determine the right-hand sides of its constraints. In an iterative way, the optimum of the objective constrained by the ecological systems model and the requirements then could be reached. Comprehensive models (that is, multidisciplinary, integrated, economic- ecological models) offer many opportunities for dealing with the requirements and demands posed for RSD issues. Integration may refer here to modules/ disciplines, techniques/models, and aggregation levels. Some authors propose a combined use of programming, econometric, and input-output models (see, for example, Isard 1986). Boyce (1988, p. 6) has stated in this context that "we should be seeking to integrate our theoretical concepts into more comprehensive model formulations." It is an interesting exercise of course to design a general conceptual RSD model (see van den Bergh and others 1990), but in the context of our study we have limited ourselves to a general discussion of relevant RSD issues (and relevant models), followed by a set of concrete case studies. Thus, we move on to illustrate our ideas on the basis of three RSD case studies. VI. CASE STUDY A: THE PEEL AREA IN THE NETHERLANDS This section describes a first illustration of RSD analysis. The Peel area in the southeast Netherlands has been selected for a pilot study on RSD because of the intensive-and problematic-interactions between the natural resource base and economic activities there. The Peel region is situated on relatively high sandy soils and includes two hydrological systems divided by a faultline. Until the nineteenth century the area was an almost impenetrable marsh, dominated by peat moss. But two centuries of human influence have brought extensive drain- Nijkamp, van den Bergh, and Soeteman 169 age and large-scale extraction of peat, which have created land for agriculture and forestry. This process has left only small pieces of the original natural system, such as the Groote Peel and the Maria Peel. These two natural fen areas are situated in an area in which intensive cattle farming and mixed agriculture are the dominant users of the land. Most farmland in the Peel areas is drained each spring, lowering the water table to allow machines to work the land and cattle to enter pastures early. During the summer, when recharge of the aquifer itself is limited because of the spring drainage, shortfalls in soil moisture are circumvented by using irrigation sprinklers to pump from reserves of groundwater. Groundwater extraction for agriculture is shallow and widespread, whereas that for municipal supply occurs at a small number of sites and involves deeper extraction. The last few fens and marshes are threatened by dropping water tables. The intensive cattle farming practice causes large emissions of ammonium. Nearby forests and fens have suffered from various forms of ammonium-related damage. The high loads of nitrogen cause changes in the regional heathland by giving grassy vegetation a competitive advantage. Excess manure is spread over both pasture and crop- land, leading to relatively high levels of soil nitrate, much of which leaches to lower strata and the groundwater aquifer. RSD thus is at stake in this region. The most important interest groups in the region are the farmers, the drinking-water companies (representing also household interests), the tourist industry, the commercial forest owners (the State Forestry Service and private owners), and the nature conservation lobby (including the state as the owner of the national park de Groote Peel). The renewable natural resources demanded by the regional and national inter- est groups are clean groundwater; clean air; vital and productive commercial forest; diverse and vital national park communities; and land for agriculture, development, protection of natural communities, and urban development. The Peel region's natural resources are used as goods-that is, groundwater for drinking and irrigation, timber for paper and pulp. The ecosystems perform services in the form of outdoor recreation and nature conservation, and they assimilate surplus ammonium and nitrate from nature. Hence, there are various questions and conflicting issues regarding the sustainability of land use in this area. And, clearly, the identification of critical success factors is crucial for RSD policies. Sustained resource use in the region forms the starting point of our analysis. Economic activities are taken into account insofar as they influence (or are influenced by) these resources. Consequently, the regional boundaries were determined primarily by ecological and geographical criteria, based on the groundwater basin around the Peel-fen reserves. The renewable natural resources central to this study are groundwater, forests, and natural vegetation. The issues associated with these may be summarized as follows: * High water tables, sandy soil, and nutrient-poor conditions have led to the development of unique ecological communities. 170 Regional Sustainable Development and Natural Resource Use * Widespread drainage of the land and multiple use of the groundwater resource (for irrigation as well as municipal supply) have lowered the water tables. * Agricultural activities that use fertilizers intensively and that result in increas- ing production of manure are causing nitrate enrichment of the groundwater, affecting the remnant vegetation, and decreasing the suitability of the ground- water for human use. * Air pollution is also causing acidification of soils, with effects on the natural vegetation as well as on forests. In an RSD analysis, the economic sector also has to be considered. Economic activities that are directly dependent on the groundwater resource include agri- cultural and municipal water supply. Other activities in the region are timber production, recreation, and nature conservation. Agriculture especially contrib- utes significantly to regional income. For some activities a further subdivision is useful. For example, timber production is based on two tree species-pine and Douglas fir, both of which are produced in plantations. Agriculture includes livestock rearing (cattle, pigs, and poultry) and crop cultivation (for livestock and human consumption). Livestock rearing can be intensive (for example, bio- industry for meat and egg products) or extensive (for example, dairy and meat). The spatial distribution of activities in the region also affects their interactions and relationships with available resources. For example, groundwater extrac- tion for agriculture is shallow and widespread, whereas that for municipal sup- ply occurs at a small number of sites and involves deeper extraction. The central focus of this case study is on the relationship of the region's SRUS to its economic activities, with a view to identifying the critical success factors for RSD. Multiple use is a prominent feature and a source of conflict, because allocation of a scarce resource among users involves tradeoffs. For example, economic activities are not the only users of groundwater, and groundwater is also crucial for the regeneration of wetland communities. In view of the foregoing observations on RSD in the Peel area, the most important critical success factors associated with these resources may be summa- rized as follows: * Maintaining the groundwater level. High water tables have been a major feature of the Peel and, together with the sandy soil and generally nutrient- poor conditions, have led to unique ecological communities. Widespread drainage of the land and increasing use of the groundwater resource (for irrigation and municipal supply) have lowered these water tables. * Curtailing nitrogen pollution. Agricultural activities using fertilizers and pro- ducing excessive amounts of manure are causing nitrate enrichment of soil and groundwater, leading to changes in the remnant fen and heathland vege- tation as well as to decreasing suitability for human consumption. Agri- culturally originated ammonium emissions, in addition to nitrogen and sul- phur oxides from various sources within and outside the region, affect the vegetation directly and contribute to the acidification of soils, resulting in changes in the natural vegetation and in decrease of forest vitality. Nijkamp, van den Bergh, and Soeteman 171 Regulating competitive demand for land. The present and expected future demand for land by the agricultural sector, nature conservation, and urban (including residential, industrial and tourist attractions) interest groups is conflicting and less sustainable. The analysis of regional system interactions in this area has resulted in the design of an exploratory dynamic simulation model (see van den Bergh and Nijkamp 1990). The submodules describe groundwater, nitrates, forestry and natural vegetation, agriculture, and regional economic activities. The sub- module that describes the economic activity records profits over time for each sector on the basis of developments of quantities, costs, prices, and technology. The time paths for quantities are for most sectors based on changes in produc- tion capacity, except for recreation, where demand for recreational activity determines the quantity. The development of the economic system is to a large extent determined by exogenous variables, for which the time paths were chosen on the basis of a relevant development scenario. The fact that a relatively low number of interactions between different sectors is modeled is because of the small size of the region. Models that include many interactions between sectors (for example, interindustry supply, or competition on factor and final markets) usually have an economywide rather than a regional orientation (see Vincent 1982). The global interrelationships between the modules are listed in figure 1 (for a full description, see van den Bergh and others 1990). These indicator variables were chosen for the assessment of RSD: * Value added in the region * Costs of measures * Nature conservation value * Recreational attractiveness * Groundwater quality with respect to concentration of nitrates * Air quality with respect to concentration of nitrates and ammonia in the air * Stock of groundwater * Stocks of vegetation and forest. The indicator for nature conservation value is based on vegetation areas. Recre- ational attractiveness is based on economic facilities, natural amenities, and disservices (arising from economic activities). The spatial scale of the model distinguishes between two subregions, the national park de Groote Peel and buffer zone, and the rest of the region. The indicators selected to provide relevant information for the judgment of RSD in this area are as follows: * Yearly value added in crops, livestock, extracted groundwater, forestry, and timber * Total recreational (amenity) value * Cost-benefit ratio of pollution control measures * Yearly average ambient air concentration of ammonium (NH3) * Yearly average nitrate concentration in groundwater (NO3). 172 Regional Sustainable Development and Natural Resource Use Figure 1. Modules and Relationships of the Peel Model _ pi~~~~~1 Economic l_ Comuntydivrstyinhenaio activiti p Forestry and . 10natural vegetation Thle Peel * Land area classified as urban, agricultural, and natural * C.ommunity diversity in the national park. As mentioned above, external developments and policy choices will be incor- porated via scenarios. Each scenario that is used for a simulation run has effects that are evaluated regarding their RSD via the indicators listed above. Effects may be compared with standards and may lead to inferences about acceptance or rejection of the relevance of the scenario used for RSD. The scenarios are determined by choices for both exogenous and management variables. To limit the number of scenarios, we have identified some plausible scenarios consisting of a set of related changes in variables. The time horizon of the scenarios is fifty years with 1980-81 as base years; the time solution is given in years. The model has been run for each of the six scenarios mentioned hereafter. Each scenario description is followed by a brief evaluation of the time paths of indicators. Scenario 1: Business as Usual The stock of grazing cattle declines from 1980 to 1985 and remains constant during the rest of the simulation period. The stock of feedlot cattle increases by 10 percent for each period of twenty years. Population will increase by 9,000 per decade for the period 1980-81 to 2000 (trends are extrapolated after that Nijkamp, van den Bergh, and Soeteman 173 year). Imports of nitrogen from outside the Peel region and sulphur oxide emis- sions decline; nitrogen oxides decrease by 30 percent and S02 emissions by some 45 percent after fifteen years. The results in figure 2 indicate that in this scenario major changes are expected after fifteen years. The trend toward grassification of heathland is somewhat delayed. An improvement in the air quality index results, implying that the forest volume of alders increases significantly and that the downward trend for douglas firs is stopped. Ammonia and nitrate emissions increase slowly, and the concentration of nitrates in deep groundwater is somewhat higher than under the steady state. After fifteen years, total value added increases as a result of forest improvement, an increase in recreational revenues resulting from an improved natural environment, and an increase in feedlot farming. Scenario 2: No Import of Sulphur Dioxide and Nitrogen Oxides This scenario is based on the same assumptions as the business-as-usual sce- nario with the exception of imported sulphur dioxide and nitrogen oxides. These emissions are set to a level of zero after fifteen years to assess the impact of foreign policies and the possibilities for regional policy. It also can be seen as an assessment of interregional tradeoffs. Compared with the business-as-usual scenario, soil pH and air quality appear to improve more dramatically (figure 3). It is significant that the trend toward grassification of heathlands would be reversed. Forest improvement, for both Douglas fir and alders, would be strengthened. Value added appears to improve because of the improvement in forestry and recreational value. Scenario 3: Present Environmental Policy On the basis of the business-as-usual scenario and the present government policy of controlling the use of manure on land, a contemporary environmental policy scenario with fairly stringent regulations is created. The simulation results in figure 4 show that after fifteen years a significant decline in ammonia emissions occurs. The downward trends in both alders and wet heathlands are clearly reversed, and the share of grassland decreases significantly after approxi- mately twenty years. The improvements in forestry and natural vegetation, already observed under the business-as-usual scenario, are strengthened. Net present value, apart from the "hiccup" after fifteen years caused by a sudden increase in yields not yet counterbalanced by manure disposal costs, eventually rises toward a higher level. Scenario 4: Present Environmental Policy plus Ammonia Scrubbing This scenario is the same as scenario 3, except that now all feedlot stables are provided with biofiltration equipment after fifteen years. Figure 5 shows that compared with the present environmental policy scenario, the improvement in 174 Regional Sustainable Development and Natuiral Resource Use Figure 2. The Business-as-Usual Scenario for the Peel Area A 0.400 }{ -0 - --- 40 million --6 - 0.325 {~ 9.0 ....' 30 miUion } _5 ----- ----- ---- 5 0.250 7.0 " ~ ~ * 20 miiion 4 _ . 0.175 { _. 5.0 10 miUlion _ _ -3 - 0.100 3 l 0 12.5 25 37.5 50 2 Number of years - 95,000 } B -{ -** 30,000 ---- 450,000 - 2,500 - 86,250 { 26,250 - - - 412,500 .-. _ - 1,875 77,500 ,_/* v-*22,500 375,000 1 22,500 68,750-{ -* 18,750 337,500 ----- 625 60,000 } - _ 15 000 300,000 0 12.5 25 37.5 50 - 0 Number of years 60 miUion _ ___ { ** 120 - - - - 2,000 - - 10.00 - 50 minion -. -"~ 1,500 } - -" 90 I -- ~~~~~~~~~9.38 40 million T - - - - - - . - _ - --- 60 - - - 1,000 - 8.75 30 miUion _ ._.30 500 } 8.13 - 0 2million 0 -20m i }0 12.5 25 37.5 50 7.50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per milions of liters); - - - * is the total ammonia release from manure (kilograms); - * . is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, - * wet heathiand, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); - -- - is an indicator for the nature conservation value based on all vegetation types; -* - is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). Nijkamp, van den Bergh, and Soeteman 175 Figure 3. Scenario Featuring no Import of Sulfur Dioxide and Nitrogen Oxide for the Peel Area 0.400 11.0 - -- 40 million --6 0,325 { 9.0 *30 million } =..5 ----- ----- ---- 5 - 0.25070 "~~ 20million } { =..0 0.175 _ 5.0 - - -.10 million 3 - 0.100 _____-_-3_0 0 12.5 25 37.5 50 B Number of years 95,000 }B { 30,000 ---- 450,000 /--2,500 412,2500 ..__* ..*..-{ "26,250 77, ~ * 841622500 }-- -_ / ' - =~ 1,875 77,500 __ -22,500 8, 375,000 12,500 } 68,750 _ _18,750 -3 337,500 s - 625 60,000 I{ o15000 ---300,000 1 _ 0 12.5 25 37.5 50 ° Number of years 60 million C _{ - 120 - - - - 2,000 -- 10.00 50 milion l , 90 ~*1,500 - 9.3 40 million --------- 60 - - - 1,000 It , _ - 8.75 30 million {..30 500 8.1 - -8.3 20 million __ _ _ _ ______._._ 0 - I } __ 7.50 0 12.5 25 37.5 50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per millions of liters); - - - is the total ammonia release from manure (kilograms); - - - is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, - wet heathland, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); - -- is an indicator for the nature conservation value based on all vegetation types; -* * is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). 176 Regional Sustainable Developmentand Natural Resource Use Figure 4. Scenario Following Present Environmental Policy for Peel Area -0.400 A A {. - - - 40 million --6 0.325 9.0 '*30 million __5 0.250 70 *20 million 4 _-__ 0.175 { _ 50 - -- * 10 miUllon __3 - 0.100 30 0 12.5 25 37.5 50 2 B Number of years 95,000 _ B { 30,000 ---- 450,000 / - 2,500 86,250 26,250 ~ 77412,500 1,875 375,000 * 22,500 3~ 575,000 } - 12,500 -68,750 , { 18,750 -- - 337,500 - , -- 625 60,000 15000 * 300,000 0 0 12.5 25 37.5 50 Number of years 60 million C {.-*.*120 - - - - 2,000 - -- 10.00 50million 90 - -- 1,500 9 --9.38 40 million _ - * J.75 1~ ,000 }: ___. -8.75 - 30 million { .. 30 * 500 } _ - --8.13 - 20 million _ * 0 }- l l -- 7.50 0 12.5 25 37.5 50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per millions of liters); - - - - is the total ammonia release from manure (kdlograms); - * is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, - * wet heathiand, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); - - -- is an indicator for the nature conservation value based on all vegetation types; -* - is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). Nijkamp, van den Bergh, and Soeteman 177 wet heathland as well as the reduction in grassland is postponed, although an increase in Douglas firs is reached sooner. This is caused by an increase in nitrate resulting from biofiltration. Total value added does not alter much in terms of size, but it does so in terms of composition. The costs of biofiltration have a significant impact on the value added in the feedlot industry. This is counter- balanced by the increase in recreational demand and timber production. Scenario 5: Maximum Technical Efforts This is the same as scenario 4, except that more stringent technical standards are required. The standards are based in this scenario on the uptake of minerals by vegetation and are approximately twice as low as the governmental standards for 1995. Accordingly, we have reduced the application of manure on land by 50 percent compared with scenario 4. Figure 6 indicates that ammonia emissions already are clearly lower after ten years. This has a beneficial impact on natural vegetation and forestry, where recovery will take place sooner. The cost of this action, however, has to be borne earlier, especially by feedlot farming. Total value added in the region is not significantly affected. Scenario 6: Land Use Shifts These shifts are based on the business-as-usual scenario with the exception of the area allocated to arable land. This was reduced by 50 percent compared with 1980. The area of land allocated to forestry and natural vegetation increases by approximately 125 percent, with the exception of grassland area, whose size is constant. The volumes of natural vegetation are significantly higher (see figure 7) than under the first scenario. Also, because of less crop irrigation, the stock of surface groundwater is higher, which positively influences natural vegetation. As can be seen from figure 7, initially, total value added is some 25 percent higher, but then it drops and remains only slightly above the business-as-usual level. Further research is required to improve the empirical robustness of the model, with a special emphasis on policy strategies. Some of the data would have to be improved in precision. Some equations require more reliable data to enable a realistic specification (for example, recreational amenity, hydrological pro- cesses, and output from crops as a function of fertilizer and groundwater use). The model also might be validated by means of a historical run to track histori- cal data. Therefore, it is clear that for the time being the above results are mainly illustrative for RSD planning. Nevertheless, it is clear from the model experi- ments that the groundwater level, production of nitrogen, and conflicting demand for land in the area concerned may be regarded as critical success factors for a balanced policy regarding SRU of the Peel area. 178 Regional Sustainable Development and Natural Resource Use Figure 5. Scenario Following Present Environmental Policy for Peel Area but with Addition to Feedlots of Biofiltration Equipment for Ammonia Scrubbing After Fifteen Years 0.400 A {--11.0 -- - 40 milion -~6 - 0.325 { _.. 9.0 ---~ 30 miUion _ _- 5 0.250 7.0 ---- 20 mitlion 4{ _ _7. 0.175 5.0 - --- 10 million . __ _ 0.100 lll _ 3.0 * O2_ - -0 -_12.5 25 37.5 5( 0 2 Number of years 95,000 B -. - { 30,000 - - - - 450,000 -_ .7 -2,500 - 86,250/ N - --- 412,500 } / - =- 126,250 / N ~~~~~1,875 77,500 _ - 122,500 ---- 375,000 12/ -- 122500 68,750 /18,750 --- 337,500 62. . --625 60,000 } I I I _{ 15,000 ~~~ 300,000 v)0 12.5 25 37.5 50 0 Number of years 60 million _ 120 ----~ ~ 2,000 1 , _-10.00 - 50minion _ { =..9g ---1,500 - 9.38 40million {- 60 ---- t,000 -8.75 30 million .130 *.-. 500 f 8.13 20 million _ _,_,_ ___._._ 0 H ~~~~~~I I - 7.50 0 12.5 25. 37.5 50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per millions of liters); - - - - is the total ammonia release from manure (kdlograms); - . . is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, - wet heathliand, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); --- - is an indicator for the nature conservation value based on all vegetation types; -* * is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). Nijkamp, van den Bergh, and Soeteman 179 Figure 6. Scenario Using Maximum Technical Efforts to Curtail Ammonia Emissions in the Peel Area 0.400 A - 11.0 - - -- 40 million 6 - 0.325 **9.0 *30 million }5{=. - 0.250 7.0 ----~ 20 million 4~_S 0.175 I.0 *10million _.. __3 - 0.100 3____________0_____ *0 2-lll{ = 3 0 12.5 25 37.5 50 Number of years - 95,000 }* B * { -*-30,000 ---- 450,000 / . - 2,500 - 86,250 _ . , { 26,250 ~ ~ ~ *412,500 1* / -1,875 77,500 . _ { - - 22,500 - - - 375,000 1i/ \ - 12,500 - 68,750 { -* 18,750 - - ~ 337,500 625 60,000 ____ _ - *10 --- 300,000 _ 15,000 0 12.5 25 37.5 50 l 0 Number of years - 60million C { .. 120 - - - - 2,000 - - 10.00 50 million ,.. 90 -- *1,500 ... / -- 9.38 40million --- - .60 * LO8 1,000.75 30 million . { _ 30 --- 5001} - - 3 20 million r 0 .0 }llll- 7.50 0 12.5 25 37.5 50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per millions of liters); - - - is the total ammonia release from manure (kdlograms); -* - -is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, -* wet heathiand, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); - - -* is an indicator for the nature conservation value based on all vegetation types; -* * is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). 180 Regional Sustainable Development and Natural Resource Use Figure 7. Scenario Postulating Shift in Land Use (Reduction in Arable Land) in the Peel Area - 0.400 A 11.0 - - -- 40 million -- 6 - 0.325 { ~ 9.0 '*30 million }. .--- :{= 5 ----- ----- --- - 5 0.250 { . 7.0 *20 million _. __4 0.175 5 0 - -. 10 million __3 0.100 30 0 12.5 25 37.5 50 Number of years - 95,000 } B { -* - 30,000 ---- 450,000 _ - 2,500 - 86,250 26X250 --- 412,500 }~ - . -26,25 --1,875 - 77,500 22,500 - - * 375,000 12,500 68,750 l - 18,750 - ' 337,500 , - 625 - 60,000 15,000 -~~ 300,000 110 12.5 25 37.5 50 - - 0 Number of years - 60 million C 120 ---- 2,000 1 10.00 50 million ~ ' 1,500 } 9.38 40 million l_____. 60 - - *1,000 } 8.75 "~0 ~ _30 million - .. ~ -30 20 milliion -75 0 12.5 25 37.5 50 Number of years Key: In graph A, - is the concentration of nitrates in the deep groundwater reservoir (in kilograms per millions of liters); - - - is the total ammonia release from manure (kilograms); - . . is the pH level in the soil; and - - is the volume of surface groundwater (millions of liters). In graph B, the lines denote the stocks of natural vegetation; - alders, - - - Douglas fir, - - wet heathland, and - - grass (all in cubic meters per hectare). In graph C, - is the stock of deep groundwater (millions of liters); - - - is an indicator for the nature conservation value based on all vegetation types; -* is a general index of air quality; and - - is the total value added of regional economic activity (in billions of Dutch guilders). Nijkamp, van den Bergh, and Soeteman 181 VII. CASE STUDY B: THE SPORADES ISLANDS IN GREECE The Sporades Islands form a complex of ecologically extremely important but vulnerable islands located off central Greece in the Aegean sea. The islands' geological, archaeological, and cultural value is very high. From an environmen- tal viewpoint they are rich in vegetation. In general, the flora, fauna, and ornithology provide many interesting examples of unique wildlife. The sur- rounding sea hosts sea coral, dolphins, whales, and monk seals-the last a rare species that has been declining drastically in recent decades. New pressures from the tourist and fishery sector for increased exploitation of resources have led to severe conflicts with priorities of ecological balance and conservation. In search of a resolution of mutually conflicting growth trajectories, the Greek govern- ment, in cooperation with the European Community (EC) has established an international marine and terrestrial research center on the island of Alonnissos, and, as part of an environmental preservation policy, a strictly protected area has been designated in the Northern Sporades Marine Park (see Giaoutzi and Nijkamp 1990). Recently a project commissioned by the EC has begun to design a planning- oriented monitoring and information system and an operational strategic policy model. The aim of the model is to support management decisionmaking for the economic development of the islands and the marine park with surrounding areas. Using simulation experiments, a descriptive prototype systems model has been developed. The model uses a time horizon of twenty years, has a time resolution in terms of years, and is designed as a modular system using a so- called satellite principle (see Brouwer and Nijkamp 1988). Because the tourist and fishery sectors are the dominant factors in the stress on the ecosystem in the Sporades region, they are core modules in the integrated environmental model. The modular structure includes three parts (see figure 8): * A regional economic submodel concentrates on the relatively large tourist sector and the fishery sector. The original economic structure of the island, agriculture, is included, as are the governmental services and utilities (although in a more elementary way). Also, simple models for population, tourism, housing, labor and land markets, waste generation, and materials and financial balances are included. * A marine submodel concentrates on the dynamics in the food chain (the top of which is formed by the monk seal) and describes the behavior of passing dolphins and whales, as well as the activities of fishermen. * A terrestrial submodel is developed, including a land use model, a groundwa- ter model, and a vegetation-fauna model. This submodel has many interac- tions with land use development and erosion processes. Some indicators for RSD in the Sporades are: * Tourist services and accommodation * Regional capital (for example, fishing vessels, water pumping capacity) 182 Regional Sustainable Development and Natural Resource Use Figure 8. Structure of the Sporades Model The Sporades * Ratio of profitsintheritourist rit Tors |fseco an ecosystseml * Aquifer stock ~ ~ fiher 4- Rat osfpumptiong raeOn ihn aeotheetok.evlsofaquifern labor, and econormc . ecosystem , housing activities • Water pumping capacity T Regional employment * Income per capita * Congestion, noise, and waste dump * Ratio of profits in the tourist sector to profits in other sectors m Stocks of terrestrial vegetation and animals • Stocks of sea animals * Aquifer stock • Ratios of pumping rates and fishing rates to the stock levels of aquifer and fish, respectively. The RSD issues on the Sporades are described in the prototype model by means of a stock-flow structure. The model is formulated with a spatial resolu- tion of biological zones, based on the spatial coverage of the islands and the marine park, and the spatial aspects of fishery and tourism. The spatial subdivi- sion includes the main islands of Skiathos, Skopelos, and Alonnissos. In a later stage of the research the time resolution will be changed to seasons, because the dominant activities on the islands have a seasonal pattern. The simulation model for the Sporades has a strict modular design. It has an Nijkamp, van den Bergh, and Soeteman 183 extensive economic module, in which economic activities, income, and invest- ments are linked to land use, fishery, and tourism. Tourism is also connected with land use, the marine ecosystem, and the terrestrial system, as well as with fishery. The model has a detailed description of all these components with a special focus on the potential conflicts (and their resolution) between different competing activities. The following factors appear to act as critical success factors in environmental planning and management in this area: * Managing the development of tourism. Tourism increases both the level of welfare and the environmental pressure and plays a critical role in RSD for the Sporades. * Maintaining the qualitative and quantitative aspects of the aquifer. The avail- ability of water is of crucial importance for the economic development and the ecological quality of these islands. However, a proper balance is hard to find. L Limiting the spatial transfer of economic revenues on the islands. If the Spo- rades are only seen and used as a source of benefits for external investors, their ecological and economic base will be eroded, and RSD will not be attained. Fairly extensive fieldwork had to be undertaken, because no reliable data base for these islands was available. A prototype model has been developed that includes in a detailed manner the indicators sketched above. A set of simulation experiments has been undertaken to test the validity of model specifications and the sensitivity of the data set. Some first policy scenarios have been developed, based, for example, on land use zoning, quota systems for the fishery sector, and new agricultural activities. In these exercises, use is also made of geographic information systems to present detailed impacts on land use and the terrestrial system on the islands. VIII. CASE STUDY C: RURAL DEVELOPMENT IN BOTSWANA Despite much effort, the income gap between the developed and the develop- ing world has not diminished. Furthermore, environmental problems have been persistent, both at worldwide and at local and regional levels. In many develop- ing countries we are witnessing resource depletion such as deforestation and desertification. Short-term interests arising from subsistence and survival prob- lems are often eroding balanced long-term resource management. Thus, sustain- able development is often an illusion. In a case study on Botswana, Arntzen (1989) thoroughly investigated the possibilities of RSD. The starting point of RSD analysis here was a blend of the ecological complex approach (an abstract model of relevant relationships within and between var- ious modules in a regional system that are relevant from a sustainable develop- ment perspective) and Wilkinson's (1973) ecological theory of economic devel- opment (including natural resources and negative feedbacks of human activities 184 Regional Sustainable Development and Natural Resource Use on the environment). The imbalance between sustainable resource use and human activities may emerge from various factors, such as socioeconomic strati- fication, external factors, market imperfections, and other development constraints. The current information on Botswana is hardly sufficient for proactive poli- cymaking because the available data do not include a direct valuation of various natural resources (and hence no representation of the actual costs of the resource use), whereas in some cases (for example, timber) even implicit zero prices are assigned. This unfortunate situation, which favors unsustainable resource use, is codetermined by the communal rights to many resources in Botswana. Thus, the property right system works to the detriment of RSD in Botswana. A necessary condition for improving this situation would be to set up an up-to-date system of regional resource accounts. Arntzen's (1989) study focused on RSD in Botswana in an attempt to identify the extent and nature of land pressure, the adjustment strategies used by people, and the impacts of socioeconomic strata and external factors. Particular atten- tion was given to land pressure in various regions in Botswana-Southeast district, Kgatleng district, Central district, and Palapye. Land pressure in various regions in Botswana is fairly high, particularly in small districts, given the available technology and the production constraints in such semi-arid regions. Land pressure is particularly important in case of overstocking. The critical success factors in coping with the problems of land pressure appeared to be the following: * Controlling population growth in the areas concerned * Ensuring balanced livestock development * Controlling land alienation, mainly in smaller districts * Controlling technological developments and related policies regarding boreholes. The analysis was based not on a formal RSD model but rather on extensive fieldwork and statistical data analysis. Particular attention was given to the adaptation mechanisms in situations of high land pressure, such as expansion of the resource base, intensification of existing activities, economic diversification, and population adjustments. Also the impact of dislocating factors for farmers was examined more thoroughly-that is, socioeconomic stratification and exter- nal influences (for example, labor market developments and government policies). It turned out that there was a significant variation in regional adaptation and behavioral responses. Thus, RSD is not a concept that can be handled by means of a single measuring rod. It requires a simultaneous consideration of relevant welfare considerations (including ecological indicators) from the viewpoint of a balanced development in specific regional situations. Nijkamp, van den Bergh, and Soeteman 185 IX. CONCLUSIONS Ecologically sustainable economic development has become a policy issue of major significance. This paper has focused attention on sustainable development in a regional context. Conceptualizing and analyzing sustainable development is clearly important at a regional level. Various advantages of a regional approach have been spelled out, for example, in relation to regional causes and effects of environmental problems, the local character of economic processes, interregion- al interactions, and the possibility of operationalizing sustainable development on a regional scale. Sustainable resource use was presented as an important element of RSD. In our analysis of RSD we started with a stocktaking of the internal charac- teristics of the region, its interactions, and relevant external (global) phenom- ena. In discussing a regional resource base we argued that the present and potential dependence of regional activities on the resource base as well as the specific characteristics of the resource base should be assessed. With regard to the use of models, we mentioned the main problems in the context of RSD issues, with a particular emphasis on critical success factors. Finally, we presented three case studies in which some of the general discussions were illustrated, indicators for RSD were mentioned, and critical success factors were identified. In one case study, on the Peel area, we presented simulation results for scenarios including different, but realistic, future options with regard to continuation of present level of cattle breeding, environmental policy, technologically feasible emission levels, and land use shifts. Various lessons and issues for further research can be drawn from the above exposition. RSD refers to a systemic view of interacting regions of our world economy, not only in terms of economic linkages but also in terms of environ- mental interactions. Thus space and time are the two essential dimensions of RSD. The openness of a spatial system evokes the question of environmental sacri- fices in space and time. The issue of the substitutability of environmental decay (in other periods of time, in other regions) is at stake here; the related concept of weak and strong sustainability requires further exploration on the basis of a thorough investigation of underlying welfare concepts (and associated collective welfare functions). In order to find a balanced equilibrium of resources in space and time, much attention would have to be given to regional resource accounting. Efficiency, equity, and externality aspects of resource use can then be given due emphasis, while attempts at valuing unpriced resources may provide a sound basis for project appraisal. Geographic information systems may be helpful here to visu- alize all relevant policy aspects. Environmental externalities should be looked at not only from the viewpoint of market failures (or signal failures) but also from the viewpoint of intervention 186 Regional Sustainable Development and Natural Resource Use (or response) failures. In many cases governments-in an attempt to cope with environmental deterioration-have misjudged the causes of such decay and have even aggravated the original conditions (witness the many failures in transport policy worldwide, for example). Thus, a closer analysis of the costs and remedies regarding both signal and response failures would be desirable. Inter- nalization of social costs then might also be considered at an early stage of decisionmaking-that is, in the design phase rather than ex post, when all decisions have already been made. RSD also calls for integrated community impact analysis and related project appraisal methods (for example, in the form of multi-objective decision analysis and multicriteria analysis). In the context of RSD, critical success factors refer to major focal points for policy intervention. Identification of such factors may induce proactive sus- tainability of policies. In the context of RSD planning, where regions are open, critical success factors also may be found outside the specific region. As shown in this paper, simulation experiments leading to concerted evalua- tion procedures for RSD may be helpful analytical tools. Some experiments demonstrated in this paper have pointed out the value of simulation models for RSD analysis, particularly with a view toward identifying critical success factors in a balanced regional development. REFERENCES Archibugi, Franco, and Peter Nijkamp, eds. 1989. Economy and Ecology: Towards Sustainable Development. Dordrecht, Netherlands: Kluwer Nijhoff. Arntzen, Jaap. 1989. "Environmental Pressure and Adaptation in Rural Botswana." Ph.D. diss., Department of Economics, Free University of Amsterdam (Netherlands). Bartelmus, P. 1986. Environment and Development. London: Allen & Unwin. Baumol, William J., and J. Benhabib. 1989. "Chaos: Significance, Mechanism, and Economic Applications." Journal of Economic Perspectives 3: 77-106. Baumol, William J., and W. E. Oates. 1988. The Theory of Environmental Policy. Cambridge, U.K.: Cambridge University Press. Boyce, David E., 1988. "Presidential Address: Renaissance of Large-Scale Models." Papers of the Regional Science Association 65: 1-10. Brouwer, Floor M., and Peter Nijkamp. 1988. "Design and Structure Analysis of Inte- grated Environmental Planning Models." European Review of Agricultural Eco- nomics 15: 1-38. Ciriacy-Wantrup, S. V. 1952. Resource Conservation. Berkeley: University of Califor- nia Press. Clark, William C. 1986. "Sustainable Development of the Biosphere: Themes for a Research Plan." In W. C. Clark and R. E. Munn, eds., Sustainable Development of the Biosphere. Cambridge, U.K.: Cambridge University Press. Clark, W. C., and Ted Munn, eds. 1986. Sustainable Development of the Biosphere. Cambridge, U.K.: Cambridge University Press. Collard, D., David Pearce, and D. Ulph. 1989. Economics, Growth and Sustainable Environment. London: Macmillan. Nijkamp, van den Bergh, and Soeteman 187 Ewringmann, D., and K. Hansmeyer. 1980. "The Institutional Setting of Regional Environmental Policy." In H. Siebert, I. Walter, and K. Zimmermann, eds., Regional Environmental Policy. London: Macmillan. Frey, Rene L. 1980. "Interregional Welfare Comparisons and Environmental Policy." In H. Siebert, I. Walter, and K. Zimmermann, eds., Regional Environmental Policy. London: Macmillan. Giaoutzi, Maria, and Peter Nijkamp. 1990. "A Strategic Environmental Information System and Planning Model for Marine Park Development." Research Memorandum, Department of Economics, Free University of Amsterdam (Netherlands). Processed. Isard, Walter. 1986. "Reflections on the Relevance of Integrated Multiregion Models." Regional Science and Urban Economics 16: 165-80. Kairiukstis, Leo, ed. 1989. Ecological Sustainability of Regional Development. Lax- enburg, Austria: International Institute for Applied Systems Analysis. Klijn, Fred. 1988. Milieubeheergebieden (Environmental Management Areas). CML Information Bulletin 37. Leiden, Netherlands: Centrum voor Milieukunde Rijksuniversiteit Leiden. Law, A. M., and W. S. Kelton. 1982. Simulation Modeling and Analysis. New York: McGraw-Hill. Lonergan, Stephen C. 1981. "A Methodological Framework for Resolving Ecological- Economic Problems." Papers of the Regional Science Association 3: 117-34. Nijkamp, Peter, and Aura Reggiani. Forthcoming. "Logit Models and Chaotic Behav- ior." Environment and Planning. Nijkamp, Peter, and Frits Soeteman. 1988. "Ecologically Sustainable Economic Devel- opment." InternationalJournal of Social Economics 15, no. 3/4: 88-102. 1989. "Land Use, Economy and Ecology." Futures (December): 621-34. Norgaard, R. B. 1984. "Co-evolutionary Development Potential." Land Economics 60, no. 2:160-73. Odum, E. P. 1971. Fundamentals of Ecology. Philadelphia: Saunders. Paelinck, Jean, and Peter Nijkamp. 1976. Operational Theory and Method in Regional Economics. Aldershot, U.K.: Gower. Pezzey, J. 1989. "Economic Analysis of Sustainable Growth and Sustainable Develop- ment." Working Paper 15, Environment Department, World Bank. Washington, D.C. Processed. Randall, A., and E. N. Castle. 1985. "Land Resources and Land Markets." In A. V. Kneese and J. L. Sweeney, eds., Handbook of Natural Resource and Energy Eco- nomics. Amsterdam, Netherlands: North-Holland. Shefer, Daniel. 1974. "The Optimal Use of Natural Resources." Journal of Environmen- tal Systems 4, no. 4: 269-79. Siebert, Horst. 1985. "Spatial Aspects of Environmental Economics." In A. V. Kneese and J. L. Sweeney, eds., Handbook of Natural Resource and Energy Economics. Amsterdam, Netherlands: North-Holland. Siebert, Horst. 1987. Economics of the Environment: Theory and Policy. Berlin, Ger- many: Springer. Smith, V. L. 1977. "Control Theory Applied to Natural and Environmental Resources." Journal of Environmental Economics and Management 4: 1-24. van den Bergh, Jeroen, Leon C. Braat, Alison J. Gilbert, David E. James, Ger Klaassen, and Frits J. Soeteman. 1990. "Sustainable Use of Natural Resources and Economic 188 Regional Sustainable Development and Natural Resource Use Development in the Peel." SPIDER paper, Institute for Environmental Studies, Free University of Amsterdam (Netherlands). van den Bergh, Jeroen, and Peter Nijkamp. 1990. "Ecologically Sustainable Economic Development in a Regional System: A Case Study in Agricultural Development Plan- ning in the Netherlands." Research Memorandum, Department of Economics, Free University of Amsterdam (Netherlands). van den Bergh, Jeroen, and Frits Soeteman. 1990. "Sustainable Economic Development of Global and Regional Systems." Research Memorandum, Department of Eco- nomics, Free University of Amsterdam (Netherlands). Vincent, D. P. 1982. "The ORANI Approach to Quantitative Modelling of the Australian Agricultural Sector." European Review of Agricultural Economics 9: 415-42. Wilkinson, R. S. 1973. Poverty and Progress. London: Methuen. World Bank. 1987. Environment, Growth and Development. Development Committee Pamphlet 14. Washington, D.C. WCED (World Commission on Environment and Development). 1987. Our Common Future. New York: Oxford University Press. Zimmerrnann, Klaus. 1982. "Institutional Aspects of Integrated Planning Processes." In Y. J. Ahmed and F. G. Muller, eds., Integrated Physical, Socio-Economic and Envi- ronmental Planning. Dublin, Ireland: Tycooly International Publishing. PRO C E ED ING S OF T HE W OR L D BANK ANNUAL CON FE RE NCE ON DE V EL O PM ENT EC ONO MI CS 1 9 9 0 COMMENT ON "REGIONAL SUSTAINABLE DEVELOPMENT AND NATURAL RESOURCE USE," BY NIJKAMP Kirit S. Parikh Professor Nijkamp's paper raises some difficult but very important issues con- cerning sustainable development at the regional level. Somewhere between the Stockholm conference of 1972 and the latest report of the Brundtland Commis- sion (WCED [World Commission on Environment and Development] 1987), a global consensus emerged on the need for sustainable development, and both developed and developing countries now endorse it. At the Stockholm conference the environmental problem was seen largely as a problem of pollu- tion, and Indira Gandhi summarized the viewpoint of developing countries when she asked, "Are not poverty and need the worst polluters?" The implica- tion was that taking care of poverty would solve the problems of pollution in developing countries. Since then, developing countries have noticed that devel- opment has not taken care of their environmental problems and that issues of air and water pollution, urban slums, soil degradation, desertification, and deforestation are widely present today. Moreover, these problems have the most severe effects on the poorest segments of society. Thus, developing countries have recognized that not all development is desirable and that only sustainable development is worthwhile. Although industrial countries have made significant progress in cleaning up their own pollution of air and water, they have been made more keenly con- scious of the need for development that would sustain the global commons by the emergence of global environmental problems such as the greenhouse effect and the hole in the ozone layer. Still, what is meant by sustainability is not yet clear. The definition given by the Brundtland Commission, which I come to shortly, needs to be translated into an operational, measurable, and monitorable concept. Sustainability is essentially a holistic notion, and all aspects of the system need to be included in it. It is therefore difficult to define precisely. Defining sustainability for a sector, a specific resource, a subsystem, or a subregion is even more difficult. Yet, even when one thinks globally, one has to act locally. It is only local populations that are likely to have the necessary knowledge and understanding as well as the ability to take the actions needed to preserve the sustainability of the local Kirit S. Parikh is the director of the Indira Gandhi Institute of Development Research, Bombay, India. K 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 189 190 Comment resource base. It is therefore essential to define the concept of regional sustain- able development and to provide a framework for guiding local action. Nijkamp's paper presents a definition of regional sustainable development that involves only sustainable resource use (SRU), which he does not define. He then proceeds to provide a framework for developing a sustainable development plan for a region, in the course of which various factors that one might consider as necessary conditions to promote sustainable resource use are to be identified. These factors he calls the critical success factors (CSFS). CSFS may be looked upon as the factors that help monitor whether development is proceeding along a sustainable development path. Nijkamp also outlines what analytical models for sustainable development should look like. The special institutional arrange- ments and policies needed to implement a sustainable development plan are also discussed, and three case studies are presented to illustrate his ideas on regional sustainable development. Perhaps because Nijkamp's paper covers such a large scope, he has not pro- vided details at the level one would have liked. My comments on his paper, therefore, may be unfair. The first problem I see is the difficulty of defining sustainable development at a regional level. Nijkamp defines sustainable development as "development that ensures that the regional population can attain an acceptable level of welfare- both at present and in the future-and that this regional development is compat- ible with ecological circumstances in the long run while it also tries to accom- plish a globally sustainable development." In a sense, this is an extension of the definition given in the Brundtland Commission report, which defines sustainable development as the development that meets the needs of the present without compromising the ability of future generations to meet their needs. It also recog- nizes that sustainable development is not a fixed state but a process of change in which exploitation of resources does take place and that it is not realistic to suppose that every resource can be kept intact everywhere. The question is at what level should one think of sustainability and over what period. Should sustainability of forests, say, be defined at the level of the globe, nation, region, forest, hectare, or tree? Obviously, it is impossible to preserve every forest or tree. Spatial and temporal depletion of resources must take place, and one has to recognize that one can also restore some resources, augment them, and improve their quality. If we were to try to preserve every resource base for all regions for all times, we would have to forgo so many opportunities that development would become extremely expensive. When one recognizes this possibility-of spatial and temporal exploitation of resources-the problem of defining regional sustainable development becomes obvious. Does one assume that the population of the region should remain the same? What level of migration in and out of it should be considered a tolerable limit for sustainable development? What about the imports of commodities and resources from outside the region? Once one permits such imports into a region, it is easy to ensure that the development within the region remains sustainable. As Nijkamp himself points out, the high productivity of Dutch agriculture Parikh 191 involves Thailand's exports cassava to Netherlands. Dutch livestock activities may be sustainable as a result, but is Thailand's cassava production sustainable? Seasonal migrants from Athens can appropriate the gains from development of tourism of the Sporades Islands, leaving little gain for the islanders. If one does not care what happens outside the region, one can make its development sus- tainable. Every region in this sense can have regional sustainable development, but such development need not be globally consistent, and it would not result in a globally sustainable development. This problem of defining the boundary conditions or the conditions of trade with other regions is not addressed in Nijkamp's paper. Yet without such a discussion, it seems to me, the problem of defining regional development remains somewhat incomplete. Sustainability may be defined as the preservation of the production possi- bilities of an economy to produce the same useful goods and services, including services obtained from the state of nature. Such a definition could provide some guidance (see Parikh 1989) on the rate at which one can exploit exhaustible resources or the level of degradation of renewable resources that may be permit- ted. However, this raises another issue. In the production of such goods and services, man-made capital and natural resources (we can call the latter environ- mental capital) can substitute for each other, at least within some limits. Figure 1 shows that with the present technical knowledge and skills, different combinations of the two types of capital yield the same level of production and define the isoquant qo of production. Initially one may be at a point, P0, corre- sponding to the initial stock of nature, No and accumulated capital, Ko. If, as production takes place and some natural resource capital is used up, one remains on the isoquant qo, one still may say that one has sustainable develop- ment in the sense that the future is able to produce the same output. Thus, one may be at point Pl, where man-made capital stock has been augmented to the level of K1, which is used to produce the same level of output even though the natural resource is depleted, to the level of N1. Such a description illustrates the possibilities of substituting resources or natu- ral environmental capital by man-made capital to sustain output. But we should recognize that isoquant qo is a function of the state of human knowledge and skills. Thus, as technical progress takes place, the isoquant can shift inside to q1, in which case the same level of output as before can be produced with less of N as well as of K. Adequate account of these possibilities of technical progress and accumula- tion of human skills and knowledge is generally not taken in discussions of sustainable development, perhaps for understandable reasons. To the ecologists and environmentalists the notions of substitutability and technological change may ring too much of economists' sophistry, formulated to justify continued neglect of pressing environmental problems. Nonetheless, in a logical system, one ought to account for these possibilities-particularly the possibility that the introduction of new technology and techniques at the level of a region can significantly alter the nature of sustainable development. In a sense, Nijkamp recognizes all these limitations when he says that the 192 Comment Figure 1. Substitution between Man-Made Capital and Environmental Capital at Different Levels of Skills and Knowledge KA Stock of man-made capit Ko Production here produces the same output as at PO with less of both K and N Isoquant qo (with initial skill \ ~~~~and knowledge) Isoquant q0 (with higher skill and knowledge) N1 No N Stock of natural resources (or environmental apital) regions are open systems and that sustainable development at the global level may even require sacrificing certain regions. He also recognizes that the carrying capacity of a region is essentially a dynamic concept and that it depends on biophysical behavior as well as on economic activities. Thus, the question that Nijkamp's paper neglects to answer-how does one define in an operational way the boundary conditions for a region-becomes particularly important. Let me now turn to the operational aspects of the paper. The CSFS, which are defined as the necessary conditions to promote SRUS, are to be identified in a process that involves delineating the characteristics of the region, including its classification based on its socioeconomic circumstances, phase of development, policy of consumption of its natural resources, and generation and depletion rates of its resources. This should be followed with a stocktaking of the charac- teristics of industrial structure of the region and its interaction with other regions and its relevant global impacts. This is to be followed by a third step of assessment of both probable and possible developments and finally by an identi- fication and evaluation of different development paths. Unfortunately, other than listing these steps in the process, Nijkamp gives no accounting of the Parikh 193 conceptual difficulty or the analytical processes needed to do this work. In a sense, he provides us with a checklist of things that need to be done but not a full recipe for how to do them. He tells us what are the sights on the way but does not provide the route map. Perhaps Nijkamp's section, on the models of sustainable development can be looked upon as providing some indication as to how one goes about filling his prescription. As Nijkamp rightly points out, the most significant elements needed in models for sustainable development-which distinguish them from other models used for environmental problems-are two: a module that describes the dynamics of resource bases and ecosystems, and a module that captures the ecological feedback impacts of economic activity on the economic system. These aspects are rightly emphasized, but, unfortunately, Nijkamp does not underline the difficulties of constructing such models. Many of the ecologi- cal, biological, physical, and agronomic processes are poorly understood. To incorporate these aspects into models of development will require enormous efforts of communication across different disciplines. Some years ago, in trying to develop a framework for sustainable agricultural development (see Parikh and Rabar 1981), I was frustrated by the reluctance of soil scientists to extend their knowledge of soil science from the level of a field to that of a region. In exasperation, I posed one of them a question: "If you can generalize from your knowledge based on soil samples that are 3 centimeters in diameter to the level of a field, why are you reluctant to extend this knowledge to a larger region?" I was met by the response, "Perhaps we are being unscientific when we generalize from a sample to the field." How difficult and complex applied studies on sustainable development become can be seen in the case studies reported in J. Parikh (1988). I should mention that Nijkamp does list the different kinds of models, the multiplicity of techniques that one can use, and the role of models in obtaining an understanding in the alternative strategies of sustainability for a region. Finally, let me turn to the case studies Nijkamp provides to illustrate the use of the ideas and approaches he describes in the first five sections of his paper. The case study of the Peel region in the Netherlands is the only one he describes in enough detail for me to discuss. He notes the different interest groups in that region-the farmers, the drinking-water companies (representing also house- hold interests), the tourist industry, the commercial forest owners (the State Forestry Service and private owners), and the nature conservation lobby (among others the state as the owner of the national park de Groote Peel)-and describes how they compete for the region's resources. The region's natural resources used as goods are groundwater for drinking and irrigation and timber for paper and pulp. The ecosystems provide services of outdoor recreation and nature conservation while they assimilate surplus ammo- nium and nitrate from nature. A business-as-usual scenario serves as a reference for various alternative sce- narios Nijkamp develops: no import of SO2 or NO, from outside the region, 194 Comment continuation of present environmental policy, additional ammonia scrubbing, maximum technical effort, and shifts in land use pattern. For each of these scenarios, the paper describes the impact on concentration of nitrates in deep groundwater, total ammonia release from manure, the soil pH, the volume of shallow groundwater, the stock of different types of trees in the forests, the quality of grass, the volume of deep groundwater, the quality of air, and total value added in the region. The scenarios illustrate the possibilities of nature conservation and the trade- off between economic and ecological considerations under alternative policies. They raise a number of questions, however. How realistic or descriptive are the model equations? Is the technical knowledge sound? Are the parameters empiri- cally estimated? What policy instruments are available to realize any one par- ticular desired development path? The fundamental cause of many environmen- tal problems-that of externalities and possibility of free riders-requires that the behavioral responses of the different groups in the region are modeled in an appropriate way to reflect their responses to policy changes. It would be interest- ing to know how one can design a set of policy instruments that would imple- ment any of the desired solutions at minimal cost to the society. Land is the most important resource in many developing countries, and any models that deal with problems of land use and changes in the quality of land are of considerable interest to developing countries. Moreover, these problems are further complicated by the importance of distributional considerations in devel- oping countries. Such considerations are perhaps not of particular significance in the Peel area. Yet the Peel area did have a number of interest groups, and it would be interesting to know how the impacts of different policies on their interests were modeled and whether any policy was Pareto-optimal or not. Nijkamp has addressed difficult and important issues of how to translate the vague notion of sustainable development at the global level to a local action- oriented regional development plan. I only wish he had given us some guidance and details on how to do it rather than the mere tantalizing list of things that need to be done. REFERENCES Parikh, Jyoti K., ed. 1988. Sustainable Development in Agriculture. Amsterdam, Netherlands: Nijhoff. Parikh, Kirit S. 1989. "An Operational, Measurable Definition of Sustainable Develop- ment." Indira Gandhi Institute of Development Research Discussion Paper 21. Bom- bay: Indira Gandhi Institute. Parikh, Kirit S., and Ferenc Rabar, eds. 1981. Food for All in a Sustainable World: The iiasa Food and Agriculture Program. Report SR-81-2. Laxenburg, Austria: Interna- tional Institute for Applied Systems Analysis. WCED (World Commission on Environment and Development). 1987. Our Common Future. Oxford, U.K.: Oxford University Press. PRO CE ED ING S OF T HE W OR LD BANK ANN U AL CON FE REN CE ON DE V EL O PMENT EC ONO MI CS 1 9 9 0 COMMENT ON "REGIONAL SUSTAINABLE DEVELOPMENT," BY NIJKAMP William B. Magrath If I had known how difficult it would be to discuss a paper on sustainable development, I never would have accepted this assignment. Now, having thought about Nijkamp's paper, and Dasgupta and Maler's, I'm surprised that anyone would be willing to take on the job of writing one. Sustainability is one of those concepts that defies definition, and therefore it ends up with as many definitions as experts (of which there seems to be an infinite supply). As if sustainability alone weren't bad enough, Nijkamp has chosen to discuss it in the context of regional development and natural resource use. If I'd had to write a paper on sustainable development, I certainly wouldn't have wanted to write this one. Nevertheless, sustainability, regional development, and natural resource use are important topics, they are all interrelated, and I did manage to get myself into this, so I will try to raise some questions and suggest some issues that I think are important for the World Bank and for World Bank economists in particular. I should note from the start that I would go along with many of the recommendations Nijkamp offers, especially the need for natural resource accounting, for taking a holistic approach to problems, and for a multidisciplin- ary approach. Starting with my narrowest concern, I'm interested in exactly how Nijkamp quantitatively incorporates natural resources in his three case studies. Given the limitations of space, his description of the three models is understandably quite meager. As someone who also tries-and I admit with a great deal of discomfort-to find real numbers for his models, I would be very interested in how Nijkamp views the generation of model coefficients for processes such as drainage, groundwater recharge, agricultural productivity, erosion- productivity, and so on. I presume that essentially mechanistic relationships, drawn from the technical literature on agronomy, soil physics, and environmen- tal chemistry, are the basis for these models. This "engineering economics" approach is increasingly being used in applied natural resource economics, including World Bank work (see, for example, Dixon, James, and Sherman 1989; Anderson 1987; Bishop and Allen 1989; and Magrath and Arens 1989). Although I don't know of any practical alternative, I would like to see some William B. Magrath is a land resource economist in the Agriculture Division of the Asia Technical Department at the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 195 196 Comment careful analysis of the validity-both technical and economic-of this approach. Beyond the biophysical dimensions, which are complicated enough, some quite strict conditions must be necessary in the production functions if this engineer- ing economics approach is to be considered reasonable. Working out these necessary conditions would not be especially difficult. We know that elasticities of substitution are not zero, and if we are to have any confidence in the policy implications of models such as these, we need some work in this area. I would particularly like to know more about the technical efficiency of natu- ral resource use in these models. One of the major lessons that emerges from recent work on environmental issues is the importance of technical inefficiency. That is, on almost any environmental issue-energy output and the greenhouse effect, soil erosion, or deforestation-we know that it is possible to get more out of less using known technology. Energy is perhaps the best example. It takes twice as much energy to produce a unit of gross national product in the United States as it does in Japan, and four times the U.S. rate to produce one unit in China. Failing to account for the possibility of improvement within the frame- work of existing technology and policies overstates the costs of resolving envi- ronmental problems and makes agreement on further policy reform even more difficult than it needs to be. More interesting, if less concrete, questions are raised by the regional develop- ment focus of Nijkamp's paper. To be honest, I'm not sure what defines a region. In Nijkamp's paper a country can be a region, so can a chain of islands, and so can the southeast corner of the Netherlands. These regions are more or less open or closed, in terms of their links to other "regions," and both from a modeling and a policy perspective it is important to ask how closed, if at all, a space must be to qualify as a region. To provide some perspective on this question of what is a region, let me refer to the work the World Bank is just completing on watershed management in Asia. When we started that work we assumed that we would find that physical linkages from the downhill flow of water would provide an inescapable logic for integrated planning based on watersheds as regions. But over time we found that the technical scope for investment aimed at reducing sedimentation and flooding was severely limited (Magrath and Doolette 1990). Moreover, we have pretty much concluded that various institutional problems make integrated planning in all but the smallest watersheds a practical impossibility. Thus, even though environmental problems occur across space, geography does not always provide the right unit for addressing them. In the watershed work, because the farm was the unit most susceptible to influence by decentral- ized incentives, we essentially ended up with the farm as the unit of account, even though it is only partly a physical unit and even though we are concerned with erosion and sedimentation problems that clearly transcend farm bound- aries. In truth, the definition of a region, and whether it is a useful unit of account, all depends on the objectives of the analysis. Also in the context of defining regions, I'm interested in what Nijkamp thinks Magrath 197 about trade between regions and self-sufficiency. Nijkamp suggests that regional sustainable development "should ensure the regional population an acceptable level of welfare, which can be sustained in the future" and that "it should not be in contrast with sustainable development at a supraregional level." I wonder if this really requires that a region be self-sufficient. What if there is the possibility of trade with a nonsustainably developing region? Other analysts of sustainable development (for example, Daly and Cobb 1990) clearly associate sustainable development with self-sufficiency. They reject trade based on the principle of comparative advantage. Where does Nijkamp stand on this issue? Another way of putting this is to ask if there is some sort of adding-up requirement implied by Nijkamp's notions of regional sustainable development and globally sustainable development. If so, I think it is more than just a political issue, as he suggests; it seems to further undermine the value of regions as a unit of analysis of sustainability. Another concern about regions: Nijkamp's use of critical success factors (CSFS) conjures up an image of regional development as a linear programming prob- lem. CSFs are essentially constraints; they can have economic or ecological and local or far-removed dimensions. This is fair enough, conceptually quite useful, and not very unusual. My only problem is, who is to be charged with solving the problem? And once the problem is solved, how will the optimal solution be implemented? We all know that real-world physical problems rarely coincide with political boundaries. In the United States, the experience of the Water Resources Planning Council shows that regional planning of the type Nijkamp seems to envision is difficult to achieve and sustain. For example, one of my old employers, the Great Lakes Basin Commission (GLBC), was charged with providing planning services to the region comprising the nine states bordering the Great Lakes. The GLBC pro- duced an outstanding planning study for the region. However, the GLBC was closed down as one of the first acts of the Reagan administration, with barely a whimper from Congress or the states concerned. The entire planning exercise had no lasting impact on the region. Even if Nijkamp's planning function can be sustained, how will proposed solutions be implemented? Presumably, he expects that there would be some need for bribery or coercion, or that the optimal solution would emerge from the operation of market forces. It is worth thinking about the fact that the compre- hensive resolution of Botswana's land degradation problems in Nijkamp's exam- ple will involve changing the behavior of thousands of herders and bushmen. The role of centralized planning in such a situation is clearly limited to diagnosis of problems, but decentralized incentives or direct investments will prove to be essential to their resolution. Finally, just a few comments on sustainability. Nijkamp uses two, to some extent conflicting, definitions of sustainable development. At one level he seems to embrace the World Commission on Environment and Development (WCED 1987) definition that sustainable development is development that improves the 198 Comment standard of living of the current generation without compromising that of future generations. Further on, he describes sustainable development or a "Pareto- optimal trajectory in which progress in either the economic or the ecological system would not be to the detriment of the other system" (I tried to draw an Edgeworth Box to describe this, but had to give up). Finally, he seems to settle on sustainable development as something like Norgaard's (1984) notion of coevolution, in which changes in the environment and society induce still further changes in both themselves and in each other. Where does all this lead us? We've certainly not heard the last of sustainable development. I believe that the future is important and that running out of resources is worth worrying about. But the prospects for reaching an opera- tional consensus on what sustainability requires are, I think, quite slim. Aside from being intellectually interesting, the sustainability debate does offer the World Bank and economists a few opportunities for productive work. To realize this potential, however, we will need to recast the question. Instead of talking about "sustainable" or "unsustainable," we should be trying to distin- guish more from less sustainable. We seldom have the opportunity to put an economy on a sustainable path. Even if we did, I don't think the kinds of rules that have been put forward to describe steady-state economies would prove very useful. Dasgupta and Maler discussed the problems with Pearce, Markandya, and Barbier's (1989) "blue- print for a green economy" rule of maintaining natural resource stocks intact. The notion of countervailing projects, "green" projects to make up for dirty ones, an idea that is present in the World Bank's wildland policy, is also quite limited. Big projects only account for a small part of environmental degradation, and no one can seriously think it possible to micromanage all the potentially degrading activities in an economy. Like it or not, we need to focus on marginal improvements in the environ- ment. Economics is pretty good at working at the margins. We already have some perfectly serviceable analytical tools, and with some minor revisions and extensions, such as expanding national accounts to include natural resources correctly (something that really should not seem unusual), mainstream eco- nomics can go a long way toward understanding and alleviating environmental problems. REFEREN CES Anderson, Dennis. 1987. The Economics of Afforestation: A Case Study in Africa. Baltimore, Md.: Johns Hopkins University Press. Bishop, Joshua, and Jennifer Allen. 1989. "The Onsite Costs of Soil Erosion in Mali." World Bank Environment Department Working Paper 21. Washington, D.C. Processed. Daly, Herman E., and John B. Cobb. 1989. For the Common Good: Redirecting the Economy toward Community, the Environment, and a Sustainable Future. Boston, Mass.: Beacon Press. Magrath 199 Dixon, John, David James, and Paul Sherman. 1989. The Economics of Dryland Man- agement. London: Earthscan. Magrath, William B., and Peter Arens. 1989. "The Costs of Soil Erosion on Java: A Natural Resource Accounting Approach." World Bank Environment Department Working Paper 18. Washington, D.C. Processed. Magrath, William B., and John B. Doolette. 1990. "Watershed Management in Asia: Strategies and Technologies." World Bank Technical Paper 127. Washington, D.C. Processed. Norgaard, Richard B. 1984. "Co-evolutionary Development Potential." Land Eco- nomics 60, no. 2: 160-73. Pearce, David, Anil Markandya, and Edward Barbier. 1989. Blue Print for a Green Economy. London: Earthscan. WCED (World Commission on Environment and Development). 1987. Our Common Future. New York: Oxford University Press. PROCE ED ING S OF T HE W O RLD BANK ANN U AL CON FE RE NC E ON DE V EL O PM ENT E CON OM IC S 1 9 9 0 FLOOR DISCUSSION OF THE NIJKAMP PAPER A World Bank participant questioned the focus on sustainability in regional growth. An alternative operational focus, he suggested, might be on a strategy of regional resource use that could help the development of other regions. For example, nitrate exploitation in the Atacama Desert of Chile, as a depletion strategy, could have been a rational policy for developing the rest of the country during the nineteenth century. Depletion strategies can be rational, both eco- nomically and ecologically, he suggested. Another World Bank participant asked about the practical role of Peter Nijkamp's model in resolving disputes among the interested parties in the Peel region of the Netherlands. Did the model actually resolve some of the disputes, and did it put in place effective and sustainable policies? In his experience studying and participating in the resolution of resources disputes, the participant argued, fancy mathematical models built by experts were deemed suspect by nearly all of the disputants. Graham Pyatt (chair) noted that there may be factors other than the current international trade patterns that deserve consideration in explaining high agri- cultural productivity in the Netherlands. He recalled that Colin Clark's Condi- tions of Economic Progress had cited William Pettit on the Netherlands' high domestic agricultural productivity from around the end of the seventeenth century. Pyatt also commented on Nijkamp's Botswana case study and the observation made by William Magrath (discussant) that massive changes in herding behavior in Botswana would be needed to deal with land erosion. Pyatt's understanding of the situation was that it had turned out that the best way to modify herding behavior was to send children to school. Previously, the children had looked after the cattle; after the children began school, it was amazing how quickly electric fences went up. This was an interesting example of the interaction between the environment and household behavior. Nijkamp began his response to the discussants' comments by noting that he did not think it was meaningful to have a general discussion on terminology about sustainability. The interest and challenge of thinking about sustainability, he argued, came in considering open regional systems-that is, systems with international and interregional trade. It may be better to produce commodities in, or extract raw materials from, places where less harm would be done to the This session was chaired by Graham Pyatt, professor of economics at Warwick University, U.K. ( 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 201 202 Floor Discussion of Nijkamp Paper environment than others. So trade is linked to environmental advantage, not just to the standard notions of comparative advantage. Of course, Nijkamp continued, one could argue that these environmental costs should be internalized through sound economic policies, and the resulting market equilibrium would incorporate externalities. The real problem is that no institution or policy regime exists that can internalize all ecological costs. The question then is a practical one: do regions wait until such a comprehensive system or institutional arrangement comes into being, or do they go ahead and design and implement policies anyway? Nijkamp gave the example of the debate in the Netherlands about the import of timber from Malaysia and Thailand. The destruction of forests in those countries had motivated parliamentary discussion, and it was thought that the Dutch should use domestic raw material substitutes for tropical wood. How- ever, people soon realized that stopping the import of timber might not solve the problem of deforestation. Timber producers abroad might respond by lowering prices and seeking new markets. Depending on demand elasticities, the timber producers might actually end up depleting more forests as a result of a Dutch policy of substituting domestic raw materials in housing construction. So, inter- national negotiations, the price mechanism, and many other institutions are required in taking account of environmental costs, but even this may not be fully sufficient to internalize them. Nijkamp mentioned that there are dynamic concerns about the definition of sustainability. If one considers environmental capital per head and uses the Pearce definition of sustainability, discussed by Dasgupta and Maler in their conference paper, one may then need to consider the expansion of environmen- tal capital to keep pace with population growth. Public preferences regarding the environment are also not stable, and they vary across regions. As an exam- ple, Nijkamp cited the fact that environmental concerns were pronounced in the mid-1970s, were relatively muted at the beginning of the 1980s, and are very high now. Policies have to respond to these popular and political swings. Nijkamp then turned to the issue of technological change, which he thought was a greatly underestimated factor in sustainability. However, the important question was: which technologies are important from the point of view of sus- tainability, particularly regional sustainability? It was not enough to speak of the average rate of technological change in aggregate terms. Nijkamp said that the economist's problem lay in trying to measure welfare with a very narrow economic measure-gross national product per capita. This single measure can hardly capture the whole range of welfare effects, including equity and environ- mental considerations. It is quite evident, Nijkamp said, that the issues cannot be undertaken by an economist alone: major contributions have to come from environmentalists and resource specialists. Yet the reconciliation of methodologies and the scope of different disciplines is a major task. The problem of coordination becomes Floor Discussion of Nijkamp Paper 203 pressing for ecological phenomena, which never coincide with our administra- tive boundaries, be they regional, national, or even international. On the case studies, Nijkamp noted that one of the most difficult tasks is the problem of choosing the policy or control variables. It may sound simple, but in practice, it becomes a very difficult, interactive question. To Magrath's (discus- sant) question about whether regional sustainability as defined by Nijkamp would also be Pareto-optimal, Nijkamp replied that he agreed with Magrath that the important thing was to distinguish more sustainable from less sustain- able. Nijkamp noted in this context that the recent literature makes a distinction between "weak sustainability" and "strong sustainability." Weak sustainability would exist if one made considerable progress on one objective-say, growth- while progress on other objectives such as environmental preservation suffered, but by not too much, so that the net result for social welfare was still positive. He said that his model was about weak sustainability. Of course, he added, everything depends on how social welfare is defined and estimated, and the estimation of social welfare functions remains relatively poor in economics, so perhaps we have to depend on decision theory and operations research for empirical methodologies. That was why he had introduced multi-objective deci- sionmaking in his models. A participant from the International Monetary Fund, referring to the confer- ence paper by Dasgupta and Maler earlier in the morning and their point about the tie between poverty and environmental degradation, wondered how poverty and the environment interacted, because environmental deterioration and activ- ities such as the search for water and firewood are both outside the national income accounts. Dasgupta replied that one had to be somewhat catholic in the use of data. There are many activities that are not going to be counted, but in the case suggested, the amount of time spent by women and children in collecting fire- wood or carrying water provides some signal about the state of the environment and the state of the resource base in that community. So time use or calorie use data are needed to supplement the national or regional income data. Maler said that national income accounts could include estimates of the value of time and energy used in activities such as collecting fuelwood and fetching water by deriving appropriate accounting prices based on production functions for these activities. A Bank participant noted that household sample surveys provide much infor- mation useful to environmental policy analysis. The challenge is to make effec- tive, systematic use of, for example, time-use surveys and to show that the derived indexes for policy analysis are not sensitive to the large measurement error that comes from using data that originate outside the market. To do that, one's interest must be focused on detecting change rather than levels as such. On this issue, Pyatt observed that the primary problem was that the economists generally did not demand this kind of data from statisticians. 204 Floor Discussion of Nijkamp Paper A participant found the lack of any discussion of population growth in either of the two conference papers puzzling, because one constantly hears strong claims about its impact on global ecology. Put another way, there could be a large negative externality from reproductive decisionmaking, because environ- mental assets are collectively owned. Dasgupta agreed that population growth was obviously important. He and Maler had not dealt with it simply because there were to be papers on popula- tion specifically at the conference. Nijkamp also agreed with the importance of incorporating population growth in ecological models; however, to do so, he warned, will require tremendous amounts of new data-for example, on the use of time and the resources of people more than sixty-five years old for countries with aging populations. A Bank participant wished to follow up on the point about threshold resource use levels; he felt that besides the measurement reasons Maler had cited for deriving such levels, there were other very practical and operational reasons for doing so, particularly in irreversible situations. What if a seemingly prohibitive price put on a particular resource to prevent its further use did not, in fact, ultimately prevent its use because someone was willing to pay that price? Often, to deal with such circumstances, a threshold level makes more sense, the partici- pant argued. Maler agreed that in many circumstances an environmentally con- scious development policy based on rules of thumb could be preferable to relying on the price system. Another Bank participant said that authors of the papers under discussion seemed to have different ideas on the substitutability of capital and natural resources: Maler seemed to think that it was an empirical question; Dasgupta seemed to say that there was not much substitutability; and Nijkamp suggested that one's beliefs about the degree of substitutability would influence the defini- tion of sustainability. The questioner asked for an elaboration of their positions. On the question of defining sustainability (either as nondecreasing per capita welfare or a nondecreasing natural resource stock over time) and the issue of substitutability, Maler argued that the basis should be welfare. If substitutability between natural and man-made capital is nil, he said, then we should preserve natural resources; this is a consequence, however, not the basis for the definition of sustainability. He reiterated that there is no general answer on substitutability of capital; it depends on the specific circumstances. Dasgupta said that spending a lot of time arguing whether there was or was not substitutability, or what it might mean, is not likely to yield much progress. Dasgupta argued instead for paying attention to why existing social arrange- ments may not be right in the context of the environment, and what the econo- mist's tools tell us in terms of the policies to be followed in such cases. Regarding substitutability between natural and man-made capital, Nijkamp said that much depended on the time horizon being considered. A pesticide ban in the short run may lead to reduction in agricultural activity, but it may pro- Floor Discussion of Nijkamp Paper 205 duce favorable long-run environmental conditions that are also a necessity for agricultural production. In closing, Nijkamp reiterated the central importance of incorporating social externalities, through shadow prices, to the maximum extent possible in pro- duction, technology, and factor choices. Too often economists wait until exter- nalities begin to appear, and then it is very difficult to internalize them into the market mechanism. Pyatt thanked the speakers, discussants, and conference participants for their stimulating contributions. PRO CE ED ING S OF THE W OR L D BANK ANNUAL CO NFE RE NCE ON DE VE L OPME NT EC ONO M ICS 1 99 0 The Soft Underbelly of Development: Demographic Transition in Conditions of Limited Economic Change John C. Caldwell Population policy is driven by two sets of theories, which are examined here. Thefirst, demographic transition theory, proclaims the inevitability of a global transition from high to low birth and death rates as a world society and economy develop. Experience to date supports this generalization, but researchers have had little success in identifying threshold socioeconomic or demographic indexes that would predict the onset of fertil- ity decline. The second, demographic-economic theory, predicts that slower population growth rates will lead to faster rises in per capita income. The historical evidence does not confirm these predictions, however. This paper examines the case of Sub-Saharan Africa, where mortality has been falling consistently, although often driven more by educational and other social changes than by rising incomes or enhanced medical care, and halted only by war and civil disorder. However, declining fertility in Sub-Saharan Africa is not a certainty, despite population projections by international organizations that assume that widespread decline is imminent. Evidence indicates that African fertil- ity is peculiarly sensitive to infant and child mortality levels, and therefore amenable to decline through service-intensive combined health and family planning programs; but it is not clear that governments can implement such programs on a national scale. On the demographic-economic nexus, some evidence suggests that in Sub-Saharan Africa high levels of natural population increase do slow growth in per capita income. Family planning programs might also catalyze changes in family social and economic structures that would contribute to faster economic growth. Finally, as the full scale of the African AIDS epidemic becomes clearer, it will come to dominate all other population-related concerns, creating a compelling case for huge expenditures on comprehensive health, family planning, and social services aimed at reducing mortality, curing sexually trans- mitted diseases, detecting HIV infection, and providing follow-up services. The population element is described here as the "soft underbelly" of develop- ment only in the sense that until now its theoretical relationships to development have not been sustained by empirical testing. Thus, although policy interven- John C. Caldwell is professor of demography at the Australian National University, Canberra, Austra- lia, and director of the Health Transition Centre at the National Centre of Epidemiology and Population Health at the Australian National University. The author thanks Betty Kavunenko and Jennifer Braid for research assistance and Mary Gilmour for typing. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 207 208 The Soft Underbelly of Development tions in demographic matters probably are necessary and important, policy- makers have lacked certainty and direction in formulating and implementing them. Debate on population policy has always been fiercest about the propriety and likelihood of success in controlling fertility-so much so that population policy and fertility control policy have often been regarded as synonymous. However, clearly mortality matters as well, and this paper deals with both mortality and fertility changes. Regarding mortality, the important questions this study addresses are whether continuing decline is occurring; if so, what drives it; and what impact the spread of acquired immune deficiency syndrome (AIDS) is likely to have. Regarding fertility, the key questions are whether fertility is likely to fall at all; whether there are advantages in economic development to be gained from declines in fertility; and whether policies adopted successfully elsewhere in the world to achieve declines in fertility are relevant for Africa. The paper examines the general questions regarding relevant demographic theory and evidence in this order: (1) Is it certain that fertility will soon be falling everywhere in the developing world? (2) Is it certain that economic gains will accrue from such decline? (3) What drives mortality decline? (4) What do local studies in Asia show about why populations begin to control their fertility? (5) Can intensive small-area family planning programs lower the birth rate even in areas where there has been only limited socioeconomic change? The major geographic focus of the study is on Sub-Saharan Africa and in particular on whether the decline in mortality is continuing there (in contrast to much-publicized assessments that the mortality decline is, in fact, slowing to a halt) and whether, or after the implementation of which policies, fertility is likely to fall at all. A salient reason for focusing on Sub-Saharan Africa is that the demographic indicators for this region are among the most adverse in the world. Life expec- tancy in Sub-Saharan Africa is barely fifty years, compared with fifty-five years in South Asia, sixty years in North Africa and the Middle East, more than sixty years in Southeast Asia, and more than sixty-five years in East Asia and Latin America (and seventy-five years in industrial countries). Fertility in Sub-Saharan Africa averages more than 6.5 children per woman, compared with 4.8 in South Asia, 5.5 in North Africa and the Middle East, 3.9 in Southeast Asia, 2.3 in East Asia, and 3.7 in Latin America (and 1.7 in industrial countries). More signifi- cantly, Sub-Saharan Africa is the only major region in the world that so far has shown no certain signs of a decline in fertility. Demographic expectations and plans for Africa usually cannot be understood solely in terms of African evidence. In Asia-especially in Southeast and South Asia-the first fertility declines often were achieved in the late 1960s with gov- ernment assistance and at income and mortality levels similar to those of a range of African countries at that time. Comparative data on Latin American fertility Caldwell 209 are not very useful for Africa, however, because the fertility transition in Latin America from the mid-1960s onward was largely a spontaneous social move- ment in circumstances similar to those obtaining decades earlier in Europe. Usually, then, expectations for Africa are largely derived from assumptions based on Asian precedents or on theories largely formulated from the Asian experience. This paper thus attempts to identify and build on the theories and Asian experiences that are most likely to influence policy planning in Africa. Besides the five questions above, the paper poses the following questions on Sub-Saharan Africa: (6) Is mortality decline-significant both as a developmen- tal gain in its own right and as a basis for the control of fertility-faltering, and has the potential impact of AIDS been addressed adequately by policymakers? (7) Are African societies more pronatalist than other societies, and are there mate- rial reasons why this should be so? (8) How is African fertilitv determined? (9) Why has Africa failed to implement fertility control programs that exhibit the Asian pattern of success? (10) Given the foregoing problems, are current popula- tion projections providing us with a realistic guide? (11) Given also that there are thirty-five independent countries with populations exceeding 1 million in Sub-Saharan Africa-thus allowing national comparisons on a scale impossible anywhere else-does the evidence support the case that material economic advantages accrue from slower rates of population growth? (12) What does the future hold for Sub-Saharan Africa, and what should be done to make it the best future possible? I. GENERAL QUESTIONS Population interventions have been based on two interrelated groups of theo- ries. The first centers on the notion of demographic transition. This theory describes and seeks to explain a process whereby, as economic development occurs and as a global economy (and perhaps a global society) comes into being, both mortality and fertility ultimately and inevitably will decline to low levels. Demographic theories have interested development planners who have come to suspect that as the trends in global economic development and population proceed, demographic change-especially declines in fertility-may not be a purely dependent variable and that such change, particularly fertility declines, may be self-sustaining and may accelerate economic development. The second group of theories, demographic-economic theories, purport to show how such interrelations work. Is It Certain that Fertility Will Soon Be Falling Everywhere in the Developing World? The concept of demographic transition-the parallel fall in both fertility and mortality rates-gained currency in the early twentieth century, as theorists 210 The Soft Underbelly of Development began to generalize about observations of declining mortality and fertility in France and in other Western countries. ' Undoubtedly the greatest impact during the last forty-five years on the atti- tude of development planners toward population derived from the work done during World War II by a group headed by Frank Notestein at Princeton Univer- sity's Office of Population Research. Notestein and his team carried out a series of studies of national population trends, first for the League of Nations and then for the U.S. Department of State (see Notestein 1943; Notestein and others 1944; Davis 1951; Kirk 1944, 1946; Taeuber 1958; and Lorimer 1946). They concluded that all societies would eventually reach low levels of fertility and mortality. At the 1943 Hot Springs Conference, which created the basis for the U. N. Food and Agriculture Organization (FAO), Notestein proposed some of the early ideas on demographic-economic theory, particularly that economic and social development not only would bring down fertility and population growth but would eventually "help stimulate parents to new aspirations for themselves and their children-aspirations that are incompatible with large fam- ilies" (Schultz 1945, p. 210). Blacker (1947) placed even stronger emphasis on demographic transition as beginning and ending with zero population growth, a notion that has reappeared in recent years, once again with little more justifica- tion than the appeal of symmetry. An attempt to formulate demographic transition theory's postulated relation- ship between development and the onset of fertility decline in terms of economic and social indexes was made by the United Nations (1965) in terms of what has become known as threshold theory. To determine thresholds, the United Nations study examined twelve measures ranging from per capita income to urbanization, mortality levels, female literacy, and even cinema attendance. Such an approach has a theoretical attraction, but empirical testing has shown it to be of little value. In fact-although the authors were reluctant to admit it- the real value of the U.N. exercise was to discount completely the possibility of generalizable thresholds across societies. The research showed that the decline in fertility had begun in some societies at one-third the per capita income level that had been needed in others and that the finding held with regard to energy consumption at one-ninth the level, urbanization at one-seventh, and infant mortality at one-fourth. Even the more culturally homogeneous population of Europe, studied by the Princeton Office of Population Research project of the 1970s, did not yield a set of meaningful thresholds (Coale and Watkins 1986). Studies of historical data at Princeton and elsewhere have permitted compari- son of the onset of fertility transition in Europe, which took place generally in 1. French concern about low birth rates was mentioned frequently in both the French- and English- language press of the period. Willcox (1916) formulated a demographic transition theory from American experience, and Knibbs (1917) published data showing the course of a global transition. Knibbs (1917) and Pearl (1939) produced mathematical formulations of the path of population growth during the transition. Thompson (1929) began to characterize the necessary stages of transition. Caldwell 211 the late nineteenth century, with the onset of fertility transition in Asia, in the middle to late twentieth century, and to demonstrate how different the two are in terms of threshold indexes of fertility decline. Knodel and van de Walle (1979) compared European economies, with fertil- ity declines beginning in the period 1882-1900 (as measured by a decline of 10 percent in fertility), with Taiwan and Thailand, with onsets in 1963 and 1970, respectively. Two differences in the European and Asian samples are striking: the European economies began their transitions with much smaller initial family sizes and with much higher infant mortality rates. Overall fertility in Europe was two-thirds the Asian level, as measured by an index of fertility (If, based on the birth rate of the Hutterite religious sect) of 0.32, compared with the Taiwanese and Thai samples of 0.42 and 0.51, respectively.2 European infant mortality ranged between 150 and 200, as measured by deaths during the first year per 1,000 live births, whereas the rates for Taiwan and Thailand were 49 and 77, respectively. Europeans of the late nineteenth century thus were far less worried that subsequent mortality would erode family size or even result in childlessness. The other indexes are not particularly instructive as indicators of fertility decline. At the onset of transition, 20 to 40 percent of Europeans lived in towns with more than 20,000 inhabitants. Taiwan, too, fell within this range, although Thailand had a lower level of urbanization. Nevertheless, the sim- ilarity may mean only that we are considering a rather homogeneous range of European economies. European illiteracy levels were generally below 20 per- cent, and Thailand, too, fell within this range at the time of transition, although Taiwan was at 30 percent. Yet this may indicate little more than the importance of biblical literacy in some Northern European economies or the greater diffi- culty in achieving literacy in the ideographic Chinese writing. F. van de Walle (1986) showed that there was some doubt about whether low child mortality was a threshold for fertility decline, or low fertility a threshold for mortality decline, and concluded that both occurred in Europe with modern- ization. Coale (1973, p. 65) concluded that the European project had shown that (1) fertility must be within the calculus of conscious choice; (2) reduced fertility must be seen to be advantageous; and (3) effective techniques of fertility reduction must be available. The 1950s saw a transitory interest in the macroscopic aspects of demographic-economic interrelations. Leibenstein (1957) and Nelson (1956) argued that high levels of population growth could prevent both economic takeoff and the onset of demographic transition. But rapid economic growth in parts of East and Southeast Asia-despite high rates of population growth- undercut the notion of a demographic trap (that is, a situation in which any economic growth would lower mortality levels, thus speeding up population 2. If is a fertility-weighted index that assumes maximum possible fertility as that of a religious sect in the United States in the 1930s, the Hutterites, who did not control their fertility and for whom the index is set at 1.00. The index was developed by Coale (1967). 212 The Soft Underbelly of Development growth to a point where the original per capita income and mortality gains were nullified-essentially Malthus's position). The last fifteen years have seen a strong revival in interest in the economic determination of fertility decline. The revival focused on the economics of the family as a way of explaining why many societies were failing to seize the opportunity provided by the new family planning programs to control family size. The central argument was that in these societies each child represented a net economic gain to the parents because of such contributions as child labor or support to parents in their old age, especially in circumstances in which child rearing was not very expensive. The focus was often on both the conditions of stable high fertility and the circumstances when instability-that is, fertility declines-first occurred. Easterlin (1975; Easterlin, Pollak, and Wachter 1980) placed the transition in a "market" context, whereby before transition the demand for children was higher than their supply and where the cost of reducing that demand included the cost of practicing contraception-both economically, in terms of money and time, and psychically, in terms of the disapproval of relatives and the commu- nity. Becker (1960), Schultz (1969), and others of the New Household Eco- nomics School saw transition as occurring when the cost of children became greater, and they included not only expenditure on the children but the costs to mothers, both in income forgone and also in the value of their time, even in terms of non-income-generating time. Caldwell (1976, 1978, 1982), in the wealtb flows theory, argued that fertility decline began when there was a reversal of the net flow of resources-toward children rather than parents-but that this economic change was the result of social changes that concentrated greater family concern on the children. Ben-Porath (1980) provided an economic analysis of stable high fertility, arguing that in this situation family transactions were often preferable to exter- nal transactions in markets, and implying that fertility transition began as the market competed ever more effectively with family-produced goods and services. Cain (1981) emphasized the value of children as risk insurance for their parents and argued that fertility decline was much less likely in rural Bangladesh than in rural India because the greater development of capital markets and public relief employment in India substituted for some of the assistance that could be provided only by children in Bangladesh. Nugent (1985) asserted that the whole question of old age security and its relation to fertility was far more complex than current theorizing or field research suggested. Most of these theories emphasize that fertility decline arising from social change affects the family's economic calculus, as does market penetration. There are vaguer suggestions that market penetration may accelerate social change and that family change may allow swifter market penetration. The chief assault on theories citing economic and socioeconomic determinants has come from those maintaining the importance of the diffusion of ideas about controlling fertility or even of knowledge of and access to contraception. Caldwell 213 At first, the "innovationist" interpretation was discounted by social scientists. For example, Stix and Notestein (1934, 1940) concluded from an early study of a family planning clinic in New York City that everyone already knew the so- called natural means of fertility control and hence that the onset of fertility transition was the result not of the diffusion or legitimation of birth control ideas or the greater availability of contraception but of broader socioeconomic phenomena. Later, anthropological demographers countered the innovationist argument by advancing the view of an "ancient" practice of family size limitation-that is, maintaining that even pretransitional societies had practiced deliberate fertility control. This perspective was accidentally reinforced by the Princeton European indexes of marital fertility (see Coale 1967), which showed that birth rates in nearly all societies were below the theoretical maximum. For example, Leiben- stein (1975, p. 2) concluded that rates well below the Hutterite level show that even in circumstances where fertility is traditionally high, there is a "consider- able degree of control." In fact, however, most pretransitional populations have fertility levels well below those of the Hutterites-even when they are not making decisions to control family size. Such "natural fertility" regimes were the general case, even in England as late as the first half of the nineteenth century, as Wilson (1984) demonstrated. Thus, the Hutterites are not a good comparator population because they were a twentieth-century American population, well fed, with high levels of protein intake and declining durations of breastfeeding-all conditions conducive to high birth rates that did not necessarily obtain for other popula- tions. The innovationist viewpoint has gained renewed currency as critics of the socioeconomic determinist viewpoint have poked holes in that theory's concep- tions of fertility control. According to Caldwell, Caldwell, and Caldwell (1987), the notion of "ancient" fertility control was based on successive misinterpreta- tions of the evidence. Moreover, data from ongoing family planning programs and historical research have increasingly validated the innovationist perspective. Coale (1973, p. 62) agreed that the evidence from the European project that illegitimate fertility had fallen at about the same rate as legitimate fertility in Europe between 1850 and 1930 gave strong support to the innovationists. Mauldin and Berelson (1978) concluded that about 55 percent of the 1965- 75 fertility decline in developing countries could be explained by the new avail- ability of contraceptives, but they attributed two-thirds of this effect to latent demand rather than to the idea that providing contraception itself changed viewpoints and behavior. A new review focusing on the 1986-87 period comes to a similar conclusion (Westoff, Moreno, and Goldman 1989). Further buttressing the innovationist case, studies of several Asian popula- tions have shown that respondents attributed their own ability to control fertil- ity to the availability of family planning facilities and that they attributed the inability of their parents to do so to the lack of such facilities (Caldwell, Reddy, and Caldwell 1985, 1988; Basu 1984; Tuladhar 1987; Jain 1989). However, 214 The Soft Underbelly of Development Caldwell, Gaminiratne, and others (1987) showed that in Sri Lanka, because the demand for fertility control emerged long before the government provided fam- ily planning facilities, people gradually had resorted to withdrawal, rhythm, and abortion to limit births, as also happened in the West in an earlier period. In a self-styled iconoclastic but nevertheless persuasive view, Cleland and Wilson (1987) surveyed both the historical and contemporary evidence, espe- cially the World Fertility Survey (WFs) data, and concluded "that attitudes toward birth control, broadly defined, are of central explanatory importance for the timing of fertility transition" (p. 29). They identified the need for the prior existence of a latent desire for fertility reduction, and clearly regarded the avail- ability of contraception as actualizing and legitimizing that demand. They agreed with Caldwell and Ruzicka's (1978) conclusion, from their analysis of the Australian experience, that whole societies moved at a rather similar pace to reduce their fertility. This view is in conflict with the embourgeoisement theory, enunciated by Lesthaeghe and Surkyn (1988), which places emphasis once again on the essential leadership of upper-class inrovators. The conflict among social theorists about the onset of fertility decline is now more in emphasis than in substance. Most agree that social change can be accelerated by providing contraception and legitimizing its use, so long as the economic calculus is not completely against fertility decline; that is, so long as there is some latent demand. The major controversy is beginning to focus on whether there is already any significant latent demand in Sub-Saharan Africa. Researchers have gained experience over the last two decades with fertility transitions that go far beyond the availability and legitimation of contraception. Although social scientists have tended to regard these cases as abnormal and have conspicuously failed to bring them within demographic transition theory, these measures warrant serious consideration. At their mildest, such additional measures include incentives. Satia and Maru (1986) concluded that in India incentives and disincentives had some impact. The evidence from China's expe- rience over the last dozen years is that Asian societies with traditions such as China's, and governments of its type, can reduce fertility to levels well below anything suggested by latent demand. India achieved no less remarkable declines of fertility during the Emergency of 1975-77, and only the return of democratic government prevented India's fertility from following a similar course to that of China. A less extreme case is provided by indonesia. There, strong government lead- ership has been reflected in considerable local pressure in securing fertility decline. It is doubtful whether such programs would work, or whether govern- ments would survive their imposition, either in the Moslem and Christian heart- lands or in the very different circumstances of Sub-Saharan Africa. Perhaps the most general conclusion that can be made is that there is no specific developmental threshold at which fertility begins to fall spontaneously: France was probably an example of unusually low thresholds; Africa may prove to have some high thresholds. Government intervention, if strong enough, can Caldwell 2 15 produce fertility decline at any level of socioeconomic development. The real question is whether such governments are likely to survive and whether their subjects or planners would feel that the achievement was worthwhile. Is It Certain that Slower Population Growth Brings Economic Gains? Theories of Demographic-Economic Interrelations The oldest thesis on demographic-economic interrelations is that of carrying capacity, especially in terms of the food supply. Demeny (1988) has pointed out that this view is at least as old as Ecclesiastes (5 :11): "When goods increase, they are increased that eat them." The most famous modern variant of the theory is that of Malthus, which depended not only on ceilings to carrying capacity but also on a concept of diminishing returns. Both concepts exist in the contempo- rary ecological literature. One would imagine that a test case would be provided by Sub-Saharan Africa, given its trend toward importing an ever greater propor- tion of the food consumed and the occurrence of successive famines. Somewhat surprisingly, the 1982 FAO-International Institute for Applied Systems Analysis study (Higgins and others 1982, p. 34) claimed that the carrying capacity of the continent was 1.6 times the population projected for the year 2000 with no improvement in agriculture; 5.8 times that population with improved tools and modest inputs of fertilizer and pesticides; and 16.5 times with the full develop- ment of industrialized scientific agriculture. These ratios can be compared with the eventual multiplication of the continental population projected for 2000 of only 3.5, anticipated by the most recent World Bank population projections to be fully attained only after more than another hundred years (Bulatao and others 1989). These views are in direct conflict with the analysis of Allan (1965), who reported twenty-five years ago that shifting cultivation was already encroaching dangerously on the fallow with resultant declining soil fertility. In contrast, Mortimore (1968) demonstrated that manure and other inputs had permitted the establishment of relatively dense sedentary agricultural populations in the grassland soils around Kano in the West African savanna. Probably the demographic-economic analysis with the most influence on modern development activities was that of Coale and Hoover (1958). They argued that high rates of population growth led to slow increases in per capita income because of the disadvantages of high population densities; of the impact on capital accumulation; and of the maintenance of a less than optimal age distribution. The heart of this argument was that the higher the rate of popula- tion growth the higher the proportion of saving that had to be spent on creating new social infTastructure such as schools for the growing population, or on duplicating productive capital rather than increasing the ratio of productive capital to the existing work force. In their analysis of projected changes in India over the period 1956-86, Coale and Hoover calculated that per capita income would rise only 38 percent with high fertility but could rise by 95 percent with significant declines in the level of fertility. This argument was put most force- fully by Enke (1968), who wrote, "The value of a birth prevented in a develop- 216 The Soft Underbelly of Development ing country is one to two times the annual per capita income of that country. . . . One percent of the total development budget spent on reducing births is as effective in raising output per person as the other 99 percent of the budget." These arguments have come under heavy attack. Leibenstein (1969) said that the results depended largely on the assumptions and changed with them. The age distribution disadvantages depend on the ages at which people work, although it is possible to argue that although fertility decline is not economically advantageous in a society where children work from young ages, it would be advisable if the society was succeeding in putting most children in school for substantial periods. Empirical confirmation proved impossible to find. Kuznets (1956) examined eleven now-developed countries over the half-century up to 1914 and could find no correlation between the rates of growth of population and per capita income. He later (1967) failed to show any such relationship after examining the history of either developed or developing countries. The same conclusions were reached by Easterlin (1967), who examined the experience of all developing countries with more than 2 million people in the six years fol- lowing 1957-58, and by Chesnais and Sauvy (1973) in an examination of sixteen countries in Western Europe and seventy-six developing countries during the decade 1960-70. Kuznets (1967, p. 189, n. 15) was not surprised at the absence of a clear demographic-economic relationship because he argued that the per capita gains shown by Coale and Hoover during the first twenty years of fertility decline could be more than offset by a minor change in capital-output ratios. The second version of the United Nations' Determinants and Conse- quences of Population Trends (1973), which, in fact, contained little on conse- quences, admitted the lack of correlation in the historical experience to date but anticipated future correlations. Simon (1981, pp. 81-82) has argued that much of the gain identified by Coale and Hoover was transient, arising from changing age structure as fertility falls, but not existing over a longer period. One possible response might be that many countries would value such a transient period as giving them a chance to get all children into school. The Coale and Hoover arguments were put with renewed force in the World Bank's World Development Report 1984 (World Bank 1984; hereafter WDR), prepared immediately before the 1984 International Population Conference in Mexico City. The arguments placed great emphasis on the problems of school- ing everyone where population growth rates were high, and on widening pro- ductive capital. It concluded (p. 105) that "Population and development are interrelated in many ways, not all of them fully understood.. . . The complexity of the subject makes it tempting to be agnostic about the consequences of rapid population growth. Nevertheless, the evidence discussed above points over- whelmingly to the conclusion that population growth at the rapid rates common in most of the developing world slows development." The WDR (p. 121) did concede that in Kenya (and presumably by implication in the rest of Sub- Saharan Africa) the diverting of the majority of expenditure to lowering child mortality would probably be more cost-effective in reducing fertility than spend- ing the majority of it directly on family planning. Caldwell 217 Many economists found the WDR evidence less than convincing, as reported in a review symposium published in Easterlin, Clark, and Lee (1985). Leibenstein argued, "To the extent that data are reported in the Report, they are almost entirely simulations or projections. That is, these calculations are based on assumptions compounded onto population data rather than on observations of population and economic variables to seek causal connections." Lee criticized the unwillingness to come to terms with the historical data; the dismissal with- out serious discussion of "the possible influence of population size, population growth rate or population density on technological progress . . . although it is the centerpiece of the work of Boserup and Simon," and the leaving of the relation of education to population growth rates and age distributions on a priori arguments rather than on empirical relations. Leibenstein quoted Perlman (1975, p. 256): "If we use antinatalist programs, we do so for reasons other than those simply offered by what we as economists know." At this stage of our thinking on the demographic-economic interactions, the safest conclusion would be to adopt Perlman's. The best arguments for such interventions are likely to prove to be ones of historical analogy and of those family changes that suit economic development. Mortality Change: What Drives It? Life expectancy at birth in Northwest Europe, North America, and Austral- asia rose from around forty years in the early nineteenth century to fifty years in 1900 and is now almost seventy-five years. The swiftest rise occurred between 1880 and 1950, partly propelled in the first quarter of the present century by a decline in infant and child mortality that was at its steepest after the onset of fertility decline and not before. Life expectancy in the developing countries had reached fifty years by the late 1960s, but this was an uneven mixture of a life expectancy of sixty years in Latin America ranging down to forty years in Sub- Saharan Africa. Those levels have now reached around sixty-three years, although Sub-Saharan Africa still averages around fifty years and South Asia little more than fifty-five. The forces driving the mortality decline are once again the subject of debate. McKeown (1967) attempted to discount the progress of scientific medicine before the present century and identified the driving force as improved living conditions. He probably overestimated improvements in nineteenth-century nutrition while understating improvements in water supply and sanitation (Pres- ton and van de Walle 1978), the impact of educational and other social changes, and possibly improvements in the medical system that were not associated with breakthroughs in medical science. Impressive declines in mortality in much of the developing world following World War II revived interest in the causes of such change. Stolnitz (1965) identified the cause as the growth of a global society and the transfer of medical technology after the war. Kuznets (1956), conversely, showed that these trends had begun earlier, in the prewar colonies. The present author (Caldwell 1986) has been associated with some of the 218 The Soft Underbelly of Development revisionist thinking that has been termed health transition. It is clear that in many countries improvements in the health services have been associated with subsequent unusual declines in mortality. Life expectancy increased by twelve years between 1946 and 1953 in Sri Lanka, at seven times the rate of the previous half-century. Similarly, it jumped twelve years in the Indian state of Kerala in 1956-71 and seven years in Costa Rica in 1970-80. The immediate cause of these unusual advances was in each case a health revolution, but, except for the DDT campaign against malaria in Sri Lanka, it did not take the form of the application of new knowledge. In each case, the success was achieved by democratizing the health services by widening access to free or cheap services to the rural and urban poor populations, often with the kindling of a kind of revolutionary fervor. This recipe has not succeeded everywhere. The great suc- cesses were achieved in democratic countries where the electorate had an increasing awareness of its rights; in societies that were fairly egalitarian and well educated; and where women possessed a substantial degree of autonomy. An examination of mortality levels for 1982 in developing countries showed the best predictors of low mortality were the proportions in school-especially the female proportions-a generation earlier, and the extent to which family plan- ning is practiced. The supply of physicians was of lesser importance, although still more important than nutritional levels or income. The educational findings relate of course to the educational levels of parents. The examination of the high levels of correlation between maternal education and child survival has become a major academic industry (Caldwell 1979; Cleland and van Ginneken 1988; Cleland 1990). Caldwell (1986, p. 174) identified eleven countries that in 1982 had mortality rankings well above what would have been predicted by income levels and another eleven with rankings well below. The first group contained countries or states such as China, Kerala, and Sri Lanka, with life expectancies approaching seventy years despite per capita incomes around US$300. The latter were predominantly Middle Eastern and North African countries, many with life expectancies in the fifty- to sixty-year age range and per capita incomes of several thousand dollars. The former had much lower incomes and no more doctors, but the women were not secluded and had higher education levels. Child survival seems to depend to a considerable degree on the mother's self-assurance and the mother's capacity to take action. What is clear is that continuing improvement in health depends on changing the structure of the health system so as to give more universal and equitable access; and on continuing social change in a democratic direction, with more social and gender equality. It might be noted that mortality is much more sensitive to educational differences in the contemporary develop- ing world than in the comparable case at the beginning of the century in the West-probably because education in developing countries imports the social and behavioral adjustments that the West has made to medicine and science (Preston 1985; Ewbank and Preston 1990; Caldwell 1990). Mortality change has been the product of massive social change, economic Caldwell 219 growth, and the development of medical science. Future policies are likely to be most successful if they concentrate on the democratic provision of health ser- vices and the enhancement of the social changes that multiply its effectiveness- female education and female autonomy foremost. What Do Asian Local Studies Show Us about the Onset of Fertility Control? This section focuses on three studies in fertility transition: one in Thailand; one in Sri Lanka; and the other in a small area of rural India. Knodel and colleagues (Knodel, Havanon, and Pramualratana 1984; Knodel, Chamratrithirong, and Debavalya 1987) have reported on Thailand's fertility decline, in which fertility halved in two decades from a total fertility rate (the average number of live births women would have in a lifetime if fertility remained constant at the current level) of 6.3 around 1965 to 3.4 in 1983, and to near replacement levels now. Between 1969 and 1981 contraceptive preva- lence increased from 15 to 60 percent, at first with very little organized assis- tance from government. The researchers, employing a focus group approach as well as surveys, came to a series of conclusions with regard to one of the world's fastest fertility declines. First, the Thais, although not traditionally opposed to large families, had never been strongly pronatalist and had not made high fertility a central aspect of their religion or culture. Partly in consequence of this, fertility deci- sions in recent decades have mainly been a matter for the young couple and have not been regarded as lying within the domain of the larger family. Second, there has been rapid economic and social change. The economic change singled out as most important was not so much the rise in per capita incomes-although this has occurred, moderately at first and more rapidly lately-as the penetration of the market economy and the conversion of most transactions to a cash basis. On the social side, the major aspect of change was in a turning downward of the wealth flow, with an ever-increasing demand for child schooling and child care. Now, almost all Thai children of primary school age are in school, and one- third of those of secondary school age; and infant mortality has dropped since 1965 from almost 100 to less than 40. Third, the research identified a latent and recognized demand for smaller families well before 1965. Fourth, the research identified as an important factor the provision of family planning services, espe- cially by the government. This had the pretransitional effect of crystallizing attitudes toward family size, as well as allowing those attitudes to be translated into smaller families. The researchers found no evidence of pretransitional sub- stantial knowledge of natural methods nor any evidence that those methods could have effected the fertility transition. An interesting comparison is provided by a study of Sri Lanka (Caldwell, Gaminiratne, and others 1987), probably the only Asian country other than Japan to have experienced any significant fertility control before World War II. This control appears to have taken place in the economic depression of the 1930s and is a measure of how market-oriented the economy was by that time, 220 The Soft Underbelly of Development and how hard it was hit by declining world prices for its plantation exports. Moore (1985) has calculated that a majority of rural Sri Lankan households depended on some nonfarming income as early as 1931. Demand for fertility control was encouraged by a range of factors. Sri Lanka's education levels were unusually high, with the 1921 census recording the same proportion of girls in school as did the 1971 Pakistan census. Buddhism, especially as it emerged from the Buddhist Reform Movement of the nineteenth century, gave no special emphasis to fertility-certainly not the emphasis it gave to enlightenment and education. Nevertheless, what fertility control did take place in the 1930s appears not to have been autonomous but to have been affected by the West's family planning debate of the 1930s. It was given direction by the translation into Sinhalese and Tamil of family planning manuals, and, in a population that was one-tenth Catholic, by the teaching of the Catholic Church on rhythm (a method of self-restraint that appealed to Buddhists). When marital fertility decline began again in the 1960s, rhythm and withdrawal predominated as the methods of fertility control, not to be overtaken until the 1980s by the steriliza- tion and contraceptive services provided by the government program. An instructive minor theme in the Sri Lankan experience was a decline in fertility during the 1950s among the poor and poorly educated Indian Tamil tea estate workers (Langford 1982; Caldwell, Gaminiratne, and others 1987). This appears to have been achieved by partial or total sexual abstinence and some abortion and was carried out by a wage-earning proletariat in circumstances in which women's employment was central to the family economy. It also affected their right to retain housing, a matter of great insecurity following Sri Lankan independence, partly because of questions of residence status, and partly because of changing estate ownership and doubt about the future of the estates. The final study is that of nine villages with a total population of around 5,000 in rural Karnataka State, India, studied between 1979 and 1984 (Caldwell and Caldwell 1984; Caldwell, Reddy, and Caldwell 1988). In this area, infant and child mortality had been slowly falling. Differentials by family income were small but by maternal education were large and explained much of the change that had occurred. Part of the explanation for the greater survival of the children of educated mothers was that the latter felt more personal responsibility and took more effective action to prevent sickness and accidents; but even more significant was the better interrelation between mothers and the health system once sickness had occurred. In terms of fertility decline, which had been about 30 percent in the previous twenty years, there was little in the way of preexisting knowledge or practice of fertility control and apparently little in the way of latent demand. The local population had no doubt that the fertility decline that had occurred was almost entirely a product of the activities of the government family planning program, and the researchers largely concurred. The family planning program had its most direct impact during the Emer- gency of June 1975 until March 1977. In this area, which differs little in this Caldwell 221 regard from India as a whole, 86 percent of all couples classified as practicing family planning were sterilized, and sterilization probably explained well over 90 percent of the fertility decline. Although the first sterilization in the area occurred as early as 1962, half of all sterilizations during the next eighteen years were performed in an eighteen-month period during the Emergency, probably accounting by the early 1980s for two-thirds of the extent to which fertility had been depressed below its predecline level. Even during the 1980s sterilization was adopted largely because of the persis- tence and thoroughness of the auxiliary nurse midwives and lady health visitors, supplemented by frequent comments by the government doctor to women with many children, or to those who had recently given birth, that they really should be sterilized for their health's sake. There was no memory, even among the old, of the population ever having been pronatalist in the African sense. That is, most couples wanted at least two sons and a daughter to survive, but their lives were not blighted if they did not reach this target. Few were convinced that lower fertility would bring economic benefits, and, when comparing the ster- ilized with those who had refused, we could find little evidence of improvement except a somewhat greater persistence at school by their children. The benefits may be largely a second-generation gain. In fact, the main reason that some couples definitely decided to limit family size was the problem of keeping chil- dren at school because of the costs associated with schooling, even though there were no fees. Everyone knew that a large family of educated children would bring the greatest long-term benefits, but most found it impossible to keep children at school when they were too close together in age. It should be added that they were rarely closer together than three years because of an average breastfeeding period of two years. So there was not the fear of very close births and the demand for temporary contraceptive methods now found in Bangalore City among young middle-class couples who are greatly reducing the lactation period. In contrast, a pervasive fear of permanent health damage from steriliza- tion weighted the scales against acceptance of family planning. In these rather forbidding circumstances, the decisive force was the female family planning team. The team turned up regularly at every house with three or more children and argued that it was time for a sterilization. The team offered no other method, except to a small elite that knew the doctor socially. Team members concentrated on the young wife, thus breaking with tradition, but increasingly the parents of the young couple-noting the authority of govern- ment, caste, and class-allowed that fertility and family planning decisions were no longer their concern. In the course of their job the family planning workers never wavered in their conviction that sterilization would bring the family eco- nomic benefits and that the operation had no side effects, even though they seemed far less certain of this when talking privately to us. The Hindu population was certainly affected by the fact that limiting family size was the wish of the government and of the local elite that appeared to represent it, and following the newly identified direction toward small families 222 The Soft Underbelly of Development was taken by them to be a mark of virtue. Indeed, this largely self-appointed local establishment, consisting of panchayat (council) members, government functionaries, teachers, bank clerks, Brahmins, Jains, Lingayats, large land- owners, and successful merchants, had been instrumental in ensuring the success of the Emergency family planning program even to the extent of refusing day laborers employment unless they or their wives were sterilized. In this way they both represented the official morality of the time and ensured that the district administration was not penalized by the loss of the 10 percent of its income that would have been incurred had it failed to meet its monthly family planning target. Interestingly, the real opposition was from the Muslim population, whose leaders claimed that Hindus have always worshipped government, whereas they had a Book of God that set out religion and morality and that would not bend to secular power. Certainly, Muslim family planning acceptance was only half that of Hindus and, as is common in India, their fertility was 10 percent higher. Ultimately, the success of the family planning program was due not only to the assistance and persuasiveness of the auxiliary nurse-midwives but to their decisiveness when their clients wavered; they also made the bookings for the monthly sterilization camp and arranged transport to it. It might be noted that vasectomies were common before rural government surgeons were confident of their ability to perform tubectomies, but now a man has a difficult task convinc- ing the system that he, rather than his wife, should be operated on. Global economic and social revolution does encourage fertility decline. How- ever, national family planning programs can accelerate the process by introduc- ing the idea of fertility control, by legitimizing its practice, and by providing the means. Can Intensive Family Planning Areas Reduce Fertility Even Where There Was Little Prior Demand for Restricting Family Size? There is growing experience in translating findings of this kind into compre- hensive and intensive family planning programs in limited areas. The best- known example is probably that in the Matlab Division of Bangladesh, where the International Centre for Diarrhoeal Disease Research has provided 89,000 people with comprehensive family planning services in the treatment area for comparison with 86,000 people in the comparison area, which has only the ordinary government services (Phillips and others 1988). By historical accident, there had been a comprehensive experiment in the same division, in the earlier Contraceptive Distribution Project (CDP) of 1975-76, which assumed latent demand and merely provided contraception in the form of a pill saturation project. Oral contraceptive use, or at least acceptance at the door, rose quickly to 30 percent of target women and then declined, with minimal fertility impact. The subsequent comprehensive program is based on three principles: (1) Female family planning workers are recruited from each village and are selected so that they are literate, young, of low parity, have some Caldwell 223 contraceptive experience, and are from influential families. (2) The family plan- ning workers are allowed a substantial measure of authority and independence so that they really run their own system without feeling like the lower tier in a hierarchical system. (3) There is accountability throughout the system, and checks are made to ensure that each house is regularly visited. It should be noted that this system is similar to those that allowed infant and child mortality to be reduced so dramatically in Sri Lanka after 1945 and in Botswana recently. Since 1977, contraceptive prevalence in the treatment area has risen from 7 to 45 percent compared with a rise in the comparison area from 4 to 16 percent. Even now there is great variation in levels between villages determined largely, it appears, by the degree of commitment of the family planning worker and by the attitudes of the local community leaders (Rahman 1986). It might be noted that such systems depend on a pool of educated girls in each village, preferably familiar with family planning, and this may have been made more possible by the prior existence of the Matlab project. Phillips and others (1988, p. 324) concluded that there was latent demand of a kind, but that "The CDP findings lend support to the view that the provision of contraceptive technology alone will not induce demographic change in tradi- tional societies where demand is fragile. Rather, service demand must be informed by the organizational imperatives of service delivery where social sup- port for contraceptive behavior is weak." They also concluded that the latent demand arose, despite stagnant income growth, from rising economic aspira- tions and an increase in off-farm activities in a situation of growing market penetration. The Matlab experiment was important because it was carried out not only in one of the poorest parts of Asia but among a Muslim population. Even this success may not have been achieved but for the local knowledge that govern- ment and public morality approved. One such intensive area exists in Africa (De Boer and McNeil 1989), among the 300,000 people living in the 1,000 square mile catchment area of Chogoria Hospital, run by the Church of Scotland Mission in Meru District, 100 miles north of Nairobi, Kenya. It is an area where a comprehensive health system, together with other social and emotional support, has long been established. There has been a network of satellite dispensaries through the area since 1970, and by 1985 family planning, together with health services, was being offered daily not only in the hospital but in twenty-seven clinics, and monthly in six mobile clinics. Now, 34 percent of eligible women are using modern contracep- tion, four times the level found in rural Kenya as a whole, and 50 percent say they want no more children. The total fertility rate was measured at 5.2 at a time when it was 8.1 in all of rural Kenya. Bauni (1989) compared Chogoria with two other control areas and concluded that the family planning program had succeeded because the intensive health, contraceptive, and counseling pro- grams of the area had induced broader social change, especially in spousal relationships. Unlike individuals in the control areas, most husbands and wives 224 The Soft Underbelly of Development in Chogoria now share the same beds and can easily and freely discuss family planning practice, and wives also participate in fertility control decisions. Even in the poorest parts of rural Asia and selected parts of Africa, family planning programs will work and will reduce fertility if they are embedded in a comprehensive health program with family-level counseling. To provide a Chogoria-type program across an African nation would probably mean that the health-family planning budget would cost not the 1.5 percent of gross national product (GNP) ($7 per capita) that is typical today, but 3 to 4 percent ($15 to $20 per capita). Even so, the real bottleneck might be human resources rather than money. Summary of the Essentially Asian Experience The lessons from the Asian experience are fairly consistent and agree with much of existing fertility decline theory, which is not as diverse as some of its proponents proclaim. In East and Southeast Asia, there has been enough eco- nomic and related social change to create a latent demand for restricting family size. Nevertheless, there was little prior practice of family planning, and fertility decline would have been slow in the absence of the legitimation and services provided largely by government family planning programs. In mainland South Asia, the economic case for family planning at the family level is weaker, and fertility decline has largely been the product of fairly strong governmental inter- ventionist programs or of programs with an intense and trusted local service component (almost domiciliary care). In both areas pronatalism is weak, and little distinction is made between government-proclaimed morality regarding fertility and morality rising from older religions. Neither of these conditions holds good in the Muslim area, extending from the eastern shore of the Mediterranean to Pakistan, and fertility there generally remains high. Throughout East and Southeast Asia, mortality continues to decline, driven not only by medical technology but also by changes in the way medical services are provided and to a very great extent by social change, particularly female education. An examination of time series on fertility and per capita income for Southeast Asia does little to determine which is the horse and which the cart in develop- ment. In Hong Kong, Indonesia, Malaysia, Philippines, Singapore, Taiwan, and Thailand, growth in per capita income was slower before about 1970 and has successively speeded up since. Fertility probably began to fall in Singapore from the late 1950s, in Hong Kong from the early 1960s, and more generally in the region from the late 1960s. No fertility fall occurred when there had been no prior family planning activities, although not all earlier programs were govern- ment ones. The onset of fertility decline occurred at very different per capita income levels. Perhaps one could draw the following conclusions: (1) In the right circumstances, organized family planning activities led to an earlier onset of fertility transition than would otherwise have occurred. (2) Fertility decline has Caldcvell 225 done no harm to economic growth and may have been one of the preconditions for subsequent rapid per capita increase, possibly largely because of changes in society and the family that owed something to fertility decline. (3) Income growth, fertility decline, and mortality dedine may have been different facets of the same process. (4) When the Philippines experienced economic problems from 1969 onward, fertility decline ceased earlier and resumed later, but both may well have arisen from broader social and political reasons. II. SUB-SAHARAN AFRICA Is the Asian experience, and the theory, which is largely built upon it, applica- ble to Africa? Mortality and Morbidity: Is the Decline Faltering? Sub-Saharan African mortality levels remain, as they have been since adequate statistics first became available, the highest regional levels in the world. A World Bank report (World Bank 1989) summarized the situation as follows: every 100,000 live births result in 10,000 infant deaths, a further 10,000 child deaths before the age of five, and 500 maternal deaths. There have been fears that the position would have been greatly aggravated by slow economic growth during the past decade-even negative growth in per capita income in about one-third of all countries-and severe famines in some areas beginning in the early 1970s. This has become part of popular belief with the latest UNICEF (1989, p. 1) report, The State of the World's Children 1989: "Throughout most of Africa and much of Latin America, average incomes have fallen by 10 to 25 percent in the 1980s.... [A]t least half a million young children have died in the last 12 months as a result of the slowing down or reversal in the developing world." The real situation is probably not as starkly clear as this. These calculations were based on the difference, by the end of the 1980s, between mortality pro- jected according to the mortality trends during the 1970s in ten African and six Latin American countries, and supposed actual mortality. The problem is that the Latin American countries were approaching asymptotic mortality levels, and one might have anticipated some slackening, whereas in Sub-Saharan Africa, the methods of estimation employed, in the total absence of registration data, mean that we can only talk with any confidence about retrospective trends. We have little real knowledge of actual mortality levels at the end of the 1980s, only projected figures. The best attempts at putting these trends together have been made by Hill (1987) and by Hill and Hill (1988); the latter draws on the former, and on another study by the United Nations (1988a) and one by Timaeus (1987) on adult mortality. All make exhaustive use of specialized surveys, and Hill and Hill (1988) and the United Nations (1988a) concentrate on the securer child mortality data. The chief finding of these studies is surprising. Mortality decline has occurred at rather similar rates in nearly all African countries during the 226 The Soft Underbelly of Development whole period from about 1945 into at least the early 1980s. The rate of decline in mortality has been equivalent to a gain in life expectancy of about a third of a year for every elapsed year, lower than the target of half a year that the United Nations said was globally acceptable, but nevertheless by no means discourag- ing. There are insufficient data to draw adequate conclusions with regard to the late 1980s, but by 1990 the expectation of life at birth had probably reached fifty-five years in Southern Africa and fifty years in the rest of the region, with West Africa catching up with East Africa. It is likely that the rate of mortality improvement has been slower in Ethiopia and Mozambique, and in Rwanda in the early 1970s. Evidence suggests that warfare and disorder rather than lack of economic growth block mortality decline. There have also been specific prob- lems in The Gambia and Sierra Leone. This is probably evidence of the problems presented in lowland high-rainfall areas with massive problems of malaria, a thesis that finds support from the failure of Kenya to reduce mortality along the shores of Lake Victoria to anything like the low level found in most parts of the country. A study by Caldwell and Caldwell (1987b) showed that the life expectancy in the richer half of the region's countries was 3.5 years greater than in the poorer half. This difference was surprisingly small, considering that per capita incomes in the latter averaged only 40 percent of those of the former. Furthermore, the mortality differential could probably be explained very largely by the educa- tional differential, itself a product of differences in resources. Nor was the change in mortality between the two groups very different, averaging in terms of annual increase over the previous two decades at 0.31 years in the poorer countries and 0.34 years in the richer ones. A comparison of the experience over the period 1965-85 of the ten countries that had experienced severe drought and famine with the twenty-two countries that had not showed an almost identical increase in life expectancy, even though the countries with famines exhibited only half the per capita income of the ones without famines. The lack of any Malthusian impact from Africa's famines was demonstrated by an average annual rate of population growth in the former group of countries over the whole period of 2.7 percent compared with 2.8 percent in the latter, a tribute to the African experience with famine and the mobility of its people (Caldwell 1975). The real question is why Africa has managed to sustain its mortality decline through such a difficult period over the past twenty-five years, during which per capita income has fallen in half the countries of the region. How has Ghana managed to increase its life expectancy over the last twenty-two years by about 6.5 years at an average annual rate of 0.3 years despite a decline in real per capita income of about 30 percent, a fall in the proportion of children at school, and a decline in the number of doctors per million people (many of those trained went to Nigeria, the United Kingdom, or the Middle East)? Clues may be sought from a comparison with Ghana's neighbor, Cote d'Ivoire, which now has almost twice Ghana's per capita income and more children of primary school age at Caldwell 227 school, but a life expectancy still 1.5 years shorter. A major part of the answer is probably that Ghana in the late 1950s and early 1960s was spending the world's highest proportion of GNP on education, so that by 1965 it had 1.4 times the proportion of girls in primary school and 3.5 times as many in secondary school as Cote d'Ivoire. These are today's mothers. More broadly, the answer for the whole region is the driving force of social change. The proportion of children in school has continued to rise in every country except Guinea, Ghana, and Mali. The towns have grown. The market economy has continued its penetration. This has been helped by some increase in medical services. The ratio of doctors to population has risen except in Ethiopia, Ghana, Guinea, Madagascar, Mozambique, Uganda, and Zaire. In some countries low mortality is within reach. Botswana, helped by dia- monds and a stable economy to raise its per capita income to more than $1,000, now has a life expectancy of more than sixty years and an infant mortality rate of less than 50 per 1,000. Part of the answer is more doctors; their density has quadrupled over the last quarter of a century from a ratio of one doctor to almost 30,000 people to one doctor to less than 7,000. Part of the answer has been the continuing educational revolution taking place within Southern Africa in general, where more girls than boys go to school. Girls' primary school enrollment increased from 71 percent in 1965 to universal education in 1987, and secondary school attendance multiplied eleven times, from 3 to 33 percent. The most potent force in reducing infant and child mortality has probably been the adoption of a system similar to the one that has driven the spectacular Sri Lankan mortality decline. In every community, educated local girls are selected for further extended training in a health training institute as public health workers and are subsequently returned to their communities as public health workers on central government salaries within a supervisory and consul- tative network. They can give health advice and even simple treatments, but their main task is to identify and visit women who are pregnant or who have young children and to advocate healthful antenatal and postnatal behavior, as well as checkups and competent child care. They help to identify sick children and ensure that they are taken to health centers, and they arrange for women to have supervised institutional births. They also advise on family planning. It might be noted that similarly low levels of infant and child mortality are now found in parts of central Kenya where there are high levels of education and health services, and it is probably no accident that these are two of the three countries in mainland, independent Sub-Saharan Africa where fertility may have begun to decline. A similar outcome has been the result of intensive health care in selected villages of The Gambia by the British Medical Council project. The ten World Fertility Surveys in the region revealed, on average, that 10 percent of births resulted in death by one year of age and 20 percent by five years of age. This is the highest child mortality ratio for any world region. A feature of Sub-Saharan African mortality that has long been disturbing is the high child mortality from one to four years of age relative to mortality at other ages. Of the 228 The Soft Underbelly of Development ten World Fertility Surveys (Benin, Cameroon, C6te d'lvoire, Ghana, Kenya, Lesotho, Mauritania, Nigeria, Senegal, and Sudan; see World Fertility Survey 1985, pp. 64-75, and Cleland and Scott 1987, p. 1049), six showed that 45 to 57 percent of deaths occurring before five years of age took place after the child's first birthday. These were largely associated with the highest mortality, but this is not the whole explanation. Even at life expectancies of forty-five years, the wider experience incorporated in the Coale and Demeny (1966) tables showed this proportion ranging between the four families of life tables from 29 to only 43 percent (and, at life expectancies of fifty-five years, from 26 to 40 percent). Unusual levels of malaria may be part of the explanation, although the geo- graphical spread of high toddler mortality in the World Fertility Survey does not fully support this thesis, and the suspicion must remain that a role is often played by poor weaning practices and by inadequate supervision thereafter of the share of food going to very young children. The huge demographic, and indeed developmental, unknown in Africa is the impact of AIDS. As yet, demographers have failed to gauge the situation. For example, we know the human immunodeficiency virus (HIV) levels in East and Middle Africa with approximate accuracy only for Uganda, where admittedly the level is probably above average. In Uganda, about 7 percent of the popula- tion are HIV positive-about 1.2 million people. The periods between infection and the development of AIDS and between that and death appear likely to be shorter than the cohort data reported for developed countries. Without AIDS, Uganda's death rate probably would have been about 13 per 1,000 in the mid-1990s, but it is now likely to be about 19. That means that one-third of all deaths will be due to AIDS, and that probably AIDS will account for 80 percent of deaths in the fifteen- to fifty-five-year age range. Yet neither the World Bank population projections issued in November 1989 (Bulatao and others 1989) nor the most recent United Nations (1988b) projec- tions appear to take the AIDS epidemic into account. In terms of the impact on the World Bank and other agencies, an epidemic of this magnitude will neces- sarily dominate much planning-well beyond the health field and extending to the areas of labor force and economic development, as well as family planning. The epidemic will in some ways be comparable only to the Black Death or to the experience for a few months during 1919 in India of the influenza pandemic. Although the AIDS epidemic may cause comparable deaths and have an immense social impact, it will in one way be very different from the Black Death. The latter occurred in conditions of an almost stationary population, and hence led almost immediately to very significant population decline. A death rate of 25 per 1,000 in Uganda would still, with a birth rate of 50 per 1,000, allow an annual growth rate of 2.5 percent unless the epidemic disorganizes society, marriage, and reproduction to a greater extent than is yet predictable. The major impact of the epidemic is still largely confined to East, Middle, and Southern Africa. Bongaarts (1988) has projected population growth for this region as unlikely to fall below 2 percent per annum. There are now foci of AIDS Caldwell 229 in West Africa, but we do not know as yet whether the spread of large-scale AIDS to that region is only a matter of time or whether there are real barriers to such a spread, as is suggested by recent work on the role of male circumcision (Bon- gaarts and Way 1989). Patterns of sexual networking are not decisively different between West Africa and the AIDs-affected regions (Caldwell and others 1989). In the absence of a major medical breakthrough, one obvious defense against AIDS would be substantial behavioral change, as has occurred among the homo- sexual community in the West. The evidence so far is that such change is occurring in Africa on a very limited scale, partly because of a grassroots refusal to identify the disease, to agree about behavioral risk, or to recognize its incur- ability. Nor is the condom likely to prove of much value outside urban commer- cial sexual relations. There is a need for a great deal of sophisticated social science research, most of it employing methods other than large-scale surveys. We know that untreated venereal disease, especially in rural areas, exists in parts of Africa on an unusually large scale and has had a considerable effect in causing both primary and secondary sterility (Caldwell and Caldwell 1983). There is also ample evidence that venereal and other pelvic infections, when untreated and allowed to develop lacerations or ulcers, greatly increase the rate of trans- mission of AIDS. From a public health stance, then, the most effective way of curbing AIDS iS probably for national and international bodies to cooperate in creating compre- hensive health services, partly oriented toward sexually transmitted diseases and AIDS. This would be enormously expensive and goes against much current eco- nomic ideology, but it probably would have multiple benefits in terms of con- trolling AIDS, reducing other mortality, and controlling fertility. Mortality is probably continuing to fall in Sub-Saharan Africa. It is driven to a marked degree by social change and provides additional evidence of the need for placing female education very high in developmental planning priorities. The Botswanan experience shows that sufficient expenditure will greatly reduce mor- tality and probably lead to the onset of fertility decline. But it should not be overlooked, first, that Botswana has achieved this at the cost of almost 3 percent of its GNP or nearly $30 per capita (which for the average country in the region would represent 10 percent of GNP or half the total government budget); and, second, that this expenditure has been so effective because 9 percent of GNP iS spent on education, the majority on female education (which, again, for the average country, would represent one-third of the GNP or one-and-a-half times the whole government budget). Finally, for much of the region, the expenditures necessary to control the AIDS epidemic are going to be far greater than their budgets can provide, and large-scale foreign aid will be necessary. The Persistence of High Fertility: Is Africa Uniquely Pronatalist? Africa's fertility is high, but not uniquely so, as it is subject to many con- straints. The ten World Fertility Surveys revealed an average total fertility rate of 6.7. This is lower, Cochrane and Farid (1989, pp. 6-7) point out, than the 230 The Soft Underbelly of Development Middle East Surveys (but the latter tended to be in countries of that region with above-average fertility). What is unique about Sub-Saharan Africa is the failure of fertility to decline anywhere. In fact, Cochrane and Farid suggested that fertility has risen in seven of the ten countries, with only one, Ghana, possibly showing a slight decline. Actually, the most recent Demographic and Health Survey (DHS) for Ghana makes it doubtful whether there has been any significant decline in Ghana (Ofosu 1989). However, very recent DHS data do raise the possibility that fertility has begun to decline in Botswana, Zimbabwe, and Kenya. There is as yet little certainty, because the Botswana figures may be an artifact of methods of analysis, and those of Kenya may be a readjustment from an overestimate or an irregular upswing in fertility. If the beginning of fertility decline is real, it is significant that these three countries had to reach levels of life expectancy of about sixty years and infant mortality rates below 70 per 1,000 before it occurred. In fact, most of the Kenyan decline appears to have taken place in parts of Kenya with still lower mortality. Fertility is falling in at least some parts of all other world regions, and the question arises as to whether in some way Africa is different. Caldwell and Caldwell (1985, 1987a, 1990), in a series of works originating in the Changing African Family Project and the Nigerian Family Study, argued that Africa is different. Because religion and family economics mutually reinforced the demand for high fertility, Africa is probably the most pronatalist of the world's regions. Lineage structure, especially in West Africa, is represented in terms of religion by the cult of the ancestors, who intervene in this life and who favor high fertility. The usual return from children that traditional societies offer to parents is reinforced by feelings of guilt, and even fear, if filial duties are not adequately performed. One reflection of pronatalism has been the deep horror of barrenness, which led to the ill-treatment of barren women and to a dread among women of either having no children or of becoming childless through the death of all their children. Our study (Caldwell and Caldwell 1978, 1987c) of all women in Ibadan City in 1973 who were completing their family size and had intentionally and successfully restricted that family to fewer than six children showed that they were subject to enormous pressures from their relatives to prevent them from acting in this way and risking becoming childless through the death of all their children. The result was that only 1.3 percent of women had taken this risk, the majority of whom had broken with their husbands and husbands' families as a result. It is of interest to note that, in fact, they had experienced unusually low mortality among their children, as apparently has been the case in one-child families in China (Caldwell and Srinivasan 1984), doubtless in both cases because of the extra care parents have shown because of their fear of child mortality. This pervasive fear persists in Africa despite the fact that infant mortality has already fallen below 100 per 1,000, the level reached in Northwest Europe when fertility transition had already reduced the total fertility rate to little over half that currently found in Africa, and that of France to one-third of that Caldwell 231 currently found in Africa. The onset of the French fertility decline occurred at an infant mortality rate about 300 per 1,000. The evidence is that despite the levels attained by such socioeconomic indicators as education or income, the mortality threshold for fertility decline in Sub-Saharan Africa will be well below that of other regions. Perhaps infant mortality levels of 50 to 70 per 1,000 will be required, with no more than 10 percent of all births resulting in deaths by five years of age. These are levels that few countries are projected to reach before the year 2000 and that many will reach much later. The fear of barrenness also appears as a fear of wishing for barrenness after any particular age by denying wanting any more children. This explains the uniquely small proportion of women at any parity in recent surveys who said they wanted no more (Caldwell and Caldwell 1987a, p. 413). There are other factors in the persistence of high fertility in the region. Land has traditionally been communal and cultivated with digging sticks and hoes so that, although investment was difficult or impossible in land or farming equip- ment, it was possible in farm labor, usually acquired by marriage or reproduc- tion. The lineage implies shared responsibility for the costs of children. Indeed, massive fostering of children in West Africa means that there is little relationship between reproductive decisions and reproductive economic burdens. The will- ingness to foster in can be taken as evidence that there is no economic loss in having children. This is compounded in West Africa for fathers by the fact that, although men and their families of origin control fertility decisionmaking and certainly the decision to cease childbearing, mothers bear the burden of most day-to-day costs. In a polygynous society there is a certain logic in each woman and her children forming a separate economic unit. Fortes (1978) predicted that the way children or their mothers were treated, or the investment made in them by fathers, would have little influence on the return to fathers from the children; we found this to be so (Caldwell 1982, pp. 65-69). The right of the patrilineage to make fertility decisions is paid for by bride- wealth (a payment at marriage from the bridegroom's family to the bride's family). Thus men can make reproductive decisions with little extra economic burden in raising the children and with a resulting near certainty of support in old age (and earlier). This is a guaranteed recipe for high fertility. It might be expected that the position of women would be very different, but this is not so. Because of the weakness of the spousal economic bond, women become increas- ingly dependent on their children, and few feel safe without a considerable number. Thus, the majority of women, even with very large families, tell survey interviewers that they want more children. A deep fear of terminal barrenness- or at least of expressing a desire for it-means that very few women state that they want no more children. In terms of the fertility transition, this is what makes Africa different. The lineage structure-and its accompanying beliefs in honoring living ancestors with economic support and dead ones with continuing lineage births-orients the society toward persistent high fertility. This might be undermined if, in West 232 The Soft Underbelly of Development Africa, there were a move toward common spousal budgets and, everywhere, if more expenditure were confined to the nuclear family, if more of it were spec- ified for children, and if polygyny declined. It would be further undermined if there were a related move toward common spousal decisionmaking about fertil- ity control. Family attitudinal change may be promoted by family planning programs and, indeed, by some fertility decline. However, many Africans argue that the defense of African culture and the African family is more important than fertility decline. Caldwell and Caldwell (1985, 1988a) did not claim that fertility decline would not take place in Africa but only that Africa was sufficiently different from Asia for fertility decline there to require different thresholds-lower levels of mortality and higher levels of education and possibly income. Caldwell and Caldwell argued that twenty years ago a range of African countries scored no lower than India, Indonesia, or Thailand in a series of socioeconomic indexes, but that whereas family planning programs succeeded in these Asian countries, they did not in Ghana or Kenya. They believed that only a few African countries would exhibit fertility declines before the year 2000 and that even subsequent fertility decline would be slower than the current projections. These views have been largely supported by Lesthaeghe (1989), Page (1988), and Frank and McNicoll (1987). Cleland and Wilson (1987, p. 26) concluded, after their historical, global survey of fertility transition, that "the indisputable strength of pronatalist sentiments in Africa poses a major theoretical challenge. It cannot be explained simply by low levels of development, educational attain- ment, or life expectancy." Recent World Bank reports have been prominent in their support of a one- world tradition, endorsed by Boserup (1985, p. 383), who maintained that the persistence of high fertility was merely the outcome of lower levels of develop- ment. Projections issued by other international agencies show that they, too, make this assumption. The World Bank (1986, p. 12) concluded that Africa was "probably not" different in terms of the acceptability of family planning and that the failure of fertility decline to begin was explained by generally lower levels of income, education, health, and urbanization. This stance was sustained in a later World Bank report (1989, p. 71), in which the authors cited supposed fertility declines in Botswana and Zimbabwe and family planning success in Chogoria, concluding that "there is increasing evidence that, where family plan- ning services are available, contraceptive use is high." However, Cochrane and Farid (1989), after exhaustively examining World Fertility Survey data, con- cluded that Sub-Saharan Africa was different and that at every economic level the population wanted more children than elsewhere. African social and economic structures are less conducive to providing mate- rial returns to individuals or families for fertility declines than are those of Asia. Fertility decline is likely to begin in only a few countries during the present decade, and for the region as a whole it is likely to be slower than has been the experience in most of Asia or Latin America. Caldwell 233 The Mechanics of African Fertility: How Is Fertility Determined? We need to examine at greater length just how African fertility levels are determined. In West Africa, the husband is usually now responsible for over- head costs, such as housing, which was traditionally provided by the lineage, whereas there is a considerable range in the variable cost from the wife meeting nearly all expenses to the husband-especially among the middle class-making a considerable contribution (see, on Lagos, Fapohunda and Todaro 1988). In East and Southern Africa, where women do not market on the same scale and hence are less likely to earn substantial separate incomes, there is usually a unified family budget, especially in the more patriarchal of the patrilineal soci- eties. This should facilitate fertility decline in that there is a closer relationship between reproductive decisionmaking and subsequent economic burdens, although even here parents can make substantial claims on other relatives to help meet education costs. Similarly, child fosterage appears to occur on a larger scale in West Africa. Employing Sierra Leone census data, Bledsoe and Isiugo-Abanihe (1989) found that in some chiefdoms, more than 50 percent of children under age two were away from their mothers. When Bledsoe and Isiugo-Abanihe investigated the ramifications of family economics in the society, they concluded that who meets the cost of African children is a very complex matter that bears little relationship to reproduction or to reproductive decisions. Page (1989) employed WFS data to calculate fosterage as measured by the residence of children away from their mothers. She found it highest on the West African coast; some 35 percent of children under fifteen were away in the southwest C6te d'Ivoire, and 40 percent of ten-to-fourteen-year-olds were away in parts of Cote d'Ivoire and Ghana (the World Fertility Survey did not include Sierra Leone). Page found little difference by the sex of children, and that migrant parents in urban areas were likely to send children under five back to the rural areas, whereas older children were more likely to be fostered to the towns for education and employment. Across Africa, she found the highest correlation with the level of marital instability, which can be translated as a correlation with a weak husband-wife dyad reflect- ing a stronger lineage organization-above all a characteristic of West Africa. Most Africans believe that marriages are more stable if the children present are only those of the new couple and fostered-in children, whereas each spouse's children by previous unions are fostered to their own relatives. This institution, too, should mean longer persistence of high fertility in West Africa. Traditionally, African birth spacing, largely aimed at improving the survival chances of both children and their mothers, was achieved both by prolonged breastfeeding and by postpartum female sexual abstinence, which could last as long as three years and which was still more than two years even in the city of Ibadan in 1973 (Caldwell and Caldwell 1977). The duration of postpartum abstinence is now falling and has become particularly short in this century in parts of East Africa, according to Lesthaeghe and Eelens (1989, p. 90) because 234 The Soft Underbelly of Development of early and thorough Christianization, the lack of gerontocratic rulers, and the encapsulation of women in their husbands' patrilinies. One might be somewhat doubtful about the exact role of the last two conditions for West Africa and might add instead the higher level of polygyny there and hence the greater number of men with access to more than one wife. There is an associated more general recourse to marital sexual abstinence, and many women expect little sex after their mid-thirties, which probably explains why Cochrane and Farid (1989, p. 5) found abnormally high fertility under thirty-five years (resulting from earlier female marriages than in other world regions), and lower levels among older women. The important points, in terms of fertility transition, about the reduction of both sexual abstinence and breastfeeding, are, first, that the process might nullify the impact of rising contraceptive practice on control- ling fertility for a sustained period; and second, that this nullification process is likely to end suddenly with the onset of fertility decline, once further contraction of breastfeeding and abstinence can have no more impact as their duration is no greater than that of postpartum amenorrhea. Polygyny exists in Sub-Saharan Africa on a scale unknown elsewhere. Typ- ically, more than 40 percent of married women are in polygynous unions in West Africa; between 20 and 30 percent in East Africa, and probably fewer than 20 percent in Lesotho and Botswana (Lesthaeghe, Kaufmann, and Meekers 1989, pp. 276-77). Timaeus and Graham (1989, p. 373) claim that in the nineteenth century polygyny was as prevalent in Botswana as in West Africa but has declined because of Christianization, as well as because of changes in agri- culture that made women's work less valuable. Polygyny probably does little to reduce the fertility of individual women and something to maximize societal fertility by keeping all women in marriages (Pebley and Mbugua 1989). Its main impact on fertility is that it helps to sustain such institutions as child fosterage and separate spousal budgets, which delay the onset of fertility transition. At the level of current polygyny found in West Africa, the institution can be regarded as universal, in that a higher proportion of women than those currently in poly- gynous marriages will find themselves so situated in the course of a lifetime, and nearly every woman must prepare herself emotionally and economically for this eventuality. There is little sign of a decline in polygyny in West Africa. Lower levels of polygyny with increased female education might suggest such a change, but Gaisie (1969) has shown that as educational levels have risen in Ghana, both polygyny and its differentials by education have remained stable through its incidence increasing at every educational level. Goldman and Pebley (1989) have shown that the stability of the system depends largely on the spousal age gap and widow remarriage but that it is likely to prove remarkably resistant to falls in fertility. It has been argued that the fundamental force sustaining polygyny, and indeed high fertility as well, is the value of family labor in agriculture. In Botswana, not only has polygyny steeply declined but marriage itself is disintegrating, with half of all adult women never having married. Timaeus and Graham (1989) and Caldwell 235 Lesthaeghe, Kaufmann, and Meekers (1989, p. 243) claim that this is a direct product of the declining importance of hand-tilled agriculture with the introduc- tion of the plow from South Africa in the nineteenth century and the displace- ment of agriculture by commercial stock raising more recently. Although there are currently no great differentials in fertility by marital status, these changes may assist fertility decline. Contraceptive use in Sub-Saharan Africa is both at a lower level than in any other world region and less likely to be related to the control of family size. World Fertility Survey data showed that efficient methods had been used by 6 percent of women in Sub-Saharan Africa, compared with 32 percent in Asia, 37 percent in North Africa, and 50 percent in Latin America. The region's level of current use ranged from 6 percent in Ghana to 5 percent in Kenya; 3 percent in Lesotho and Sudan; 1 percent in Benin, Cameroon, Nigeria, and Senegal; and 0 percent in Cote d'Ivoire and Mauritania (Cochrane and Farid 1989, pp. 26, 90). Supplementing these data with others from the Contraceptive Prevalence Sur- veys and the Demographic and Health Surveys, Page (1988, p. 33) has calcu- lated the current use of modern contraception for various dates, most in the 1980s, as 28 percent in Zimbabwe, 19 percent in Botswana, 10 percent in Kenya, 6 percent in Ghana and Liberia, 3 percent in Lesotho, and 1 to 2 percent in most other countries. Caldwell (1975) had assessed the use of modern con- traception in Middle and West Africa in 1970 at about 125,000 women or 0.5 percent and had projected 1980 levels as 1.6 percent with no further develop- ment of family planning programs, 3.1 percent with considerable development, and 4.7 percent with programs of the Indonesian type. The actual level reached in 1980 appears to fall between the first and second projections. Many contra- ceptive methods are not as yet particularly acceptable in Africa. Mauldin and Segal (1988, p. 341) show that even of Africa's limited use of contraception, only 10 percent is sterilization (nearly entirely female, because of prohibitive cultural pressures against vasectomy), compared with 45 percent in all develop- ing countries, whereas another 10 percent is the intrauterine device (IUD), com- pared with 24 percent in all developing countries. Indeed, only the birth-control pill has been found to be particularly acceptable, at a level of 40 percent in contrast to 12 percent in the developing world as a whole. The very low level of acceptability of condoms bodes ill for their use against AIDS. Success in African family planning programs probably depends on much higher levels of use of IUDS and tubectomies. The relationship between the use of family planning and the desire to control ultimate family size, assumed by family planning programs in Asia, does not hold in Africa. The family planning services may meet a social need without doing very much to limit family size. In southwest Nigeria, the Changing Afri- can Family Project (Caldwell and Caldwell 1976) found that 40 percent of first contraceptive use was to substitute for postpartum sexual abstinence, with the obvious possibility of raising fertility, while other major uses were to prevent conception during premarital or extramarital sexual relations or after terminal 236 The Soft Underbelly of Development abstinence was supposed to have begun. Only 11 percent of contraceptors in Ibadan City had begun to use contraception with any thought of controlling family size, and, at the time of the survey, only 17 percent were employing it for this purpose. As only one-sixth of Ibadan women had ever practiced fertility control, this was 3 percent of all women. The fairly high nonmarital use of contraception arises from the fact that premarital and extramarital sexual rela- tions are not proscribed in the region to the same extent as in Asia (Caldwell, Caldwell, and Quiggin 1989). Nichols and others (1986) showed in Ibadan in 1982 that among secondary school students over fourteen years of age, 60 percent of boys and 38 percent of girls had been sexually active. This means that African family planning programs have to be much more oriented toward the young than in Asia, although such orientation meets with considerable opposition. Bongaarts (1987, pp. 134-35) reported that African countries did not follow the relationship between contraceptive prevalence and fertility found in the rest of the world. That formula suggested a total fertility rate for Kenya in 1977 of 6.4, whereas 8.1 was actually recorded, and for Zimbabwe in 1984 of 4.5, although 6.5 was found. Kenya, in fact, had experienced rising fertility over twenty years as contraceptive prevalence increased. Part of the explanation is uses of contraception for other purposes than controlling family size, and part is the fact that fertility is not as restricted to marriage as in most of the developing world. Nevertheless, Cochrane and Farid (1989) concluded that if all informa- tion about marriage, abstinence, breastfeeding, sexual activity, and contracep- tive use were put together, then the region's fertility should be higher than recorded, and they suspected the cause to be abortion, sterility, subfecundity, and spousal separation. In fact, all of these factors play a role. Abortion occurs on a considerable scale, especially to permit schoolgirls to continue their education and to allow wives to hide extramarital conceptions (Caldwell and Caldwell 1978). Primary sterility is a major problem in Middle Africa, where fertility may rise as health services improve; and secondary sterility after ten or fifteen years of marriage is probably widespread (Caldwell and Caldwell 1988b). Many wives have little or no sex after their mid-thirties even if spousal separation has not taken place (Caldwell, Caldwell, and Quiggin 1989). In many African countries the value of family labor in agriculture still keeps levels of both polygyny and fertility high. This is reinforced by mechanisms that weaken the relationship between fertility and the economic burden arising from it-for example, weaker emotional and economic ties between spouses than between each and their lineages, separate budgets for spouses, and massive child fosterage. All are related to polygyny, and, like polygyny, are most prominent in West Africa, thus helping to explain why West African fertility decline may be slow. African levels of contraceptive use are low, and most use is not aimed at controlling family size. The use of IUDS and tubectomies is particularly low, a fact that militates against early fertility transition, as these are the major means Caldwell 237 by which contemporary developing world fertility transitions have been achieved. The Success of Population Policies: Why Is Africa Differentfrom Asia? National family planning programs began in Kenya in 1968 and in Ghana in 1970, dates comparable with those of Southeast Asian programs, although the success in controlling fertility has been much less in the African countries. This lack of success has been described as arising from a lack of efficiency in the African programs, although Caldwell and Caldwell (1988a) have argued that it is demand that drives such efficiency. They have identified lack of demand to control family size and an unwillingness of administrators and politicians to be identified with failure and with promoting programs reportedly at odds with the African way of life. Caldwell and Caldwell also argued that African govern- ments would never be able to impose Chinese-type family planning programs or even ones more like those in India or Indonesia, because of the newness of African national states; a related lack of Confucianist or Brahmanical traditions of state or elite leadership, especially in areas of personal and moral sensitivity; and a strong belief that individuals and communities know more about the morality of fertility than do governments, because fertility is the central concern of traditional religion. At the governmental level there has been a major shift toward approval of fertility control, as is shown by comparing policy statements at the 1974 Bucharest and the 1984 Mexico City World Population Conferences. The Eco- nomic Commission for Africa (1989) reports that between 1977 and 1987 the proportion of African governments claiming that their population growth rate was too high climbed from 34 to 59 percent and that those stating that they had policies to curb excessive population growth rose from 23 to 46 percent. Les- thaeghe (1989, pp. 475-76) ascribes these changes to increasing concern over food crises and lack of economic growth and to the hope that successful popula- tion policies will help to overcome these crises, together with the realization that traditional African birth spacing through postpartum sexual abstinence is being eroded. Because one-fourth of the population of Sub-Saharan Africa lives in Nigeria, a good deal hinges on just how that government implements its new population policies and on just how successful the implementation proves to be. Caldwell (1982) has suggested that fertility decline depends to a considerable extent on the increasing emotional and economic nucleation of conjugal rela- tions, and this interpretation is supported for Kenya by Frank and McNicoll (1987, p. 231ff.). Lesthaeghe (1989) is skeptical about whether the required Westernization of the family is taking place, but his view of the future is not very different (p. 36): "One can expect a close relationship to emerge between the regional pattern of fertility transition and the geographic spread of Christianity, with education acting as a major intermediate variable." The position may be more complex than this in that coastal West Africa has been no less enthusiastic about the Christian message than East or Southern Africa, but-either because 238 The Soft Underbelly of Development there were fewer European settlers or because West African cultural institutions were more resistant-that message has not been so closely related to family change. Lesthaeghe (1989, p. 149) also reports that his quantitative analysis has provided some consolation for economic determinists by showing that economic development has a strong effect on contraceptive use. Sindiga (1985, p. 175) believes that family planning programs in Africa will be more successful if they admit the dominant role of men in Africa and design their interventions accord- ingly. The World Bank (1989, p. 72) says that ultimately it will be a question of money, and that programs to be successful will need to spend 0.6 to 0.8 percent of GNP or perhaps 5 percent of the national budget, an amount similar to that usually allocated for health. African governments are changing toward antinatalist policies faster than the societies they represent. Indeed, it is the fear that their constituents will not follow their lead that gives rise to weak and often confused programs. It may well ultimately be a question of money, but it is very doubtful whether the critical expenditure is 0.6 to 0.8 percent of GNP. The fact that African fertility decline is so dependent on massive child mortality decline, and the resource outlays needed in Botswana and Chogoria in Kenya, suggest that the combined health and family planning figure could be for many African countries 3 to 4 percent of GNP or typically two to three times the size of current health budgets. Population Projections: Are We Now Guided by Realistic Estimates? The World Bank released its latest population projections in November 1989 (Bulatao and others 1989), showing Sub-Saharan African population growing from 444 million in 1985 (9 percent of the world's population) to 1,907 million in 2050 (19 percent); 2,535 million in 2100 (22 percent); and soon after, in conditions of stable global population, to 2,614 million (23 percent). The mortality projections are, in one sense, not surprising, because they assume the continuation of the trends of the last four decades, with annual gains in life expectancy of about one-third of a year per elapsed year from a level of about fifty-two years now to sixty-five years by 2030. What is surprising is the inability still to face the AIDS crisis. For example, Uganda's 1995 crude death rate is projected to be 13 per 1,000, with consequent natural increase of 3.1 percent per annum, whereas in reality the death rate will probably be at least 19 per 1,000 with natural increase about 2.5 percent. The fertility projections are much more radical and assume quite sudden and dramatic change, with net reproduction rates in Sub-Saharan Africa reaching unity-a level attained in the West temporarily during the Depression of the 1930s and probably more permanently in the 1970s, after a century of decline; by 2010 in Botswana; 2015 in Zimbabwe; 2020 in South Africa; 2030 in Lesotho and Namibia; 2035 in Benin, Ghana, Liberia, Madagascar, Sudan, Swaziland, and Togo; and, last, by 2060 in Niger. This implies fertility declines between 1985-90 and 1990-95 in twenty-eight (almost 60 percent) of the coun- tries covered; no change in fourteen (almost 30 percent); and increases, as Caldwell 239 sterility problems are overcome, in six (more than 10 percent). By 2000 to 2005, a little more than a decade from now, falls of 24 percent or more are projected for Benin, Botswana, Ghana, Liberia, Madagascar, Namibia, South Africa, Swaziland, Togo, and Zimbabwe. Indeed, the two latter, with declines of 42 and 43 percent, are shown with levels about 25 per 1,000 and rates of decline equal to those reached by Thailand between 1965 and 1985 (Knodel and others 1987, p. 56). Various points are noteworthy. The projections place great emphasis on the first reports of change in Botswana and Zimbabwe, although there has been little in the way of confirmatory analysis or follow-up surveys. They also back hunches about fast declines in a range of West African coastal countries much changed by foreign contact and in the highly Christianized Southern African countries, which are, in addition, affected by the South African economic sys- tem. They may well underestimate the strength of institutional resistances in West Africa. They do not agree with Lesthaeghe and colleagues' thesis (1985) that Kenya-with its low mortality, socioeconomic change, Christianization, and early family planning program-will be a front-runner. I share the skepti- cism of Page (1988, pp. 31-32) about whether change can be as rapid as this, but I note that the United Nations (1988b) shares the World Bank's optimism. The United Nations' total projected population for the region is, in fact, 1 percent below the Bank's projection in the year 2000, although 4 percent higher by 2025. The United Nations achieves this effect less by identifying front- runners than by postulating general fertility decline, so that by the year 2000 the United Nations gives Botswana, Ghana, Togo, and Zimbabwe all birth rates around 44 (births per annum per 1,000 population), whereas the World Bank suggests 25, 28, 38, and 39, respectively. However, the United Nations shows fertility falling everywhere, whereas the World Bank envisages delays in many countries and postulates initial rises in others. The other difference is that for the period 1985-90, the United Nations begins with a regional life expectancy of fifty years in contrast to the Bank's estimate of fifty-two years. The U.N. projections may possibly be right for the whole region, but they cannot possibly be right for individual countries, for the rather mechanical projections show, with few exceptions, declines in national total fertility rates of A approximately 5 percent in the first ten years after 1985-89 and 50 percent or a little more in the first thirty-five years. Thus Botswana, Ghana, and Zimbabwe move at a slightly slower pace than Chad, Niger, or Sierra Leone. The United Nations does select two front-runners with falls of 60-62 percent in thirty-five years: Kenya and, rather curiously, Burundi. It anticipates a rise in fertility in only one country, Gabon, and it offsets the general fertility decline by a rather steeper mortality decline than does the World Bank. Thus, life expectancy henceforth is projected to rise by 0.4 years per elapsed year (about 25 percent faster than in the last four decades), with some slackening after sixty-five years is attained. In contrast to the picture presented by the latest World Bank and U.N. popu- Table 1. Population Growth and Economic and Social Indexes, Independent Sub-Saharan African Countries with mid-1987 Populations Exceeding 1 Million Average Percentage annual of central per govern- capita D f I Number ment Otber characteristics Average annual income Percentagesobrelevant9ag of per- expendi- Total population growth, growth groupatschool, 1986 sons per ture spent fertility Life Per capita 1965-80 1965-87 Primary Secondary doctor, on health, rate, expectancy, income US$, (percent) (percent) school school 1984 1987 1987 1987 (years) 1987 Less than 2.5a 0.7(1) 58(3) 17(3) 25,786 1.4(8) 6.0 50 412(1) 2.5-3.Ob 0.1(2) 63(4) 15(4) 22,438(6) 1.0(9) 6.6 50 357 More than 3.0c 1.1 97(5) 21 10,780(7) 2.0(10) 6.6 54 688 Note: Unweighted averages. Data not available, as numbered above: (1) Guinea and Mali. (2) Mozambique. (3) Sierra Leone. (4) Congo, Nigeria, Uganda, and Zaire. (5) Liberia. (6) Cameroon, and Zaire. (7) Cote d'lvoire, Tanzania. (8) Burundi, Central African Republic, Chad, Guinea, Mali, Mauritania, and Sierra Leone. (9) Benin, Congo, Ethiopia, Madagascar, Mozambique, Niger, Senegal, Somalia, Uganda, and Zaire. (10) C6te d'lvoire, Gabon, and Rwanda. (11) Guinea. a. Burkina Faso, Burundi, Central African Republic, Chad, Ghana, Guinea, Lesotho, Mali, Mauritania, Mauritius, and Sierra Leone. b. Benin, Cameroon, Congo, Ethiopia, Madagascar, Malawi, Mozambique, Niger, Nigeria, Senegal, Somalia, Sudan, Uganda, and Zaire. c. Botswana, Cote d'lvoire, Gabon, Kenya, Liberia, Rwanda, Tanzania, Togo, Zambia, and Zimbabwe. Source: World Bank (1989). Caldwell 241 lation projections, there is little evidence that we are about to witness a wide- spread onset of fertility decline in Sub-Saharan Africa. Most research comes to the opposite conclusion, partly because of a lack of economic returns to individ- uals from restricting family size, and partly because African governments are likely to prove unwilling or unable to implement Asian-type family planning programs. On the basis of past experience, one might forecast declines begin- ning somewhat later and perhaps of no more than half the steepness of the projections. Certainly they will begin in some countries well before others, as the World Bank suggests. West Africa is likely to offer greater cultural resistance to fertility decline than either group of projections suggest. However, the AIDS epidemic may mean that the past is no guide to the future, for it may have an impact on marriage, and through it on reproduction, and it may bring into existence health services that reduce child mortality more rapidly and so acceler- ate fertility decline. Do Sub-Saharan African Countries Benefit from Slower Population Growth? A Crude Analysis Table 1 examines the relationship between the rates of population growth during 1965-80 and per capita income growth and various other measures during the late 1980s that demographic-economic theory suggests should be affected. The anticipated linear deterioration in other measures as population growth increases does not appear. Instead there is a U-shaped curve in income growth, health expenditure, and secondary schooling, and a linear increase in primary schooling and the density of doctors. Clearly, the social measures are most affected by preexisting income levels, whereas little can be said about the determinants of income growth. There are major problems in carrying out this kind of analysis in the region. Because the region is pretransitional, there is little difference in the level of fertility between the growth-rate categories. Instead, rates of population growth are largely determined by mortality and migration levels. Thus, fewer doctors and lower levels of primary schooling might be expected to yield higher mortal- ity and hence lower population growth rates. In addition, lower rates of popula- tion growth can be the product of emigration arising from a faltering economy, whereas high rates of population growth may reflect the attraction to immi- grants of a rapidly growing economy. Furthermore, population growth arising from the arrival of energetic, and sometimes skilled, adults may assist rather than retard economic growth. It is noteworthy that three of the major emigrant countries of the 1965-80 period are found among the low-growth-rate group, Burkina Faso, Ghana, and Mali, whereas two of the chief immigrant countries, C6te d'Ivoire and Gabon, are in the high-growth-rate group. In fact, the average rate of natural increase of the first three was higher during this period than that of the last two. Table 2 makes a second attempt to examine the situation. This time the focus is on the rate of natural increase halfway through the period, for it is this rate 242 The Soft Underbelly of Development Table 2. Level of Natural Increase and Economic and Social Indexes, Independent Sub-Saharan African Countries (except Mauritius) with mid-1987 Populations Exceeding 1 Million Average annual per capita Percentage of Average annual income Percentages of relevant age centralgovern- natural increase growth, group at school, 1986 Number of ment expendi- 1970-75 1965-87 Primary Secondary persons per ture spent on (percent) (percent) school school doctor, 1984 health, 1987 2.5 or lessa 1.4(1) 69(2) 15(2) 25,589(4) 1.7(6) More than 2.Sb -0.4 73(3) 18(3) 15,555(s) 1.4(7) Note: Unweighted averages: The rate of natural increase is the difference between the birth and death rate. Data not available, as numbered above: (1) Congo. (2) Congo. (3) Liberia, Nigeria, Uganda, and Zaire. (4) Cameroon. (5) C6te d'lvoire, Tanzania, and Zaire. (6) Burundi, Central African Republic, Chad, Congo, Ethiopia, Gabon, Guinea. Madagascar, Mali, Mauritania, Mozambique, Senegal, and Sierra Leone. (7) Benin, C6te d'lvoire, Niger, Rwanda, Somalia, Sudan, and Zaire. a. Botswana, Burkina Faso, Burundi, Cameroon, Central African Republic, Congo, Chad, Ethiopia, Gabon, Guinea, Lesotho, Madagascar, Malawi, Mali, Mauritania, Mozambique, Senegal, and Sierra Leone. b. Benin, Cote d'lvoire, Ghana, Kenya, Liberia, Niger, Nigeria, Rwanda, Somalia, Sudan, Tanzania, Togo, Uganda, Zaire, Zambia, and Zimbabwe. Sources: United Nations (1976), World Bank (1989). that helps determine the level of child dependency, which, if high, is held, in the dominant economic demographic theory, to be disadvantageous. This time, Mauritius, with its island location and completed demographic transition, is omitted as of little relevance to the rest of the region. The level of natural increase is merely dichotomized because it does not exhibit the extremes that population growth does. Once again, the countries that provided doctors achieved lower mortality and higher rates of natural increase. Even that position might change because by 1987 the countries that earlier had lower levels of natural increase were spend- ing a higher proportion of GNP on health and had marginally more children in primary school. However, the significant difference in per capita income growth is noteworthy. It cannot be reproduced in table 1 by dichotomizing that table. There remains the possibility, of course, that, with only thirty-four countries, the result is merely an artifact of coincidentally different histories, economic policies, and natural resources. Nevertheless, it seems likely that lower levels of natural increase do favor growth in per capita income even if they are achieved not by the control of fertility but by its depression by unusually high levels of sterility (as in the cases of Cameroon, Central African Republic, Chad, and Gabon, although the first and last almost certainly have high per capita income growth rates because of oil and other minerals) or by its offsetting by high mortality levels. Continued high mortality is not the route, however, to better living standards, because in Sub-Saharan Africa, unlike in earlier Europe, even moderately high mortality reinforces the persistence of high fertility, whereas mortality itself nevertheless gradually falls. If there is an inverse relationship Caldwell 243 between population growth and economic growth, it would merely delay the economic crisis. Wbat Does the Future Hold, and Which Developmental Policies Should Be Adopted? Demographic-economic theory does not provide a particularly good guide for development policy. On the whole, its findings have not been sustained by historical experience. A surer guide might be provided by two commonsensical observations. The first is that in the course of development to a degree of affluence, the history of countries now developed shows both that per capita income rose and that family size declined. It seems likely that each change was helped by the other in subtle interactions that are not captured by existing theory. It is reassuring that modern Western history shows very substantial declines in infant and child mortality immediately subsequent to the first fertility decline. The second consideration is that much of Sub-Saharan Africa suffers from poor soils and has little in the way of large alluvial valleys suited to irrigation. Hence, it seems unlikely that the attainment of adequate diets and an agricultural surplus for export would be assisted by a multiplication of popula- tion of more than the sixfold increase suggested by current population projec- tions, themselves characterized by a surprising belief in the immediacy of fertility decline. It would be unwise to assume that economic growth in the region will be much better in the immediate future than it has been in the immediate past. That means that improvement in per capita income will need all the marginal support that social and demographic change can provide. It also means that continued mortality decline, which surely must remain a central developmental objective, will need to be sustained by social change and by a more democratic and effi- cient distribution of the health services that can be afforded. The form of social change that yields the greatest results in mortality decline and that may be the most certain road to fertility decline is education-especially female education- and development seems to demand major inputs into this area both for direct benefits and for indirect ones via demographic change. In terms of the power of forces tending to sustain high fertility, Sub-Saharan Africa is probably unique. One powerful support for high fertility is the nature of family economics, which, especially markedly in West Africa, tends to diffuse the cost of children and even to place it predominantly on those who do not make the fertility decisions. The route to low fertility lies either through a greater conjugality in the economic burden, and a greater concentration of economic gains and losses within the nuclear family, or perhaps in a movement toward women taking responsibility for fertility control decisions. In either case, this means the kind of social change that can only be promoted by an intensive grassroots family planning program. It may be that the success of such programs will have a direct economic impact, in that family emotional and economic nuclearization, stronger conjugal emotional and economic links, and a down- 244 The Soft Underbelly of Development ward turning of the wealth flow toward children are the preconditions for sustained economic development. Such changes will happen more easily in East and Southern Africa, for the West African family system is still structured to provide few immediate returns for fertility decline. However, it should be noted that Mason and Taj (1987), after reviewing the reported global evidence, con- cluded that there was little to show that women were likely to desire fertility control earlier than men, a conclusion that may hold for Africa as well, because the economic independence of African women means that in old age they are very dependent on their children. It is also clear that Africa's cultural context and family economic systems make it likely that the child mortality threshold will be at lower levels of mortal- ity for the onset of fertility decline than has been the case elsewhere. If a great deal of investment in fertility control is not to be wasted, then there would appear to be a case in Africa for combined intensive maternal and child health and family planning programs. There is already evidence that such programs can work, but they require not only a sufficient density of local health services in rural areas and among the urban poor but also the employment of trained health and family planning workers to visit households regularly. Such programs are easier in the areas where malaria is less intense, and the latter may once again require world attention. All planned efforts in Africa are rendered more complex and more urgent by the AIDS epidemic, which is already massively affecting areas such as southwest Uganda and threatening to have a major impact on health and distort the societies and economies of much of East and Southern Africa within the next decade. This is a development challenge of unprecedented magnitude, and his- tory will not easily forgive us if we are not prepared to spend massively in ameliorating the effects of the disaster. In the absence of a decisive medical breakthrough, the most efficient way of spending such money would seem to be on universal and comprehensive health services that could clear up pelvic infec- tions early and reduce the rate of transmission of pelvic disease. The upgraded health system will also be needed to detect HIV infection and to assist AIDS victims. There is no real conflict in using this intensive system also to reduce child mortality and to provide family planning. Indeed, the kind of social changes that may accompany marital fertility control may also be the necessary changes to reduce the extent of sexual networking. One reason for a need to combine family planning with programs to reduce infant and child mortality, and the reason for employing, as the major instru- ment for both, intensive local services, is the fact that African governments will not be able to create the kind of family planning programs that have worked in China or even in India and Indonesia. In the words of Frank Notestein, such programs would be more likely to result in the fall of governments than of the birth rate. Caldwell 245 Although it is likely that there is no fundamental conflict between battling the AIDS epidemic and continuing and intensifying fertility control efforts, it is curious that so little has appeared in the literature to show that much thought is being given to the subject. What, then, might be the immediate targets in Sub-Saharan Africa for pro- gram inputs in the population field? 1. We probably have to conclude that the way to a modern economy, and to bringing the type of society compatible with it into existence, lies through fertil- ity decline. Given the weakness of the health infrastructure, this means a need for government programs that can certainly learn from the Asian experience. Government involvement is also needed to legitimize the control of family size, and especially to legitimize the use of IUDS and female sterilization. The sooner every country of significant size has its own Chogoria to prove that fertility can be reduced, and to provide a model for expansion, the better. Although government programs are absolutely necessary in the region, the weakness of governmental service networks means that full use should also be made of the private sector in the form of doctors, other health personnel, or commerce. The most significant service function of the private sector would most likely be to distribute oral contraceptives and to further the concept of small families. 2. All research has shown that both fertility and child mortality decline are strongly accelerated by female education. Africa could do better in the area of girls' education, and indeed in all education. An unweighted average of all countries in the region shows in 1986 for primary school 91 percent of boys attending compared with only 63 percent of girls (comparable Asian figures are, for India nearly all boys and 76 percent of girls, and for Indonesia universal attendance by both sexes). In Africa 27 percent of boys and 13 percent of girls are in secondary schools (in India, the respective figures are 45 and 24 percent, and in Indonesia 45 and 34 percent). These figures are not all dictated by economic levels. Among the poorer half of Sub-Saharan African countries (as measured by per capita incomes), Madagascar, Mozambique, Tanzania, Togo, and Zambia have more than 70 percent of girls in primary school, whereas Madagascar has 30 percent in secondary school. For a range of demographic and other development reasons, furthering education-especially that of girls- should be given high priority. 3. In much of rural Africa health services are still rudimentary. In any circum- stances, there would be a strong case for concentrating collaborative aid in this area. The case is strengthened by the extent to which any fertility decline is dependent on lower child mortality, and it is rendered overwhelming by the emergence of the AIDS epidemic. The latter will almost certainly move to center stage during the coming decade in terms of African developmental and aid policies. The world finds it easy to be alarmed by this unprecedented crisis but surprisingly difficult to plan for it. 246 The Soft Underbelly of Development REFERENCES Allan, William. 1965. The African Husbandman. London: Oliver and Boyd. Basu, Alaka Malwade. 1984. "Ignorance of Family Planning Methods in India: An Important Constraint on Use." Studies in Family Planning 15, no. 3: 136-42. Bauni, Evasius K. 1989. "A Quantitative Study of Fertility Behaviour in Chogoria, Maua and Kyeni of Eastern Kenya, 1989." Centre for Population Studies, London School of Hygiene and Tropical Medicine, London. Processed. Becker, Gary S. 1960. "An Economic Analysis of Fertility." In National Bureau of Economic Research, Demographic and Economic Change in Developed Countries: A Conference of the Universities-National Bureau Committee for Economic Research. Princeton, N.J.: Princeton University Press. Ben-Porath, Yoram. 1980. "The F-Connection: Families, Friends and Firms and the Organization of Exchange." Population and Development Review 6, no. 1: 1-30. Blacker, C. P. 1947. "Stages in Population Growth." Eugenics Review 39, no. 3: 88- 102. Bledsoe, Caroline, and Uche Isiugo-Abanihe. 1989. "Strategies of Child Fosterage among Mende Grannies in Sierra Leone." In Ron J. Lesthaeghe, ed., Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Bongaarts, John. 1987. "The Proximate Determinants of Exceptionally High Fertility." Population and Development Review 13, no. 1: 133-9. . 1988. Modeling the Spread of HIV and the Demographic Impact of AIDS in Africa. Center for Policy Studies Working Paper 140. New York: Population Council. Bongaarts, John, and P. Way. 1989. Geographic Variation in the HEv Epidemic and the Mortality Impact of AIDS in Africa. Population Council Research Division Working Paper 1. New York: Population Council. Boserup, Ester. 1985. "Economic and Demographic Interrelationships in Sub-Saharan Africa." Population and Development Review 11, no. 3: 383-97. Bulatao, Rodolfo A., Eduard Bos, Patience W. Stephens, and My T. Vu. 1989. Africa Region Population Projections, 1989-90 Edition. Policy, Research, and External Affairs Working Paper 330. Population and Human Resources Department, World Bank. Washington, D.C. Cain, Mead. 1981. "Risk and Fertility in India and Bangladesh." Population and Devel- opment Review 7, no. 3: 435-74. Caldwell, John C. 1975. "Fertility Control." In John C. Caldwell, ed., Population Growth and Socioeconomic Change in West Africa. New York: Columbia University Press. 1976. "Toward a Restatement of Demographic Transition Theory." Population and Development Review 2, no. 3-4: 321-66. .1978. "A Theory of Fertility: From High Plateau to Destabilization." Population and Development Review 4, no. 4: 553-77. -.1979. "Education as a Factor in Mortality Decline: An Examination of Nigerian Data." Population Studies 33, no. 3: 395-413. 1982. Theory of Fertility Decline. London and New York: Academic Press. 1986. "Routes to Low Mortality in Poor Countries." Population and Develop- ment Review 12, no. 2: 171-220. Caldwell 247 1990. "Cultural and Social Factors Influencing Mortality Levels in Developing Countries." Annals of the Academy of Political and Social Science, 1990. Philadelphia. Caldwell, John C., and Pat Caldwell. 1976. "Demographic and Contraceptive Innova- tors: A Study of Transitional African Society." Journal of Biosocial Science 8, no. 4: 347-65. .1977. "The Role of Marital Sexual Abstinence in Determining Fertility: A Study of the Yoruba in Nigeria." Population Studies 31, no. 2: 193-217. -. 1978. "The Achieved Small Family: Early Fertility Transition in an African City." Studies in Family Planning 9, no. 1: 2-18. -. 1983. "The Demographic Evidence for the Incidence and Cause of Abnormally Low Fertility in Tropical Africa." World Health Quarterly 36, no. 1: 2-34. . 1984. "The Family Planning Program at the Local Level: A Study of a Village Area in South India." In Gavin W. Jones, ed., Demographic Transition in Asia. Singapore: Maruzen Asia. -. 1985. "Cultural Forces Tending to Sustain High Fertility in Tropical Africa." PHN Technical Note 85-16. Population, Health and Nutrition Department, World Bank. Washington, D.C. Processed. -. 1987a. "The Cultural Context of High Fertility in Sub-Saharan Africa." Popula- tion and Development Review 13, no. 3: 409-37. . 1987b. "Famine in Africa." Paper presented at International Union for the Scientific Study of Population seminar on Mortality and Society in Sub-Saharan Africa. Yaound6, 19-23 October. .1987c. "The Limitation of Family Size in Ibadan City, Nigeria: An Explanation of its Comparative Rarity Derived from In-Depth Interviews." In Etienne van de Walle and J. A. Ebigbola, eds., The Cultural Roots of African Fertility Regimes. Ile-Ife: Obafemi Awolowo University; Philadelphia: University of Pennsylvania. . 1988a. "Is the Asian Family Planning Program Model Suited to Africa?" Studies in Family Planning 19, no. 1: 19-28. .1988b. "Marital Status and Abortion in Sub-Saharan Africa." Paper presented at International Union for the Scientific Study of Population seminar on Nuptiality in Sub-Saharan Africa, Paris. . 1990. "Cultural Forces Tending to Sustain High Fertility." In George T. Acsadi and Gwendolyn J. Acsadi, eds., Population Growth and Reproduction in Sub-Saharan Africa: Technical Analyses of Fertility and Its Consequences. Washington, D.C.: World Bank. Caldwell, John C., Pat Caldwell, and Bruce Caldwell. 1987. "Anthropology and Demography: The Mutual Reinforcement of Speculation and Research." Current Anthropology 28, no. 1: 25-43. Caldwell, John C., Pat Caldwell, and Pat Quiggin. 1989. "The Social Context of AIDS in Sub-Saharan Africa." Population and Development Review 15, no. 2: 185-234. Caldwell, John C., K. H. W. Gaminiratne, Pat Caldwell, Soma de Silva, Bruce Caldwell, Nanda Weeraratne, and Padmini Silva. 1987. "The Role of Traditional Fertility Regu- lation in Sri Lanka." Studies in Family Planning 18, no. 1: 1-21. Caldwell, John C., P. H. Reddy, and Pat Caldwell. 1985. "Educational Transition in Rural South India." Population and Development Review 11, no. 1: 29-51. . 1988. The Causes of Demographic Change: Experimental Research in South India. Madison: University of Wisconsin Press. 248 The Soft Underbelly of Development Caldwell, John C., and Lado T. Ruzicka. 1978. "The Australian Fertility Transition: An Analysis." Population and DevelopmentReview 4, no. 1: 81-103. Caldwell, John C., and K. Srinivasan. 1984. "New Data on Nuptuality and Fertility in China." Population and Development Review 10, no. 1: 71-79. Chesnais, J. C., and Alfred Sauvy. 1973. "Progres Economique et Accroissement de la Population: Une Experience Commentee." Population 28, no. 2: 843-57. Cleland, John. 1990. "Maternal Education and Child Survival: Further Evidence and Explanations." In John C. Caldwell and others, eds., What We Know about Health Transition: The Cultural, Social, and Behavioral Determinants of Health. The Pro- ceedings of an International Workshop, Canberra, May 1989. Health Transition Series 2. Canberra: Australian National University. Cleland, John, and Jerome K. van Ginneken. 1988. "Maternal Education and Child Survival in Developing Countries: The Search for Pathways of Influence." Social Sci- ence and Medicine 27, no. 12: 1357-68. Cleland, John, and Chris Scott, with the collaboration of David Whitelegge, eds. 1987. The World Fertility Survey: An Assessment. Oxford, England: International Statistical Institute and Oxford University Press. Cleland, John, and Christopher Wilson. 1987. "Demand Theories of the Fertility Transi- tion: An Iconoclastic View." Population Studies 41, no. 1: 5-30. Coale, Ansley J. 1967. "Factors Associated with the Development of Low Fertility: An Historic Summary." In Proceedings of the World Population Conference, Belgrade 1965. New York: United Nations. . 1973. "The Demographic Transition." In International Union for the Scientific Study of Population, International Population Conference, Liege, 1973, vol. 1. Liege, Belgium. Coale, Ansley J., and Paul Demeny. 1966. Regional Model Life Tables and Stable Populations. Princeton, N.J.: Princeton University Press. Coale, Ansley J., and Edgar M. Hoover. 1958. Population Growth and Economic Development in Low-Income Countries: A Case Study of India's Prospects. Princeton, N.J.: Princeton University Press. Coale, Ansley J., and Susan Cotts Watkins, eds. 1986. The Decline of Fertility in Europe. Princeton, N.J.: Princeton University Press. Cochrane, Susan H., and Samir M. Farid. 1989. Fertility in Sub-Saharan Africa: Analy- sis and Explanation. World Bank Discussion Paper 43. Washington, D.C. Davis, Kingsley. 1951. The Population of India and Pakistan. Princeton, N.J.: Princeton University Press. De Boer, Cornelis N., and Malcolm McNeil. 1989. "Hospital Outreach and Community-Based Health Care: The Case of Chogoria, Kenya." Social Science & Medicine 28, no. 10: 1007-17. Demeny, Paul. 1988. "Demography and the Limits to Growth." Population and Devel- opment Review. Supplement to vol. 14, Population and Resources in Western Intellec- tual Traditions. Washington, D.C.: American Association for the Advancement of Science Committee on Population, Resources, and the Environment. Easterlin, Richard A. 1967. "Effects of Population Growth on the Economic Develop- ment of Developing Countries." In John D. Durand, ed., World Population. Special Issue of Annals of the American Academy of Political and Social Sciences 369: 98- 108. Caldwell 249 1975. "An Economic Framework for Fertility Analysis." Studies in Family Plan- ning 6: 54-63. Easterlin, Richard A., Colin Clark, and Ronald D. Lee. 1985. "Review Symposium on World Development Report 1984." Population and Development Review 11 (March): 113-38. Easterlin, Richard A., Robert A. Pollak, and M. L. Wachter. 1980. "Towards a More General Model of Fertility Determination: Endogenous Preferences and Natural Fertil- ity." In Richard A. Easterlin, ed., Population and Economic Change in Developing Countries. Chicago: University of Chicago Press. Economic Commission for Africa. 1984. "Food Production and Population Growth in Africa." In Department of International Economic and Social Affairs, United Nations, Population, Resources, Environment and Development: Proceedings of an Expert Group on Population, Resources, Environment and Development, Geneva, 28-29 April 1983. New York. .1989. Status and Prospects of Population Policies in ECA Member States. African Population Studies Series 10. Addis Ababa, Ethiopia. Enke, Stephen. 1960. "The Economics of Government Payments to Limit Population." Economic Development and Cultural Change 8, no. 4, part 1: 337-48. . 1968. "Raising Per Capita Income through Fewer Births." 68 TMP-9 Tempo. Santa Barbara, Calif.: General Electric Company. Processed. Ewbank, Douglas, and Samuel H. Preston. 1990. "'Personal Health Behaviour and the Decline in Infant and Child Mortality: The United States, 1900-1930." In John C. Caldwell and others, eds., What We Know about Health Transition: The Cultural, Social, and Behavioral Determinants of Health. The Proceedings of an International Workshop, Canberra, May 1989. Health Transition Series 2. Canberra: Australian National University. Fapohunda, Eleanor R., and Michael P. Todaro. 1988. "Family Structure, Implicit Contracts, and the Demand for Children in Southern Nigeria." Population and Devel- opment Review 14, no. 4: 571-94. Fortes, Meyer. 1978. "Family, Marriage, and Fertility in West Africa." In Christine Oppong and others, eds., Marriage, Fertility and Parenthood in West Africa: Papers from the 15th Seminar of the International Sociological Association Committee on Family Research, Lome, Togo, January 1976. Changing African Family Monographs 4. Canberra: Department of Demography, Australian National University. Frank, Odile, and Geoffrey McNicoll. 1987. "Fertility and Population Policy in Kenya." Population and Development Review 13, no. 2: 209-43. Gaisie, Samuel K. 1969. Dynamics of Population Growth in Ghana. Ghana Population Studies 1. Accra: Demographic Unit, University of Ghana. Goldman, Noreen, and Anne Pebley. 1989. "The Demography of Polygyny in Sub- Saharan Africa." In Ron. J. Lesthaeghe, ed., Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Higgins, G. M., and others. 1982. Potential Population Capacities of Land in the Devel- oping World: Technical Report of Project INT/75/P13, "Land Resources for Popula- tions of the Future." Rome: FAO, in collaboration with International Institute for Applied Systems Analysis and U.N. Fund for Population Activities. Hill, Althea. 1987. "Trends in Childhood Mortality in Sub-Saharan Mainland Africa." Paper presented at International Union for the Scientific Study of Population Seminar on Mortality and Society in Sub-Saharan Africa, Yaoun&, Cameroon. 250 The Soft Underbelly of Development Hill, Althea, and Kenneth Hill. 1988. "Mortality in Africa: Levels, Trends, Differentials and Prospects." In Etienne van de Walle, Patrick Ohadike, and Mpembele D. Sala- Diakanda, eds., The State of African Demography. Liege: International Union for the Scientific Study of Population. Jain, Anrudh K. 1989. "Fertility Reduction and the Quality of Family Planning Ser- vices." Studies in Family Planning 20, no. 1: 1-16. Kirk, Dudley. 1944. "Population Changes and the Postwar World." Sociological Review 9, no. 1: 28-35. . 1946. Europe's Population in the Interwar Years. Geneva, Switzerland: League of Nations. Knibbs, G. H. 1917. The Mathematical Theory of Population, of Its Character and Fluctuations, and of Factors Which Influence Them. Appendix A of the 1911 Census of Australia, 1. Melbourne: Australian Government. Knodel, John, Aphichat Chamratrithirong, and Nibhon Debavalya. 1987. Thailand's Reproductive Revolution: Rapid Fertility Decline in a Third World Setting. Madison: University of Wisconsin Press. Knodel, John, Napaporn Havanon, and Anthony Pramualratana. 1984. "Fertility Tran- sition in Thailand: A Qualitative Analysis." Population and Development Review 10, no. 2: 297-328. Knodel, John, and Etienne van de Walle. 1979. "Lessons from the Past: Policy Implica- tions of Historical Fertility Studies." Population and Development Review 5, no. 2: 217-45. Kuznets, Simon. 1956. "Quantitative Aspects of the Economic Growth of Nations, Part 1: Levels and Variability of Rates of Growth." Economic Development and Cultural Change 5, no. 1: 5-94. -. 1967. "Population and Economic Growth." Proceedings of the American Philo- sophical Society 11: 170-93. Langford, Christopher M. 1982. "The Fertility of Tamil Estate Workers in Sri Lanka." World Fertility Survey Scientific Reports 31. Voorburg, Netherlands: International Statistical Institute. Leibenstein, Harvey. 1957. Economic Backwardness and Economic Growth: Studies in the Theory of Economic Growth. New York: Wiley. -. 1969. "Pitfalls in Benefit-Cost Analysis of Birth Prevention." Population Studies 23, no. 2:161-70. . 1975. "The Economic Theory of Fertility Decline." Quarterly Journal of Eco- nomics 89, no. 1: 1-31. Lesthaeghe, Ron J., ed. 1989. Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Lesthaeghe, Ron J., and Frank Eelens. 1989. "The Components of Sub-Saharan Repro- ductive Regimes and Their Social and Cultural Determinants: Empirical Evidence." In Ron J. Lesthaeghe, ed., Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Lesthaeghe, Ron J., Georgia Kaufmann, and Dominique Meekers. 1989. "The Nup- tiality Regimes in Sub-Saharan Africa." In Ron J. Lesthaeghe, ed., Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Lesthaeghe, Ron J., and Johan Surkyn. 1988. "Cultural Dynamics and Economic Theo- ries of Fertility Change." Population and Development Review 14, no. 1: 1-45. Caldwell 251 Lesthaeghe, Ron J., C. Vanderhoeft, S. Becker, and M. Kibet. 1985. "Individual and Contextual Effects of Education on Proximate Fertility Determinants and on Lifetime Fertility in Kenya." In John B. Casterline, ed., The Collection and Analysis of Com- munity Data. Voorburg, Netherlands: World Fertility Survey, International Statistical Institute. Lorimer, Frank. 1946. The Population of the Soviet Union: History and Prospects. Geneva, Switzerland: League of Nations. Mason, Karen O., and Anju M. Taj. 1987. "Differences between Women's and Men's Reproductive Goals in Developing Countries." Population and Development Review 13, no. 4: 611-38. McKeown, Thomas. 1967. The Modern Rise of Population. London: Edward Arnold. Mauldin, W. Parker, and Bernard Berelson. 1978. "Conditions of Fertility Decline in Developing Countries 1965-1975." Studies in Family Planning 9, no. 5: 89-147. Mauldin, W. Parker, and Sheldon J. Segal. 1988. "Prevalence of Contraceptive Use: Trends and Issues." Studies in Family Planning 19, no. 6 (part 1): 335-53. Moore, Mick. 1985. The State of Peasant Politics in Sri Lanka. Cambridge, England: Cambridge University Press. Mortimore, M. J. 1968. "Population Distribution, Settlement, and Soils in Kano Pro- vince, Northern Nigeria, 1931-62." In John C. Caldwell and Chukuka Okonjo, eds., The Population of Tropical Africa. New York: Longman. Nelson, Richard R. 1956. "A Theory of the Low-level Equilibrium Trap in Under- developed Economies." American Economic Review 46, no. 5: 894-908. Nichols, Douglas, 0. A. Ladipo, John Paxman, and E. 0. Otolorin. 1986. "Sexual Behavior, Contraceptive Practice, and Reproductive Health among Nigerian Adoles- cents." Studies in Family Planning 17, no. 2: 100-6. Notestein, Frank W. 1943. "Some Applications of Population Change for Post-War Europe." Proceedings of the American Philosophical Society 87, no. 2: 165-74. Notestein, Frank W., and others. 1944. The Future Population of Europe and the Soviet Union. Geneva, Switzerland: League of Nations. Nugent, Jeffrey B. 1985. "The Old Age Security Motive for Having Children." Popula- tion and Development Review 11, no. 1: 75-97. Ofosu, D. Y. 1989. "Hazard Models Analysis of Birth Intervals: A Study Based on West African Data." Ph.D. thesis, Department of Demography, Australian National Uni- versity, Canberra. Page, Hilary. 1988. "Fertility and Family Planning in Africa." In Etienne van de Walle, Patrick 0. Ohadike, and Mpembele D. Sala-Diakanda, eds., The State of African Demography. Liege, Belgium: International Union for the Scientific Study of Population. . 1989. "Childrearing versus Childbearing: Coresidence of Mother and Child in Sub-Saharan Africa." In Ron J. Lesthaeghe, ed., Reproduction and Social Organiza- tion in Sub-Saharan Africa. Berkeley: University of California Press. Pearl, Raymond. 1939. The Natural History of Population. London: Oxford University Press. Pebley, Anne, and Wariara Mbugua. 1989. "Polygyny and Fertility in Sub-Saharan Africa." In Ron J. Lesthaeghe, ed., Reproduction and Social Organization in Sub- Saharan Africa. Berkeley: University of Califomia Press. 252 The Soft Underbelly of Development Perlman, Mark. 1975. "Some Economic Growth Problems and the Part Population Policy Plays." Quarterly Review of Economics 89, no. 2: 247-56. Phillips, James F., Ruth Simmons, Michael A. Koenig, and J. Chakraborty. 1988. "Determinants of Reproductive Change in a Traditional Society: Evidence from Matlab, Bangladesh." Studies in Family Planning 19, no. 6 (part 1): 313-34. Preston, Samuel H. 1985. "Resources, Knowledge and Child Mortality: A Comparison of the United States in the Late Nineteenth Century and Developing Countries Today." In International Union for the Scientific Study of Population, International Population Conference, Florence, 1985, vol. 4. Liege, Belgium: International Union for the Scientific Study of Population. Preston, Samuel H., and Etienne van de Walle. 1978. "Urban French Mortality in the Nineteenth Century." Population Studies 32, no. 2: 275-97. Rahman, Makhlisur. 1986. Tradition, Development and the Individual: A Study of Conflicts and Supports to Family Planning in Rural Bangladesh. Asian Population Change Series 1. Canberra: Department of Demography, Australian National University. Satia, J. K., and Rushikesh M. Maru. 1986. "Incentives and Disincentives in the Indian Family Welfare Program." Studies in Family Planning 17, no. 3: 136-45. Schultz, T. Paul. 1969. "An Economic Model of Family Planning and Fertility." Journal of Political Economy 77: 153-80. .1981. Economics of Population. Reading, Mass.: Addison-Wesley. Schultz, Theodore W. 1945. Food for the World. Chicago: University of Chicago Press. Simon, Julian L. 1981. The Ultimate Resource. Oxford, England: Martin Robertson. Sindiga, Isaac. 1985. "The Persistence of High Fertility in Kenya." Social Science and Medicine 20, no. 1: 71-84. Stix, Regine, and Frank Notestein. 1934. "Effectiveness of Birth Control." Milbank Memorial Fund Quarterly 12, no. 1: 57-68. . 1940. Controlled Fertility: An Evaluation of Clinic Service. Baltimore, Md.: Williams and Wilkins. Stolnitz, George. 1965. "Recent Mortality Trends in Latin America, Asia, and Africa." Population Studies 19, no. 2: 117-38. Taeuber, Irene B. 1958. The Population of Japan. Princeton, N.J.: Princeton University Press. Thompson, Warren. 1929. "Population." American Journal of Sociology 34: 959-75. Timaeus, Ian. 1987. "Adult Mortality in Sub-Saharan Africa." Paper presented at World Bank Seminar on Disease and Mortality in Sub-Saharan Africa, Tonbridge Wells. Timaeus, Ian, and Wendy Graham. 1989. "Labor Circulation, Marriage and Fertility in Southern Africa." In Ron J. Lesthaeghe, ed., Reproduction and Social Organization in Sub-Saharan Africa. Berkeley: University of California Press. Tuladhar, Jayanti M. 1987. "Effect of Family Planning Availability and Accessibility on Contraceptive Use in Nepal." Studies in Family Planning 18, no. 1: 49-53. UNICEF (U.N. Children's Fund). 1989. The State of the World's Children 1989. New York: Oxford University Press. United Nations, Department of Economic and Social Affairs. 1965. Population Bulletin of the United Nations, no. 7, 1963, with Special Reference to Conditions and Trends of Fertility in the World. New York. Caldwell 253 . 1973. The Determinants and Consequences of Population Trends: New Sum- mary of Findings on Interaction of Demographic, Economic, and Social Factors. Vol. 1. New York. - 1976. Demographic Yearbook 1975. New York. - 1988a. Mortality of Children under Age 5: World Estimates and Projections, 1950-2025. New York. . 1988b. World Demographic Estimates and Projections, 1950-2025. New York. van de Walle, Francine. 1986. "Infant Mortality and the European Demographic Transi- tion." In Ansley J. Coale and Susan Cotts Watkins, eds., The Decline of Fertility in Europe. Princeton, N.J.: Princeton University Press. Westoff, Charles F., Lorenzo Moreno, and Noreen Goldman. 1989. "The Demographic Impact of Changes in Contraceptive Practices in Third World Populations." Popula- tion and Development Review 15, no. 1: 91-106. Willcox, Walter F. 1916. "The Nature and Significance of the Changes in the Birth and Death Rates in Recent Years." American Statistical Association, New Series, 113. Wilson, Christopher. 1984. "Natural Fertility in Pre-Industrial England, 1600-1799." Population Studies 38, no. 2: 225-40. World Bank. 1984. World Development Report 1984. New York: Oxford University Press. . 1986. Population Growth and Policies in Sub-Saharan Africa. Washington, D.C. * -. 1989. Sub-Saharan Africa: From Crisis to Sustainable Growth. Washington, D.C. World Fertility Survey. 1985. Final Report. Voorburg, Netherlands: International Studies Institute. PRO CE E D INGS OF THE WO RLD BANK ANN UAL CO NFE RE NCE ON D E V E LOPM E NT E C ON OM ICS 1 9 9 0 COMMENT ON "THE SOFT UNDERBELLY OF DEVELOPMENT," BY CALDWELL Ron Lesthaeghe My first reaction to Caldwell's paper was actually to its title. Why should demography or demographic transition be the "soft underbelly" of develop- ment? Although I am by no means an expert on development economics in the strict sense, I imagine that any historian reviewing economic prescriptions for development since World War II would come up with major contradictions in that field as well. It is quite some distance from early theories stressing industrial development through large-scale public works or investment in heavy industries to more recent concentrations on basic needs and now on structural adjustment. It seems therefore that the creature of "social and economic development" has more than one soft underbelly and that demographic transition may not be among the "softest." This by no means implies that contemporary demographic policies are able to draw on a neat and tidy body of theory that is comfortably supported by empirical evidence. In fact, the less than perfectly clear historical record of the currently industrialized nations on the link between demographic and economic change gives ample warning of the difficulties to come when analysts shift to development on other continents. Caldwell's lucid paper, laid out in no fewer than twelve questions and accom- panying suggested answers, testifies once again to the complexities of the link- ages between demographic and economic change. I had little difficulty in sub- scribing to most of Caldwell's premises, but this -should come as no surprise, because author and discussant both adhere to an intellectual tradition of social demography that pays ample attention to the historical contexts of patterns of social organization and culture and that does not divorce policy proposals and evaluations from these contexts. Consequently, our tradition has a definite his- torical, anthropological, and sociological flavoring, and it makes ample use of cross-cultural comparative analysis. To be more precise, the tradition of social demography is intrinsically inter- ested in the effects of changes in the socioeconomic structure-such as the growth of the wage sector, the economic roles of women, urbanization, alter- ations in land tenure and agriculture, the restructuring of social stratification Ron Lesthaeghe is dean of the Faculty of Economic, Social, and Political Sciences at the Free University of Brussels. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 255 256 Comment systems, and education-on demographic factors that condition population growth and distribution. These effects operate through a variety of channels that are often connected to changes in the structure of community and of kinship or household organization. Because this gives rise to a plethora of possible con- texts, I completely fail to see why a single and simple "universalistic" or global theory of demographic transition should hold. Hence, both of us harbor a deep skepticism toward universalistic policy prescriptions worked out on the basis of any currently fashionable economic doctrine. The basic criticism of our relativistic line of thought is that it sacrifices eco- nomic efficiency for the sake of sociological understanding. This allegation is unjustified. As Caldwell's paper shows, cross-cultural comparisons have yielded variables that have an impressive record in producing demographic change. The first of these is undoubtedly education, and female education in particular, because it cuts deeply into the fabric of lineage, community, and household organization and fosters individual autonomy in reordering priorities and in making choices. Education and female schooling have and remain therefore closely associated with declining infant and childhood mortality, later ages at marriage, and a faster fertility transition. But again, one should not look for convenient thresholds, because other contextual variables are bound to alter the size of the effects and can even temporarily change their sign as well. Schooling can serve the cause of individual autonomy, but it can also serve traditionalism, so the politics concerning the control of educational institutions enter into the picture as well. A second crucial variable, equally singled out by Caldwell, is the structural transformation of the household economy. This variable responds not only to broader structural transformation (generally moving away from the household as a unit of production) but also to social patterning of gender relations. Although William Goode (1963) holds that household systems show a global tendency of converging to the nuclear family pattern, with more egalitarian gender relations, I feel that there is still ample room for new patterns to crystal- lize that definitely do not resemble the Western pattern. Africa again is where Goode's evolutionary thesis is least appropriate, as Caldwell also makes clear. Hence, a careful monitoring of how family and household systems evolve in different contexts and of their effects on demographic variables will not yield a simple picture. So far I have sketched the points in Caldwell's summary of the state of the art with which I am in agreement. I now turn to elements to which I think he has not given due attention. Caldwell's world seems to be split into two-societies with nonfamilistic reli- gions and patterns of gender relations that easily accommodate pressures toward reducing the demand for children, and societies with strongly pronatalist cul- tures and patterns of social organization. The Islamic world and Sub-Saharan Africa squarely fall in the latter category. So far, I have no basic problem. The current demographic record speaks for itself. Lesthaeghe 257 But now I would like to introduce some caveats. First, standard demographic projections assume an uninterrupted decline in fertility to replacement level. But there is evidence that societies with more openness to fertility control are at risk of going through a "halfway transition," in which the demand for children definitely declines but in which many individuals want and need fertility levels substantially above replacement. Several cases are on record in which an initial fertility decline stalled between total fertility rates of 3 and 4.5 children and contraceptive prevalence between 45 and 65 percent (see Bongaarts 1986, p. 21). These cases include countries with large populations, such as Brazil, India, Malaysia, and the Philippines, or countries with relatively early declines, such as Costa Rica and Sri Lanka. Historically, also, France went through a marked stepwise decline with an intermediate plateau between 1835 and 1880. In my opinion, the populations of Sub-Saharan Africa-if they do experience an initial fertility decline during the 1990s-are also likely candidates for a halfway tran- sition, because the utility of children is likely to remain well above that wit- nessed in Western countries or in the Far East, and because more coercive forms of family planning stand no chance of success. My second reservation about Caldwell's paper is that it does not envisage an incipient fertility transition in response to economic hardship or crisis. This possibility has been resuscitated by Boserup (1985) specifically for use in the Sub-Saharan context. The thesis goes against the grain of classic demographic transition theory but not against that of empirical evidence. The first third of the French fertility decline occurred in tandem with a substantial fall in household income. The rapid rise in contraceptive use in Brazil's northeast is associated with emigration and economic crises. And the beginning of a fertility decline in central Kenya (and outside the laboratory setting of the comprehensive health system serving the 300,000 people living in the catchment area of Chogoria Hospital) came when migration opportunities dwindled, returns from invest- ment in education sagged, and the economic growth rate of the nation slack- ened. At present, I would not discount the "crisis-led transition," because rising relative child costs and frustrations with respect to future opportunities may equally operate in the direction of a partial fertility decline, especially where the transition started with an initial fertility bulge (associated with declining breast- feeding, postpartum abstinence, and overall child-spacing)-that is, especially in Sub-Saharan Africa. A third point I missed in Caldwell's paper is that it lacks a section dealing with the other pronatalist world: the Islamic countries. Although these nations exhibit a great deal of heterogeneity with respect to policies and economic setting, it is still obvious from a demographer's point of view that they have been lagging behind in the fertility decline. It is also not hard to argue that this is associated with the strength of traditional gender relations and low levels of female education relative to that of males. Yet there may be more movement in Islamic settings than is perceived from the outside. The debate on individual autonomy versus religious strictness is fully emerging in an important number of 258 Comment Arab countries and, in combination with an economic crisis in a number of them, is apparently producing a fertility decline as well. For example, Fargues (1989) shows that birth rates have been steadily falling in Tunisia since 1965 and in Algeria since 1975. In Tunisia, the total fertility rate of 6.9 children in 1966 had fallen to 4.9 children in 1981. In Algeria, the total fertility rate was 8.4 children in 1970 and 5.4 children in 1986. A number of classic relations hold. Both countries have considerably reduced female illiteracy; age differences between the spouses at marriage have declined; and female ages at marriage have gone up significantly. Nonfarm and nondomestic female employment, however, has evolved much less. It also seems that the Tunisian fertility decline lost momentum in the 1980s, whereas the Algerian one has been quite spectacu- lar from the early 1980s onward. I have introduced the notions of halfway transition and crisis-led transition to show that outcomes can be produced that again defeat standard textbook notions of demographic transition. They actually add grist to Caldwell's mill in contending that policy-relevant wisdom is not readily transportable from one setting to another, but that investment in female education is still the best bet in most cultural contexts and economic situations, provided that female education fosters individual autonomy and remains outside the control of traditionalistic doctrines. A word should also be said about Caldwell's view concerning the primary role attributed to what he calls the latent or underlying demand for contraception and family planning services. At several places in his paper, Caldwell stresses that family planning programs are only successful if they correspond to an emerging need. Yet he also singles out family planning experiments in two typical "pronatalist" project settings (that is, the Matlab project, providing com- prehensive health care services for some 89,000 people in Bangladesh, and the Chogoria project in Kenya) to document that success is still assured provided that one is willing to invest enough money and human resources in such proj- ects. The Matlab and the Chogoria experiments could have been picked up by a Caldwell adversary to document the point that a demand for contraception can be generated in culturally adverse conditions provided that one tries hard enough. I think that these experiments show little either way. It is attractive to use an experimental research design, because a number of confounding variables can be brought under much better control than in cross-cultural or cross-regional experimental designs, but a problem arises when resources are continuously pumped into the experimental area until some effects emerge. Caldwell admits that the resources used in these two experiments are far above those available or even producible at the national level, from which I would conclude that the experiments have failed. A much better design would be one in which the resources devoted to the experimental area were limited to what can reasonably be mustered on a national level. I fear that the Matlab and Chogoria outcomes Lesthaeghe 259 would then have been quite different and more in line with the less spectacular results of other experimental designs that are not reported in Caldwell's paper. Most of my comments have pertained to the demographic transition aspects of Caldwell's paper and not to the second subject he deals with-the economic effects of population growth. From the historical record (once again) it is obvious that economic development owes much more to nondemographic fac- tors than to population growth rates (contrast France with England, for exam- ple), and I am not surprised that this apparent conundrum about population growth and economic performance has been carried forward in time. Although rapid population growth (say, rates above 2.0 percent yearly) is a risk factor, I must admit that the historical evidence with respect to pessimistic predictions regarding food production, unemployment, or slower increase in national income have not come out in a systematic way. Nor would I reckon myself among the optimists who view population growth as a motor of technological adaptation. Rather, I would prefer to judge each outcome according to the complexity of each situation. To sum up, I find the grand debate singularly unhelpful, and, to make matters worse, it has concentrated too much on eco- nomic performance to the detriment of ecological issues. REFERENCES Bongaarts, John. 1986. "The Transition in Reproductive Behavior in the Third World." Center for Policy Studies Working Paper 125. New York: Population Council. Boserup, Esther. 1985. "Economic and Demographic Interrelationships in Sub-Saharan Africa." Population and Development Review 11, no. 3: 383-98. Fargues, Philippe. 1989. "The Decline of Arab Fertility." English selection no. 1 in Population. Paris: Institut National d'Etudes Demographiques. Goode, William. 1963. World Revolution and Family Patterns. Glencoe, Ill.: Free Press. PRO C E E D IN GS O F THE W OR L D BAN K ANNUAL CONFE RE NCE ON DE VE L OPM E NT E C ONOM ICS 1 990 COMMENT ON "THE SOFT UNDERBELLY OF DEVELOPMENT," BY CALDWELL Susan H. Cochrane I wish to congratulate Professor Caldwell for being, as always, insightful and thought-provoking and for providing us with a broad sweep of the field. I cannot possibly address all his propositions. I have therefore selected three that I think are most central. Let me start by being contradictory, because that is always easiest. I wish to disagree with Caldwell's pessimistic view that a fertility decline in Sub-Saharan Africa is unlikely. In contrast, I believe that recent evidence suggests that a fertility decline has already begun in Sub-Saharan Africa, a decline that may be associated with changes in family-size preferences that are related in part to poor economic growth, in part to the history of mortality reduction and educational expansion in the successful countries, and in part to good family planning services. I agree with Caldwell's rough estimates on the high levels of health expenditures needed to stimulate fertility decline, but I would add, again more optimistically, that all these costs need not be borne by the local governments. I do agree with Caldwell that mortality decline is an important prerequisite for fertility decline, and that the epidemic of acquired immune deficiency syndrome (AIDS) may change the population balance. Second, I want to briefly discuss the economic consequences of rapid popula- tion growth. Third, I want to concur with him on the importance of the problem of AIDS, but I will suggest some slightly different policy implications relating to economic growth because of weight that should be given to its cost in forgone income. FERTILITY DECLINE IN SUB-SAHARAN AFRICA I believe that fertility decline has begun in Sub-Saharan Africa. Its expansion is perhaps uncertain, but there are hints that it has even begun in the western part of Sub-Saharan Africa. Five years ago I was much less optimistic. The scarcity of data in Sub-Saharan Africa has always plagued researchers, but two rounds of fertility surveys have become available over the last decade that have vastly expanded our knowledge base. The first is the World Fertility Survey (wFs) in Susan H. Cochrane is principal population economist in the Division of Population, Health, and Nutrition of the World Bank's Population and Human Resources Department. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 261 262 Comment the late 1970s and early 1980s; the second is the Demographic and Health Survey (DHS) in the late 1980s and early 1990s. We do know that fertility is very high in Africa. We all agree on this. In that respect and in the high family-size preference, Sub-Saharan Africa appears different from other regions today. There are similarities with other regions as well, such as the very high fertility in parts of the Middle East, the very high fertility among the least educated in Latin America until very recently, and the high levels of fertility among older women in North Africa. If we had evidence on Asia, or even more on Latin America, two, three, or four decades ago, we might find other similarities. A paper I did with Samir Farid using data from ten WFS surveys (Cochrane and Farid 1989) of ten countries around 1980 showed that in comparing the com- pleted fertility of older women with the current childbearing patterns in seven of the countries, fertility had probably risen in the past decade or two. In two countries it had remained constant, and in Ghana there was some evidence of a small decline. That paper was completed in 1986 but was not published as a World Bank working paper until last year for a variety of reasons, so it is already dated. In it we concluded that fertility decline had not begun in Sub-Saharan Africa, despite the fact that it had proceeded very rapidly in other countries in Latin America and Asia, even in very poor conditions. We also saw no evidence of impending decline, given both very high prefer- ences for children and small socioeconomic differentials in fertility. A new review of preliminary data from eleven DHS surveys for the late 1980s shows a different picture (van de Walle and Foster 1990). Part of the difference arises from the fact that the sample of countries is quite different (see table 1). The picture from the new surveys, some ten years after the first, suggests that in the past couple of decades fertility has possibly declined in all of them over the long run, and that this decline was as much as 10 percent in Botswana, Ghana, Kenya, Ondo State in Nigeria, Togo, and especially in Zimbabwe. Only Ghana, Kenya, and Senegal have data from both sets of surveys. I will focus on these. Botswana and Zimbabwe have data from other censuses and surveys that would allow a longitudinal analysis as well, but I don't think Caldwell is questioning that these declines have been real-only that they are special cases. I should add that I'm not particularly worried that we may have been blind in not seeing impending decline in fertility in our earlier paper. In a paper presented at a seminar last summer on "The Role of Family Planning Programs in Fertility Decline," one of the notables in the field, Ronald Freedman, concluded that "although the magnitude of fertility change falls far short of most optimistic predictions, the achieved change in Third World countries was not expected by most observers twenty to twenty-five years ago. Even ten years ago the subse- quent changes were not anticipated" (Freedman and Freedman 1989). With this in mind, let's look at the most recent data. First, let's examine Ghana, where we had seen some evidence of an earlier slight decline (table 1). Cochrane 263 Table 1. Fertility of Women Aged 45-49 Years Cumulative Current fertility fertility for all rate (in 0-5 years women aged 45-49 before each survey) Country (P) (F) F World Fertility Surveys Benin, 1981-82 6.27 7.08 0.89 Cameroon, 1978 5.18 6.40 0.81 C6te d'Ivoire, 1980-81 6.84 7.36 0.92 Ghana, 1979-80 6.71 6.47 1.04 Kenya, 1977-78 7.88 8.25 0.96 Lesotho, 1977 S.29 5.76 0.92 Mauritania, 1981 6.00 6.25 0.96 Nigeria, 1981-82 5.84 6.34 0.92 Senegal, 1978 7.16 7.15 1.00 Sudan (N.), 1979 5.98 6.02 0.99 Demographic and Health Surveys Botswana, 1987 5.8 5.0 1.16 Burundi, 1987 6.9 6.8 1.01 Ghana, 1986 7.25 6.43 1.13 Kenya, 1988 7.5 6.7 1.12 Liberia, 1986 6.4 6.3 1.02 Mali, 1987 7.06 6.73 1.05 OndoState,Nigeria, 1986 6.9 6.0 1.15 Senegal, 1986 7.0 6.4 1.09 Togo, 1988 7.28 6.21 1.17 Uganda, 1988-89 7.77 7.30 1.06 Zimbabwe, 1988 6.87 5.70 1.21 Sources: van de Walle and Foster 1990. This historical decline is confirmed in the most recent data. But whatever decline there was in previous decades has not continued. The current fertility rate in the most recent survey is 6.43, compared with 6.47 for 1979-80 (these fertility rates are measures of average number of live births per woman if the woman experi- enced the current age-specific fertility rates throughout her life). The other West African case, Senegal, does give some evidence of a decline both historically (9 percent) and, recently, more dramatically, 7.15 to 6.4 in the total fertility rate. Ondo State is difficult to evaluate, because the WFS was not available on a statewide basis, and southwestern Nigeria has lower fertility. Table 2 outlines the trend in fertility in Senegal, according to the WFS and the preliminary report of DHS. For all women up to the age of thirty-five years, the number of live births has decreased by about 8 to 10 percent. Socioeconomic differentials are also emerging among women twenty-five to twenty-nine, but are not apparent among older women forty-five to forty-nine. Knowledge of contraception has increased from 59 to 90 percent of women, and contraceptive use has increased more modestly from 4 to 12 percent, but modern methods have only increased from 0.6 to 2.7 percent. Although detailed analysis has not yet been done, 264 Comment Table 2. Fertility Indicators for Senegal World Fertility Survey Demographic and Health Survey Indicator (1978) (1986) Number of live births at age in years 15-19 0.44 0.32 20-24 1.69 1.57 25-29 3.39 3.09 30-34 5.28 4.74 35-39 5.94 6.16 40-44 6.80 6.83 45-49 7.16 7.27 Percentage knowing a method of contraception 59 90 Percentage using contraception Modern methods 0.6 2.7 Traditional methods 3.3 9.0 Source: Senegal Institute for Resource Development (1986). preliminary indications are that fertility decline may be incipient. Unfortunately, the preliminary data do not contain information on family-size preferences. These data are necessary to evaluate the unmet need for family planning, which would help determine what is likely to occur as family planning programs become more widely available. Kenya presents a more dramatic picture, as indicated in table 3. The number of children ever born to older women appears to have decreased slightly, and the current measure of fertility (that is, the measure among women of childbearing age in the year of the survey) has dropped from 8.25 to 6.7. Whether the drop in fertility is as large as it appears is uncertain, but there is substantial backup evidence to indicate that there is a real decline even if it has not been as large as 1.5 children. For example, substantial fertility differentials between rural and urban dwellers and among educational groups are now evident, which is an indicator of change. The mean number of children desired has also declined dramatically among all education groups. Contraceptive use has increased as well from 6 percent in 1977-78 to 27 percent in 1989. Further data analysis is required to reconcile various data sources and to check for the internal consis- tency of information on marriage, breastfeeding, and contraceptive practice, but I think that there is little doubt that something real is going on. Although there is, therefore, some basis for optimism about fertility decline in Sub-Saharan Africa, we need to examine carefully the policy conclusions of the decline that we have observed in Botswana, Kenya, and Zimbabwe and the decline that has stagnated in Ghana. I want to congratulate Professor Caldwell for pointing out that it does take resources to bring about decline. He has taken the discussion away from the false dichotomy over whether it is demand or the Cochrane 265 Table 3. Fertility Indicators for Kenya World Fertility Survey Demographic and Health Survey Indicator 1977-78 1986-88 Fertility trends P: Children ever born (of women ages 45-49) 7.88 7.50 F: Current fertility 8.25 6.70 Ratio: P/F 0.92 1.12 Totalfertility rate, by place of dwelling Major urban 5.90 Other urban 6.08 All urban - 4.30 Rural 8.48 7.10 Education None 8.28 7.50 1-3 years 9.21 Some primary - 7.50 4-6 years 8.43 Complete primary - 6.40 7 or more years 7.34 4.80 Mean number of children desired, by years of education None 7.64 5.40 1-3 years 6.88 Some primary - 4.60 4-6 years 6.36 Complete primary - 4.10 7-9 years 5.98 i0 or more years 5.13 3.60 Family planning prevalence Percentage who used contraceptivesa 6 27b Percentage who knew source of family planning 38-45 89b - Not available. a. Any method, among people currently married. b. Data from DHS 1989. Sources: World Fertility Survey and Demographic and Health Survey reports. supply of family planning programs that matters. It is clearly both, and there is interaction between the two elements that was left out of much of the earlier thinking. Ghana is an example. Although the percentage of women who want no more children in Ghana has doubled over the last decade, and more than half the women want to postpone their next birth, contraceptive use has only increased from 10 to 13 percent. Use of the pill has fallen while the proportion using abstinence has doubled. The single most frequent reason stated for not using contraception is lack of knowledge. All these factors indicate that lack of a well-funded and well-executed family planning program has played some role in the failure for fertility decline to continue in Ghana. How much resources are required is, however, uncertain. Programs such as Matlab in Bangladesh cost about US$15 per person a year for pill use based on a home-visit program such as that advocated by Caldwell (see Family Health 266 Comment International 1990). To achieve a 50 percent coverage of all women of repro- ductive age in each country by 1995 would require about US$1.85 per capita of the total population a year based on the Bangladesh costs. The costs of preventing a birth or achieving a couple of years of contraceptive protection, however, appear to depend on the level of demand to restrict fertility in the population. For example, in a recent paper (Cochrane and Sai 1989), we indicated that the cost of averting a birth declined by about $5 from an intercept of $350 for every additional percent of women wanting no more children. Thus data from about 1977 to 1982 indicated that the cost of a birth averted in Kenya was between $350 and $386 in 1987 prices. If the relation holds, one would expect that currently the cost might be as low as $125, because the proportion wanting no more children has risen to almost 50 percent. Thus how much it will cost will depend on the level of demand to restrict fertility. This gets back to Caldwell's hypotheses about what motivates fertility in Africa. He is far better able to judge this than I am, but it should be pointed out that the very sharp drop in desired family size in Kenya and Zimbabwe indicates that those preferences are not as stable as was once believed, even by me. The shift in preference may be the result of lagged responses to modernization, sharp changes in the calculus of the cost of children, or maybe even changes in the cost to the family of restricting fertility. This is a subject on which we are trying to gain some insight from a recently funded World Bank research project on Colombia, Tunisia, and Zimbabwe, all countries in which fertility has shown substantial decreases. Caldwell, in his paper, and Cochrane and Zachariah (1983) identify the role of infant mortality as of major importance in determining fertility decline in Sub- Saharan Africa, in part at least through its effect on the desire to restrict fertility. As Caldwell suggested, this implies that expenditures on family planning need to be incorporated into a substantial health program. Rough estimates of the cen- tral government health expenditures in Botswana, Kenya, and Zimbabwe indi- cate that they spend $21, $5, and $9 per capita, respectively. This is 2 percent, 1.5 percent, and 2 percent of gross national product, respectively. These are about half the percentages cited by Caldwell. In Kenya and Zimbabwe, however-I don't know about Botswana-the private sector, nongovernmental organizations, and so on provide an additional amount of coverage equal to that of the central government. This would indicate that health expenditures are about twice the indicated figures, and it shows a level of expenditure on health in these successful countries that is very similar to what Caldwell estimates. Until this point I have been focusing on fertility decline without addressing the question of whether fertility decline is a good thing. I now think it is necessary to move beyond the data-based discussion to the broader questions of conse- quences of both population growth and of AIDS and the policies that they imply. ECONOMIC CONSEQUENCES OF RAPID POPULATION GROWTH Caldwell is skeptical about the negative consequences of rapid population growth in Sub-Saharan Africa. His view is shared by many and for some very Cochrane 267 good reasons. I do not have the time or inclination to get into a long debate here. I am sure Ronald Lee's paper (this volume) will generate substantial debate on the topic. Let me say just three things. Just as Caldwell separated the effect of population growth into the rate of natural increase and the rate of migration, it is also necessary to separate out the differential impacts of mortality and fertility in assessing population growth's effect. It is necessary to make that separation because there are two directions of causation between population growth and income growth. The natural rate of increase is simply the difference between the crude birth rate and the crude death rate. In the long run, mortality is affected by the level of income. Thus increases in income will "cause" an increase in population growth by reducing mortality. This can easily mask the effect of population growth on income and income growth. Moreover, changes of population growth can be expected to have differential impacts depending on the combination of mortality and fertility rates that generate a rate of population growth. A country with high mortality and high fertility and one with low mortality and low fertility will both have low population growth rates, but they may have very different effects on the econ- omy. Reductions in mortality that represent improvements in the health of the population can be expected to stimulate productivity, possibly open up previ- ously disease-ridden areas, and result in savings in health care. Thus the two low-growth situations would have very different expected consequences. This hypothesis has been tested by Blanchet (1988). His paper deserves more attention than I can give it here, but his simple correlations for approximately eighty countries between population growth and income growth and between population growth and birth and death rates are instructive. For the three time periods he uses (1960-70. 1960-80, and 1970-80), whether samples weighted by population or unweighted samples are used, the death rate is always signifi- cantly negatively related to income growth, whereas the birth rate is also always significantly negatively related to income growth. Given this finding, what is the relation between population growth that com- bines birth and death rates and income growth? That depends on whether weighted or unweighted samples are used and on the time period. With the weighted samples, population growth is positively associated with income growth for the 1960-80 and 1960-70 periods and insignificant for 1970-80. For unweighted samples, population growth is not related to income growth until 1970-80, when its correlation is negative. What this tells us-which is what we should have known all along-is that the interactions between popula- tion growth and income growth vary, depending both on the balance of factors causing the population growth and on the economic circumstances. This pattern of interaction is borne out very realistically by the changes in population policies in Algeria and Nigeria when oil prices plummeted. Both of these countries could afford the burden of rapid population growth as long as oil prices were high. Once prices fell, however, they felt the pinch and shifted policy in favor of reductions in population growth. The second point I wish to make on economic consequences is that the conse- 268 Comment quences of high fertility and rapid population growth may not appear in the growth of income but in its distribution. This point seems to have been neglected in the focus on growth and efficiency alone. There are two ways that population growth will affect distribution adversely. The first, and most obvious, is that because the most rapidly growing part of the population is usually the poor, whose only resource is their labor, their wages will be suppressed by rapid growth of the labor force, all things being equal. The second is less researched. Many who discount the "population problem" say that markets will adjust through increased prices to ration scarce resources. The problem is that the poor will be the hardest hit by the adjustment, whether it is increases in the price of water or fuel or food. Third, the problem of population growth in the 1960s was as compelling as the problem of the greenhouse effect is today. A recent meeting of scientists and policymakers on the greenhouse effect said that the problem was so serious that we couldn't wait for data to confirm it. This is what happened in population. As Caldwell mentioned, a great deal of the work on the effect of population growth has been prospective in nature, and not enough has been done to explore care- fully the mechanisms whereby population growth's multiple effects on income and equity have been modified or magnified by policies. A nice exception is the paper Paul Schultz (1987) wrote for the National Academy of Sciences' study of consequences. He showed that rapid population growth in the developing world did not decrease enrollment rates but did reduce the quality of education. PROBLEMS AND POLICY IMPLICATIONS OF AIDS Let me conclude with a few words on AIDS. There can be no doubt that the human, economic, and demographic impacts of AIDS in Africa will be dramatic. Mead Over and colleagues (1988) have calculated the direct costs of treating MDS and the indirect costs of the loss of productivity and loss of life for two African countries. The range of estimates is quite large, given our small knowl- edge base. Nonetheless, they indicate two important things: (1) the direct plus indirect cost per human-immunodeficiency-virus-positive individual ranges from five to nineteen times per capita income in Zaire and Tanzania, and (2) the indirect cost of forgone earnings is much larger than the direct cost of treatment, ranging from five to sixty-six times as great (Over and others 1988). These drastic economic effects do not begin to convey how disruptive to life in general and to any orderly evolution of development such an epidemic may be. It is undoubtedly extremely important to address this issue. The cost of AIDS programs will have to cover treatment and prevention. In prevention there is the education on causes of AIDS, the supply of barrier methods of contraception, and the treatment of sexually transmitted diseases (STDS). To adequately deal with the last problem, as Caldwell correctly indicates, will require substantial resources. In particular, family planning programs will have to be altered to provide the clinical base for the diagnosis and treatment of STDS. This is in addition to the community- and home-visit-based distribution of family planning. Cochrane 269 In conclusion, I am not as convinced as Caldwell that the underbelly is so soft. I think that continuing rapid population growth will begin or has begun to have negative consequences for the countries of Africa. (Household negative conse- quences determine family-size preferences; in addition, they determine some of the negative effects of high fertility for society.) If these negative consequences only emerged with faster development, Caldwell might be right in his pessimism about a fertility decline in Sub-Saharan Africa, but what I am hearing is that the stagnation or even declines in incomes have made people become more sensitive to the economic costs of children, and this has led to substantial reductions in family-size preferences in parts of Africa. Where family planning services are available, this has translated into increased use of contraception and reduced fertility. The family consequences of high fertility are, thus, a major area of research that needs to be pursued in Africa if we are to understand the future course of fertility decline. That said, we must conclude that AIDS may change everything. REFERENCES Blanchet, Didier. 1988. "Estimating the Relationship between Population Growth and Aggregate Economic Growth in LDCS: Methodological Problems." In Consequences of Rapid Population Growth in Developing Countries: Proceedings of a United Nations Expert Group Meeting. New York: United Nations. Cochrane, Susan H., and Samir M. Farid. 1989. Fertility in Sub-Saharan Africa: Analy- sis and Explanation. World Bank Discussion Paper 43. Washington, D.C. Cochrane, Susan H., and Fred T. Sai. 1989. "Excess Fertility." In Dean T. Jamison and W. Henry Mosely, eds., "The World Bank Health Sector Priorities Review." World Bank Population and Human Resources Department. Washington, D.C. Processed. Cochrane, Susan H., and K. C. Zachariah. 1983. "Infant and Child Mortality as a Determinant of Fertility: Its Policy Implications." In Fertility and Family Planning: Proceedings of the Expert Group on Fertility and Family. New Delhi: United Nations. Family Health International. 1990. Investing in Reduced Population Growth: Costs of Family Planning in the Year 2000. Research Triangle Park, N.C. Freedman, Ronald, and Deborah Freedman. 1989. "The Role of Family Planning Pro- grams as a Fertility Determinant." Paper Presented at an International Union for the Scientific Study of Population seminar on the Role of Family Planning Programs as a Fertility Determinant. Tunis, Tunisia. Over, Mead, Stefano Berdozzi, James Chin, B. N'galy, and K. Nyamurekung'e. 1988. "The Direct and Indirect Costs of HIV Infection in Developing Countries: The Cases of Zaire and Tanzania." In Alan Fleming, ed., The Global Impact of AIDS. New York: Alan R. Less. Schultz, Paul. 1987. "School Expenditures and Enrollments, 1960-80." In D. Gale Johnson and Ronald D. Lee, eds., Population Growth and Economic Development: Issues and Evidence. Madison: University of Wisconsin Press. Senegal Institute for Resource Development. 1986. Enquete Demographique et de Sante au Senegal: Rapport Preliminaire. Dakar. Van de Walle, Etienne, and Andrew Foster. 1990. Fertility Decline in Africa: Assessment and Prospects. World Bank Technical Paper 125. Washington, D.C. PROCE E DING S OF THE W O R L D BAN K ANN UAL C ON FE RE NCE ON DE V E LO PME NT E C O NOM I C S 1 9 90 FLOOR DISCUSSION OF THE CALDWELL PAPER The discussion opened with observations from members of the audience. A World Bank participant emphasized the role of international and domestic migration in the population dynamics of Africa. He noted that the economic crisis in Africa clearly has reduced mobility and has hampered the efforts of individuals to survive and develop, and he argued that population redistribution deserves much more attention in research on development policy. The same participant also noted that economists have so far not understood the full con- tours of economic growth in Africa, especially as regards activity in the informal or unrecorded economy. Revising the economic indicators to account for these phenomena would change the picture of the relationship between population growth and economic growth, he said. Another member of the audience agreed with Caldwell's judgment that the failure of mortality to decline, especially child mortality, may slow the decline of fertility in Africa. But she cited recent Demographic and Health Survey (DHS) data from Nigeria as indicating significant fertility declines stemming from increased use of contraceptives, especially among educated women. She also wondered whether Caldwell, in expressing skepticism about fertility transition in Sub-Saharan Africa, had fully considered the effects of economic crisis on population decline, especially for a country such as Nigeria, in the light of the recent austerity measures there. A member of the audience expressed surprise at Caldwell's statement in the paper about the efficacy of government intervention in producing fertility declines at any level of socioeconomic development. He noted that even where government antinatalist policies have been very strong, as in China, it remains questionable whether development and broader socioeconomic factors have not been substantially more responsible for fertility declines. Moreover, even if government fertility-limiting policies can produce fertility declines at varying levels of development, the question is whether such declines are sustainable. This same participant agreed with Caldwell on the importance of the public and democratic provision of health services and education, especially of females, in moderating population growth. He suggested that World Bank policies toward the pricing and provision of such public services are narrowing access and thereby might delay the fertility decline. This session was chaired by Frederick T. Sai, senior population adviser in the Population and Human Resources Department of the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 271 272 Floor Discussion of Caldwell Paper Jack Caldwell began by replying to Ron Lesthaeghe's comments on his paper. To Lesthaeghe's comment about the need for caution in exporting population prescriptions learned in one context to another, and the need to stay away from grand theory, Caldwell replied that despite the need to tailor transition theories to the specific socioeconomic circumstances of the countries or regions in ques- tion, broad forces-such as the evolution of nuclear families, female education, duration of postpartum sexual abstinence and breastfeeding, and so on- provided useful cross-cultural indicators of fertility transition. Caldwell noted that demographers had correctly predicted fertility transition in Asia as far back as the 1950s by looking at such indicators. He had himself produced a popula- tion projection showing declining fertility in Singapore in 1959 and was assailed because it contradicted prevailing stereotypes about Chinese family structures and marital practices. In fact, the decline was already under way, but the data were not yet reflecting it. Caldwell also briefly acknowledged Lesthaeghe's comment that his paper had not given consideration to "halfway" fertility transitions, which are followed by a plateau of stable population growth, or even temporary reversals. Caldwell said, however, that such phenomena were part of the generally complex path of many transitions, and he was doubtful if there was a simple model of halfway transition. Caldwell also responded to Lesthaeghe's observation that Sub- Saharan Africa might be subject to a "crisis-led" decline reflecting a rising cost of children, perhaps in the context of structural adjustment. Caldwell acknowl- edged that Sub-Saharan governments were responding to economic crisis and that if strong enough, this response might have some effect in catalyzing fertility transition. He remained skeptical, however, about the individual-based, crisis- led fertility transition. In fact, he noted, in times of economic crisis most people saw themselves become more dependent on children. Responding to the observation about the possibility of imminent fertility declines in Nigeria, Caldwell noted that he was not quite so sanguine. He interpreted the DHS results to show lower levels of current contraception use by married women than government statistics on contraception provision would indicate. He said he stood to be corrected, but at least in the Ekiti area of Ondo State in Nigeria, there was evidence that fertility had risen due to a decline in postpartum sexual abstinence from the relatively high levels relative to the rest of Sub-Saharan Africa. The evidence suggests that the duration of postpartum sexual abstinence was linked to the duration of breastfeeding, which had declined steeply during the Nigerian oil boom, partly because there was more supplementary food. The rise in fertility, Caldwell felt, was also consistent with the World Bank's projections, which show in the immediate future a substantial number of countries where fertility is likely to rise. Caldwell responded to the participant who had questioned the efficacy of government intervention in fertility in China by arguing that the methods used there had been so absolute and thoroughgoing that they would produce fertility declines at any stage of development. He pointed out, however, that the social Floor Discussion of Caldwell Paper 273 and political culture of Africa made it unlikely that any African government could get away with such extensive intervention; so, he agreed that it would not be sustainable in Africa. Regarding Susan Cochrane's (discussant) remarks, Caldwell agreed that Bot- swana, Zimbabwe, and central (and adjoining parts of eastern and western) Kenya evidenced clear fertility declines, but he wondered whether the World Bank's projections for rapid fertility declines in Botswana and Zimbabwe could be supported by the existing evidence. He cautioned against attributing the lagging fertility decline in Ghana and West Africa to economic stagnation and program failures, such as the poor availability of birth control pills in Ghana. Caldwell suggested-as a reason for his skepticism about the Bank's population projections-that although the West African coast has had a longer history of development in some ways than much of the rest of Africa, it also has more cultural structures that are likely to oppose fertility decline as compared with East and Southern Africa. For example, Caldwell was skeptical about the accu- racy of research findings-especially comparisons of different general surveys studying fertility control-indicating that under economic stagnation, West African women were increasing sexual abstinence, amounting to a doubling of the use of abstinence in Ghana. Female abstinence in various forms has always been used in West Africa, and actual figures that the general-purpose surveys yield for how much female abstinence is being applied for contraception there- fore are likely to be haywire. To study this, Caldwell maintained, one needs a specific research project looking at the role of female abstinence. Caldwell felt that Cochrane had reported some very interesting findings about the relation- ship between fertility, mortality, and income growth, which he would like to investigate further. Caldwell fully agreed with Cochrane that the implications of the AIDS epi- demic in Africa needed to be studied more extensively, and he suggested that all the international bodies dealing with family planning were avoiding dealing with the epidemic's implications for family planning programs. Caldwell reiterated his belief that tackling AIDS in Sub-Saharan Africa is one of the world's great obligations of the late twentieth and early twenty-first centuries. He suggested that this effort required devoting significantly increased levels of resources to health services in Sub-Saharan Africa, partly with the focus on treating sexually transmitted diseases that increase the transmission rate of AIDS. He also noted that this effort would have the additional benefit of enhancing child survival and so decreasing the motivation to have additional children. Cochrane followed up on this topic by emphasizing that to diagnose and treat sexually transmitted diseases will require clinics-community-based or house- hold visit systems will not suffice-and this will require additional resources. Cochrane added that Caldwell is absolutely right in emphasizing the relevance of the AIDS crisis for family planning efforts, especially in light of the protocols for interventionist family planning-whether injections, intrauterine devices, implants, or sterilization. 274 FloorDiscussion of Caldwell Paper Sai (chair) noted that the AIDS crisis was now being addressed in family plan- ning work for West Africa by the International Planned Parenthood Federation, with funding from the U.K. Overseas Development Agency. He added that as a West African, he agreed with Caldwell that weight should be given to the cultural factors in Sub-Saharan Africa that discourage fertility decline. But he also suggested that the lagging economic development of some areas of Africa that were once the equal of Asian economies, such as the Republic of Korea, can also be attributed to "bankrupt ideologies" and dilatory implementation in the policy realm. Sai thanked Caldwell, the commentators, and the audience and adjourned the session. PROCE ED ING S OF T HE W OR LD BANK ANN UAL CO NFE RE N CE ON DE V EL O PME N T EC ONO MI CS 1 9 9 0 Population Growth, Externalities to Childbearing, and Fertility Policy in Developing Countries Ronald D. Lee and Timothy Miller Government-financed family planning programs that assist individual couples to attain their desired number of children are easily justified. Butgovernment policies that coerce or use financial incentives to influence couples to alter their desired number of children require stronger justification. Such justification may reside in the externalities to childbearing-the costs and benefits of children that are passed on by parents to soci- ety. Externalities to childbearing might include public costs of education, health, and pensions, as well as taxes to be paid by children in the future; cost sharing for public goods and social infrastructure over an enlarged tax base; the dilution of per capita value of various forms of collective wealth; and the reduction of wages and per capita incomes in the future. We estimated these externalities for a number of developing countries. Although the net total estimated externality was typically negative, it domi- nated measurement error only when public holdings of natural resources were impor- tant. Public expenditures on health, education, and pensions, financed by proportional taxes, led to negative externalities in most developing countries. There are many sources of positive and negative externalities, and each estimate is uncertain, so the total externality is itself highly uncertain and often does not provide a clear case for policies going beyond family planning. Inclusion of environmental effects might alter this conclusion. Why should governments take it upon themselves to intervene in the family- building decisions of their citizens? Is this simply a matter of unwarranted government meddling, as some might argue, or is it a justified effort to improve the public welfare, as most believe? Roughly speaking, we can divide the completed fertility of a couple into two categories. The first of these, wanted births, means here the number of children that the couple would choose based solely on their tastes and socioeconomic circumstances. The second category, excess births, means here any additional children the couple have because of imperfect contraception resulting from lack of information; social or religious opposition; and costs of contraception, Ronald D. Lee is professor of demography and economics at the University of California at Berkeley. Timothy Miller is a graduate student in demography and economics at the University of California at Berkeley. The authors are solely responsible for the views and results of this paper. They are grateful to John Quiggen for pointing out the externality arising from proportional taxation. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 275 276 Population Growth, Externalities to Childbearing, and Fertility Policy including time, travel, and embarrassment. Governments have many excellent reasons for sponsoring projects such as family planning programs, which enable couples to approximate more closely their wanted births by reducing or elim- inating excess births. Although actual family planning programs often attempt to persuade couples to choose low fertility, a pure family planning program limits its activities to assisting couples attain their reproductive desires without seeking to alter them. As Kingsley Davis (1967, p. 731) pointed out some years ago, however, ''there is no reason to expect that the millions of decisions about family size made by couples in their own interest will automatically control population for the benefit of society." Davis's influential article concluded that "in under- developed countries . . . the elimination of unwanted births would still leave an extremely high rate of multiplication." In such circumstances, policymakers might well consider fertility policies going beyond pure family planning-that is, policies designed to alter the number of births that a couple wants or chooses. One such policy is a system of financial incentives to influence fertility outcomes or contraceptive use. Another is to set the fertility level by fiat, as the one-child family policy does in China. Such policies can be justified when the desired fertility of individual couples is socially nonoptimal-when there is a difference between the social and individ- ual valuation of the costs and benefits of a birth. For example, the World Development Report (World Bank 1984) states that a justification for action by governments "is the gap between the private and social gains from having many children. . . . One reason [these gains] differ is the existence of 'externalities' " (1984, p. 54). Childbearing externalities are consequences of childbearing that do not impinge on the decisionmaking couple but rather accrue to society at large.' A good case can be made that a laissez-faire fertility policy is appropriate in the absence of externalities, and, in fact, some scholars (see Ng 1986, for example) have forcefully argued as much. Other scholars and governments believe that there are positive externalities to childbearing, so that governments actually should encourage fertility through financial incentives, by withdrawing support for contraceptive services and abortion, and possibly by making con- traception and abortion illegal. Although pronatalist policies are most common in the industrial countries, where support for the elderly dominates other con- cerns, a number of developing countries, such as Malaysia, have them as well. It is thus important from a policy point of view to determine whether there are significant externalities to childbearing, and, if so, whether on net they are 1. "Externalities" are most commonly defined as in Nerlove, Razin, and Sadka (1987), who say they occur when "the activity of one agent indirectly affects the production possibilities or utilities of other agents outside the price system" (1987, p. 38). Sometimes effects occurring through the price system are referred to as pecuniary externalities, as distinct from technical or true externalities, which are as defined above. Leeand Miller 277 positive, negative, or zero. Yet, surprisingly, hardly any efforts have been made along these lines. This statement may appear peculiar, given the amount of work that has been done on consequences of population growth in general (as reviewed, for example, in World Bank 1984; National Research Council 1986; and Kelley 1988). However, most consequences of childbearing, whether at the microeconomic or macroeconomic level, are not true (or technical) externalities. Rather, they are borne directly by the children's parents or pass through market channels (and are therefore only pecuniary externalities). For example, the birth of a child may eventually shift the supply curve of labor and thereby depress the wages and average incomes of the next generation. But this may be no more an externality than is the effect of a new shoe factory in depressing the price of shoes; it harms other shoe manufacturers but benefits consumers. When all consequences of actions are mediated by a fully competi- tive market, theory tells us, the outcome will be Pareto-optimal (provided that there are no nonconvexities, such as increasing returns to scale). Recent analysis confirms that this theorem applies in the intergenerational demographic context as well (see Blandy 1974; Willis 1987; Nerlove, Razin, and Sadka 1987; and Lee, forthcoming, a). Therefore, when we study externalities to childbearing, we must take care to isolate consequences that do not pass through the market and are not borne directly by the parents. Pecuniary externalities are also potentially important for population policy, but they fall under the general heading of consequences of population growth, and they bear quite a different relation to policy than do technical externalities. Childbearing externalities are not the only possible justification for policies that go beyond family planning. Society may have goals different frorn those of its individual members, for example. It may care more about income distribu- tion; it may discount the future less heavily than do individuals; or it may aspire to greater political, cultural, or military influence in the international arena. Furthermore, actual economies depart from the assumptions of perfect competi- tion and perfect information. In any event, childbearing externalities are assessed in the context of an institutional environment that may itself be highly inefficient. In such circumstances, the absence of externalities no longer guaran- tees a socially desirable fertility level. What we examine here is therefore only a piece of the puzzle, albeit a very important piece. In this paper, we first outline a way of thinking about reproductive exter- nalities, and we identify the categories of externalities to which this approach leads. We then present estimates of externalities arising within each of these categories for a number of developing countries: Bangladesh, Brazil, India, Kenya, Mexico, and Saudi Arabia. For comparison, we present results for the United States as well. After considering some extensions of the basic framework of analysis and some reservations about the results, we discuss the policy impli- cations of this research. 278 Population Growth, Externalities to Childbearing, and Fertility Policy I. CONCEPTUALIZING EXTERNALITIES TO CHILDBEARING It is simple and instructive to view the question of externalities in the follow- ing way, which is based on Nerlove, Razin, and Sadka (1987; Phelps 1968 developed a similar model; see appendix 1 for a more formal description of the model). Consider a generation of couples that care about their own consump- tion, about the number of children they have, and about the future consumption of their children as adults. Couples realize that their children's future consump- tion depends, to some degree, on the number of children that they, the parents, have, because this will affect the bequest, such as the share of the family farm, that the parents can leave each child. Couples may be assumed to have perfect information about the future, although this is not necessary. Each couple chooses some level of fertility that maximizes its satisfaction, given its prefer- ences and constraints as just described. The key question is whether the couples could do better if they reached a collective decision about fertility rather than choosing their fertility individually. For example, if they acted collectively, they could jointly choose the size of the labor force in the next period and thereby choose the wage level their children would face as adults; this choice is not available to them as individuals. Put differently, could a social planner with coercive powers over fertility, and acting solely in the interests of the couples as a group, increase their satisfaction over the level they could attain choosing their fertility individually under a policy of laissez-faire in reproduction? If so, then there is a difference between the implicit net cost of a child as perceived by the parents and as recognized on their behalf by the planner. The difference between these perceived costs equals the external costs of a child, and it is this difference that we seek to estimate below. A comment is required on the time frame of the calculations. The model just outlined covers two periods, each of roughly thirty years. Taking the model literally, the effects of the first generation's fertility on the second generation's consumption can be evaluated over the course of this second thirty-year period, with or without discounting and more elaborate allowances for survival. But of course life goes on, and the second generation also will have children, and so on forever, we hope. One common way to incorporate an infinite time horizon is to use an infinitely recursive altruistic utility function (see some of the other models in Nerlove, Razin, and Sadka 1987, for example). However, in regard to the optimality of laissez-faire fertility in a fully private competitive economy, the recursive altruistic model gives the same answer as the simple two-period model. A second approach, which we take here, recognizes that each additional birth to a population of N people will increase the population size forever after by a factor of (1 + 1/N), because in adding a birth we are also eventually adding an average number of descendants of that birth for every generation thereafter.2 2. A more exact calculation can be made in the same spirit using the concept of reproductive value; for the case of India, which we examined in detail, this differed only negligibly from the (I + 1 IN) approximation. Leeand Miller 279 Discounting is also appropriate, and we have done so at the rate of population growth, which should equal the real rate of interest in a society with no growth in per capita income. If per capita income does grow, then many of the quan- tities entering in the calculations would be expected to grow at the same rate, and the discount rate would be correspondingly higher, so it would make little difference.3 Obviously this is a rough assumption, but we believe that in most cases using a higher discount rate would move the estimates of externalities closer to zero rather than increase them. There are several points to note about this general way of framing the ques- tion. First, the only concern is with the well-being of the current generation; the well-being of future generations matters solely to the extent that current parents care about the subsequent well-being of their own children. Some would argue that society should have loftier goals than do individual parents and should care more about the welfare of future generations. That is a defensible position, but it would only strengthen whatever case for governmental intervention results from the present baseline analysis. Second, as posed, the question about exter- nalities has no answer, for the answer depends on the institutional context, particularly on the nature and activities of the public sector and the nature of property rights in productive factors. Third, there is no particular interest either in population size or growth per se or in the growth rate of per capita income. Interest in these quantities arises only from the fundamental interests of couples in their own consumption and fertility and in the consumption levels of their children. It might well be that under some set of initial conditions, the optimal outcome would involve growing population and declining per capita income (see Lee 1990). Fourth, so far, the society has been viewed as a homogeneous whole, with no differences among couples in tastes or wealth. We will try to relax this assumption later. Fifth, although the question may appear to be framed nar- rowly here, this approach does directly address many of the concerns that have been expressed about population growth, particularly in response to Hardin's (1968) classic article on the "tragedy of the commons." II. SOURCES OF EXTERNALITIES TO CHILDBEARING Laissez-Faire in a Fully Competitive Private Economy Analysis of the basic model (see appendix 1), with full private ownership and no public sector, yields the surprising result that the laissez-faire outcome is identical to the planner's, and no true childbearing externalities occur (see Nerlove, Razin, and Sadka 1987; Willis 1987; and Phelps 1968).4 The parental 3. If the real rate of interest equals the rate of population growth plus the rate of growth of per capita income, and if quantities such as educational expenditures per pupil and taxes grow at the rate of per capita income, then present values will be the same as for the current calculation, which assumes no growth in per capita income. 4. Pitchford (1985) appears to show that externalities do occur under laissez-faire when some resource in fixed supply leads to diminishing returns to capital and labor used in fixed proportions. It seems to us, 280 Population Growth, Externalities to Childbearing, and Fertility Policy concern that subdivision of the farm would reduce their children's future income corresponds exactly to the fertility consequences perceived by the planner. We refer to this as the baseline case. Certain deviations from the conditions of the baseline case may lead to childbearing externalities, as we now discuss in general and qualitative terms before attempting to assign quantitative estimates. Collective Wealth The most basic kind of reproductive externality arises when there is some form of asset held collectively to which all members of the population have free right of use or access. If the per capita value or usefulness of this asset is diminished as population grows, then there is an externality to childbearing. There are two problems to distinguish here. First, it is well known that a common property resource may be overused and degraded if the existing popu- lation has free access. Second, through high fertility, the population may grow too rapidly, because the effect of any couple's fertility on their own children's use of the common property resource is very small. It is this second kind of problem that concerns us here (see Lee, forthcoming, b). In the case of a privately owned family farm, each couple takes into account the effect of their fertility on their children's future share of the farm; in the case of a collectively owned asset, this effect is diluted and spread thinly over the entire population, so for any individ- ual couple it is negligible. Even in societies with well-developed systems of property rights, there are many collectively owned or shared assets: publicly owned land; government- owned industries; government-owned mineral rights; rights to fisheries and inland waterways; parks; the ozone layer; other aspects of the atmosphere, such as the temperature, the absence of acid rain, and so on; and public debt. In some socialist societies, the bulk of the productive resources may be collectively owned. The value of the externality arising through these forms of collective wealth can be calculated once the value of the collective wealth is known. Generally, it will simply equal the per capita value of the collective wealth for the next generation. For example, consider a one-class society in which all the land is collectively owned. Rural China before the "new responsibility system" might fit this description, as might certain African regions in which land is tribally owned. Couples correctly believe that their children's second-period income is independent of their own fertility decisionmaking (because the only productive asset is collectively owned). The planner, however, sees the true shadow cost of however, that the result may be mistaken. On the one hand, if a class of landlords owns the fixed resource, the laissez-faire outcome has not been shown to be non-Pareto-optimal, because the interests of the landlords are not considered. This is like the case we consider below of a two-class society. On the other hand, if ownership in the resource is evenly distributed in the population, then the budget constraint of the parents does not take into account the dilution of rents per capita to this scarce resource when fertility is higher. But it is precisely the foresight of this dilution that leads to optimal fertility choices. Lee and Miller 281 a child (for everyone) as second-period consumption minus the marginal prod- uct of labor rather than the average. The difference in perceived costs is roughly equal to the per capita value of implicit rents in the first period (see appendix 1). This is the same as saying that the childbearing externality arising from collec- tive ownership of land equals the per capita value of land, which is the total rental value of the land divided by the size of the next generation. Of course, the rent must be counted not only for a single year but over the adult life of the child (or, more realistically, the present value of the perpetual rent should be calculated). If the share of land (implicit rent) is 0.4 of average agrarian income, and if life expectancy of children in adulthood is twenty-five years, then this amounts to ten times the per capita income of the rural popula- tion, which is a very large amount. This constitutes a negative externality. (This number would be reduced by perhaps half with appropriate discounting, but it would be roughly doubled if we took into account that about half the rural population is dependent.) To evaluate the childbearing externalities arising from collective ownership of assets, or from unowned environmental wealth, we must calculate the per capita value of all such wealth. For some forms of wealth this is relatively straightfor- ward; for others, it is nearly impossible. Public Sector Provision of Public Goods Pure public goods can be enjoyed by any number of people with no congestion or diminution of individual satisfaction. The cost of providing public goods can be shared by the whole population, so public goods cost less per head when the number of taxpayers is larger. This leads to positive externalities to childbear- ing, because once again the eventual cost-spreading benefits of a couple's fertility are themselves spread widely over all couples and are negligible so far as the individual couple is concerned. The usual examples of public goods are defense expenditures, radio and television broadcasting, weather forecasting, expendi- tures on research and development and on culture and the arts. There are also quasi-public goods, which are subject to some congestion, but less than in proportion to the population using them. Important examples include many kinds of social infrastructure: transportation and communications networks, water supply and sewage systems, and harbors. In the case of pure public goods, paid for by an equal tax on each household, the two-period model of Nerlove, Razin, and Sadka (1987) can be used to show that there is a positive externality to childbearing just equal to the per capita cost of the public good (see appendix 1).5 It is a relatively simple matter, then, to calculate this cost using governmental budget data. 5. Nerlove, Razin, and Sadka (1987) analyze the special case of dynastic taxes to fund the public good; when taxes are levied in this way, no externality arises. We know of no examples of real-world dynastic taxes, so the case discussed in the text appears to be the relevant one. 282 Population Growth, Externalities to Childbearing, and Fertility Policy The Effect of Proportional Taxes If taxes are levied as a proportion of income or consumption, as is common, then they distort the trade-off between fertility and children's future income, leading parents to have more children than is optimal (see appendix 1).6 The marginal addition to the population contributes taxes equal to the tax rate, t, times the marginal product of labor (MPL) while receiving services and transfers equal to t times the average product of labor (APL), leading to a negative exter- nality equal to tAPL - MPL). This externality is superimposed on other effects resulting from the public versus private nature of the services provided and the age distribution effects to be discussed below.7 Proportional taxation converts a portion of a pecuniary externality (the effect of an incremental birth on the per capita income of the next generation) into a true externality. Intergenerational Transfers When an age group consumes more than it produces, it does so by virtue of some kind of transfer from a different age group that produces more than it consumes. Children and the elderly are the prime examples of dependent age groups. In developing countries, most such transfers take place within the family and therefore influence the couple's fertility decisions. For example, a couple may have an additional child to assure their support in old age. In the industrial welfare states, however, many substantial transfers take place through the pub- lic sector in the form of expenditures on health and education, as well as direct financial transfers through pension schemes and family allowances. The other side of the coin, of course, is that people pay taxes to support these transfer programs, and, on average, an individual's receipt of transfers and payments of taxes should balance over the life cycle, when discounted appro- priately. Analysis shows that the appropriate rate of discount is the population growth rate, and that when population growth is more rapid, tax rates must be higher if the people receive transfers at a younger average age than they pay taxes (see Lee, forthcoming, a). In many developing countries, we see that higher fertility and more rapid population growth do indeed have this effect, requiring higher taxes. In such countries there is concern that the public costs of education create negative externalities to childbearing. In some richer develop- ing countries and in the industrial countries, the importance of public sector transfers to the elderly reverses the situation, and higher fertility permits lower tax rates. In addition to these public expenditures on health, education, and pensions, 6. John Quiggen pointed out this externality in comments on an earlier draft of this paper. 7. We have calculated the full cost-diluting value of a future taxpayer for public good expenditures, but an incremental (as opposed to average) population member will contribute less than the average amount of taxes, so our calculated positive externality for public goods must be adjusted downward appropriately. The proportional tax externality, calculated from the overall tax rate, makes this adjust- ment. Lee and Miller 283 which are relatively age-specific, it may also be that population growth gener- ates increased demand for various other governmental services, such as the social infrastructure discussed above, before it generates the increased taxes to pay for them. This also is a matter of the difference between the average ages at which one requires governmental services and pays taxes. For all these reasons, externalities arise through the influence of fertility on the age distribution of the population. To calculate the externalities arising from public sector intergenerational transfers, we must first develop age profiles of receipt of transfers and of pay- ment of taxes. This requires inspection of educational enrollment rates and budgets; of health care expenditure and utilization data; and of government data on pensions. Lacking direct data on the age incidence of taxes, we have assumed them to be proportional to labor earnings and have used age-earning profiles. In many countries tax revenues are raised mainly by taxes on consumption, so this assumption may not be too inaccurate; in any case, it appears to be the best we can do for the present. The next step in the estimation is to obtain appropriate stable population age distributions, reflecting current vital rates. In many developing-country populations, these should be fairly close to the actual age distribution of the population. Using the age profiles for transfers and the stable population, the next step is to calculate the average ages at which a transfer is received and at which taxes are paid. The externality to childbearing then equals the average age of receipt of transfer minus the average age of payment of taxes, times the per capita value of transfers received annually in the stable population, divided by the average age at childbearing times the total fertility rate (see Lee, forthcoming, a).8 For non-age-specific governmental expenditures other than pure public goods, we have assumed that the mean age of demand for services is midway between the average age of the population and the average age of earning and otherwise proceeded as just described. Economies of Scale and Induced Technological Progress It has sometimes been suggested that small countries, with low gross national product (GNP), do not provide sufficiently large markets to support an efficient level of operation for many industries, or perhaps for the economy as a whole. Arguments of this sort for positive economies of scale have lost much of their force as international economic integration proceeds, and there are many exam- 8. It is assumed that the public sector budget is balanced. The tax rate is then set so as to generate exactly the revenues needed for transfers, given the age distribution. Note that with these assumptions, the present value over the life cycle of the transfers received and the taxes paid is zero when the discount rate equals the population growth rate. That is, if the population is growing, each person will be paying taxes to support more transfers than they themselves received, and discounting at the population growth rate must set the present values equal. If a birth is forgone, there will be no net gain in the transfer budget unless government funds can be invested at a rate of return greater than the population growth rate (plus the rate of growth in per capita income). The calculation described in the text captures a subtler effect- the tendency of the child to raise the population growth rate and alter the population age distribution. Table 1. Rough Estimates of Externalities to Childbearing in Selected Countries (U.S.$ of reference year) United States, Bangladesh, Saudi Arabia, Item 1985 India, 1981 1980 1986 Kenya, 1986 Mexico, 1985 Brazil, 1983 Collective wealth Mineral rights -3,798 0 to - 13,000 -289 - 189,554 -168 -10,003 -7,433 Government debt +10,013 +107 +41 0 +151 +1,075 +755 Government capital 0 -27 -1 -19,340 0 0 0 Government land -813 - 10 to -200 -11 n.a. -180 - 82 -10,600 Fisheries -224 -50 -29 n.a. -33 -165 -84 Foreign aid 0 -35 -346 0 -929 -93 -60 Subtotal (net) +5,178 -15 to - 13,205 -634 -208,894 -1,159 -9,268 -17,422 Public goods Defense, public order +57,637 +324 +98 +48,255 +472 +786 +682 Social infrastructure +9,968 +352 +429 +13,463 +692 +4,964 +1,025 Public administration +7,345 +35 +88 +19,383 +321 +2,192 +1,526 Subtotal(net) +74,950 +710 +615 +81,100 +1,484 +7,942 +3,233 Proportional taxes O to -41,000 O to -350 O to -140 O to -9,000 O to -540 O to -3,500 O to -3,500 00 Intergenerational transfers -4 Health,education,pensions +35,713 -25 -6 -1,358 -48 -187 +463 Other public expenditures -11,025 -34 -16 -1,330 -22 -247 -168 Subtotal (net) +24,688 -59 -22 -2,688 -70 -434 +295 Total net externality + 64,000 to + 640 to -40 to -130,000 to +250 to -1,800 to -13,900 to (rounded) +105,000 -12,900 -180 -139,000 -280 -5,300 -17,400 GDP per capita 16,376 257 159 6,502 316 2,350 1,448 Ratio of net externality to +4to +6 +2to -50 Oto -1 -20 to -22 +1 to-1 -ito -2 -10 to -12 GDP per capita n.a. Not available. Note: The ranges in the last row result from the ranges on specific items and do not indicate the overall degree of uncertainty, which is far greater. Entries for public wealth, public goods, and age distribution effects assume that all taxes are head taxes; the entries under proportional taxes give the added correction if instead (as is highly likely) taxes are proportional to income or consumption. "Other public expenditures" includes half of the value of quasi-public goods (most infrastructural expenditures), plus all other expenditures not counted elsewhere. These are allocated by age midway between the average ages of the population and the labor force. "Social infrastructure" includes transportation, communications, research and development, and economic services. "Public administration" includes foreign relations and general government administration. For most countries, Hotelling's principle (Miller and Upton 1985) was used to estimate the value of the country's main nonrenewable asset (oil, natural gas, coal). For Kenya and Brazil, multiple mineral values assume costs of extraction of 60 percent of resource value. The value for public land in Kenya includes total estimated value of forests, assumed to be publicly owned, and the profits of restaurants and hotels, largely attributable to tourism in the national parks. The figure for land in Brazil attempts to value the rain forest (assumed to be one-quarter publicly owned), based on sustainable harvest, and without valuing its role in the global ecology (Peters, Gentry, and Mendelsohn 1989). It also includes a valuation of current hydropower. The stable population growth rate is the discount rate implicit in the method used to calculate the externality from intergenerational transfers. For other items the discount rate is 0.02, which is close to the stable population growth rate for all countries except Kenya and the United States. For details on methodology, see Lee (forthcoming, a). For data sources, see Appendix 2. Lee and Miller 285 ples of relatively small economies experiencing very rapid growth. It may be so that larger, denser, or more rapidly growing populations stimulate technological progress by shifting both the supply schedule and demand schedule for invention and innovation (see Simon 1986; Boserup 1981; and Phelps 1968). So far, such effects have largely defied measurement, but they may be important nonetheless. In this paper, however, we will not be able to take them into account. III. ESTIMATES OF EXTERNALITIES TO CHILDBEARING Having discussed several categories of externalities to childbearing and the method to be used to evaluate each one, we now attempt to attach numbers to them. This evaluation is based on a variety of published information, including the results of household surveys, data on mineral reserves, census data, and national and subnational budgets. It should be understood that these estimates have been made without any special knowledge of the countries to which they refer and that this introduces an additional layer of uncertainty onto whatever arises from the many assumptions made. Table 1 draws together the estimates for six developing countries, with comparisons to estimates for the United States. Estimates of Externalities Arising from Collective Wealth The table reveals some dramatic differences among countries in the general importance of the various items considered under this heading. Mineral reserves dominate the entire calculation in India and Saudi Arabia. For Saudi Arabia this is surely no surprise, because its oil revenues are well known to be extremely important. Yet relative to per capita GDP, the role of coal reserves could possibly be even more important in India. The uncertainty expressed in the table results from the difficulty in valuing the reserves when the government coal industries actually operated at a loss in the period examined.9 In fact, the coal industry plays a rather minor role in the Indian economy, unlike oil in Saudi Arabia, so it would perhaps be best to minimize its role in these calculations.'I 9. The high estimate was made by valuing the underground coal at the price of Australian coal that is imported to India, less the cost of extraction. Because there are surely important differences in quality, this is doubtless too high. In fact, a number of countries operate coal industries at a loss to generate employment and to ameliorate regional economic depression. 10. Perhaps a clearer real-world example of an externality arising from collective ownership of mineral reserves is provided by the microstate of Nauru. The following passages are extracted from the United Nations (1989, p. 195) summary of their views and policies on population: "Government development objectives are strongly tied to the demographic situation since the entire economy is based on phosphate mining that, through its revenues, funds and supports all social services. Phosphate is the only resource the island has, and the supply is being rapidly depleted. The Government's response has been to make overseas investments in real estate, forming trust funds for the future welfare of its population." Notwith- standing the obvious negative externality to childbearing in this instance, the government would prefer more rapid population growth. 286 Population Growth, Externalities to Childbearing, and Fertility Policy Paralleling the value of mineral reserves in these countries is the possibly enormous value of Brazil's rain forest, listed under "government land." Valua- tion is necessarily very uncertain. The figure used estimates the present value of its sustainable yield of timber, fruit, and latex (see Peters, Gentry, and Men- delsohn 1989) and assumes that 25 percent of the rain forest is public land."1 Aside from the case of the Brazilian rain forest, government land, which is always difficult to value, is nowhere very important except perhaps in Kenya, where the value of tourism generated by the game parks is included. Government debt, which can more easily be repaid with a larger population to share the burden, ceteris paribus, leads to a significant positive externality (rela- tive to per capita GDP) in all countries other than Saudi Arabia. Surprisingly, debt is not strikingly large relative to per capita GDP in Mexico and Brazil in the mid-1980s; at 50 percent, it is comparable to debt in India and Kenya and smaller than debt in the United States, which, at 60 percent, dominates all other items of collective wealth.12 Only in Saudi Arabia does government capital lead to significant negative externalities, and our estimate there is based entirely on the assumption that a good portion of previous public oil revenues was invested in publicly owned capital. We count neither capital formed from issue of bonds or equity nor capital funded from taxes. Externalities arising from fisheries appear unimportant.13 If foreign aid is believed to be largely independent of the size of the population, then population growth reduces its per capita value. Taking a per capita present value of a stream of foreign aid at current levels over the adult lifetime of a child, this component is quite important for both Bangladesh and Kenya, where it comes to two or three times current per capita GDP. Public Goods Government budgets were inspected to select expenditures for items whose use was either not subject to congestion at all or was subject only partially to congestion. Not surprisingly, defense expenditures formed a very substantial 11. There are many problems in making such an estimate; among them is that if the entire Amazon rain forest were used in this way, the prices of the goods harvested might drop considerably. Of course, it would be impossible to use the rain forest in this way without enormous investments in social infrastruc- ture. Furthermore, it is not clear to what extent the ownership of the rain forest is public. 12. Some developed countries would have net credit in this category. Both internal and external debt is relevant here, because the obligation to repay is shared by all, whereas the value of any debt instruments held by the population is internalized in the fertility decision. 13. It is not clear how to evaluate the fisheries. Under current institutional arrangements, access is largely unregulated, and consequently rents are largely dissipated through overfishing. The entire value of output may then be attributed to inputs other than the fisheries themselves. Alternatively, we might value the fisheries according to the potential rents generated under optimal management. In principle this could exceed the value of the current catch, but in practice is likely to be only a fraction of it. In the calculations, we have used half the present value of a perpetual stream of harvests of value equal to that in the base year. This is an exceedingly rough guess at the value of rents for an optimally managed fishery. Lee and Miller 287 component of the total. Social infrastructural quasi-public goods, given a weight of one-half, were also quite important. Variation in these expenditures relative to per capita GDP was moderate, except for Saudi Arabia, which has very high relative public good expenditures funded through the oil revenues. 14 These items may appear irrelevant or peculiar, but in fact they are often of concern to scholars and policymakers. Governmental desire for a larger popula- tion to enhance military power goes back to ancient times and is well-known today. Edgeworth (1925, p. 20) wrote that a large population was desirable "for the sake of defense against or competition with foreign nations" and that "being must be secured before well-being" (quoted in Nerlove, Razin, and Sadka 1987, p. 82). A recent speech by Qaddafi (1987) strikes a very similar note, calling for rapid growth of the Arab population to a size of one billion for purposes of political and military power, while acknowledging the adverse economic conse- quences that would be likely to follow. One may well question the wisdom of heavy spending on the military in many countries, including the United States, but if we take governmental goals as given, such spending does create positive externalities to childbearing. It is also widely appreciated that denser population makes transportation and communication networks more affordable on a per capita basis, and in fact there are strong empirical associations between popula- tion density and social infrastructure capital such as roads (see Simon 1977 and Boserup 1981). Boserup (1981) has repeatedly stressed the importance of popu- lation size for major collective undertakings throughout history, including investment in and maintenance of social infrastructure such as irrigation sys- tems. Expenditure on research and development for agriculture and other areas is another public good expenditure with important potential for economic devel- opment. These uniformly positive externalities arising from public good expen- ditures should not be dismissed lightly. Proportional Taxes The incremental population member may contribute less in tax revenues than the average member while receiving the average amount of publicly provided goods and services. Evaluation of the resulting negative externality is in princi- ple straightforward: it is the survival-weighted present value of the tax rate times the difference between the average and the marginal product of labor, abstract- ing from variations by age that are considered below. The countries examined here have ratios of total tax revenues to GDP ranging from 11 percent (Bangladesh) to 33 percent (Brazil and the United States; the remaining figures are India, 17.9 percent; Saudi Arabia, 18.5 percent; Kenya, 22.9 percent; and Mexico, 19.9 percent). If production is constant returns to 14. It is correct to count the Saudi Arabian public goods as generating a positive externality, even though they are financed by oil revenues rather than taxes. The reason is that the full value of the oil has already been counted as generating a negative externality, although the portion used to fund public goods does not do so. 288 Population Growth, Externalities to Childbearing, and Fertility Policy scale, and the labor elasticity of output is 0.7, then the APL - MPL term would be 30 percent of per capita income, y. Therefore this component of externalities would have a negative value of from 0.03y to 0.1Oy per year (equals 0.11 to 0.33 times 0.3) or a negative present value of 1.5y to 5y. This is a substantial amount. More realistically, however, an incremental worker will contribute more than the short-run marginal product of labor over his or her lifetime, because the incremental worker also saves and accumulates nonhuman wealth, and has other, more complicated effects on the economy.1' For purposes of table 1, we attempted to bracket the possibilities by assuming that lifetime incomes of incremental population members lie in the range between 0.85 and 1.0 times the average life income. We then applied the tax rate for each country, calculated as the ratio of total taxes (central and local) to GDP. Obviously these evaluations are very approximate. Intergenerational Transfers In developing countries, public expenditures on education and health are frequently mentioned as important negative externalities to childbearing, whereas in more developed countries, the positive role of children in supporting the parental generation in its retirement is stressed. Each of these views is qualitatively correct in its context, as table 1 indicates. Pension expenditures in most less developed countries are relatively low, in part because of the low proportion of the population at advanced ages and in part because expenditures on public pensions per elderly person are in any event minimal, even relative to per capita GDP. Brazil is a stunning exception, as the table shows. Brazil and a few other Latin American countries have strong public pension programs. The average elderly person in Brazil receives about fifteen times as large an annual transfer as the average child, a ratio many times greater than prevails even in the developed countries of the world. Once we take into account that costly children will grow to pay taxes them- selves one day, the magnitude of the negative externality per birth in developing countries appears to be small, on the order of 10 to 25 percent of per capita GDP. (However, if the incremental child grows up to pay taxes only on the MPL, the effect of these transfer programs is far greater, by a factor of two to five. This effect has already been taken into account in the preceding section.) This is far less than many other sources of externalities and would not appear to justify the emphasis given to the costs of such public sector transfers in forming fertility policy. 15. If incremental members of the population had lifetime incomes substantially lower than the average, then population growth would lead to substantially lower per capita income, other things being equal. However, a number of recent reviews of the literature on economic consequences of population growth find the evidence on this point mixed and inconclusive in many respects (World Bank 1984; National Research Council 1986; and Kelley 1988). There may be more adverse consequences in some contexts, such as Bangladesh, than in others, such as the United States. Lee and Miller 289 Why are these estimates so low? There are two leading reasons. First, the calculations assume that the base population is homogeneous with respect to wealth and tastes, so that every child is assumed to pay the same taxes as an adult. But some children, who are born to poor families and who grow up poor, may receive average public sector transfers while paying below average taxes as adults (although the fact that the poorer population is generally in the rural areas, where they both receive poorer educational and health services and pay lower taxes, reduces the likely error arising from heterogeneity). Second, the calculations assume that the real rate of discount equals the stable population growth rate, which in these countries ranges from about 2 to 4 percent annually. Perhaps this discount rate is too low; with a discount rate of 10 percent, perhaps the earlier expenditures on education and health for children would overwhelm the later tax payments.16 Aside from these two qualifications, it appears that externalities arising from the age distribution of public expenditures on health, education, and pensions in the developing countries here examined are relatively modest, generally amount- ing to no more than a fifth of per capita GDP. When the age distribution of demand for other governmental services is considered in relation to the age distribution of payment of taxes, a negative externality of roughly equal size is found. The total negative externality arising from the age distribution of interac- tions with the public sector is therefore between 15 and 40 percent of per capita GDP. For the United States, and most likely for other industrial economies, the heavy public involvement in provision of pensions generates a far more substan- tial positive externality, equal to 150 percent of the U.S. per capita GDP. As developing-country populations age, and as their public sectors accept greater responsibility for old age support, the small negative externalities from inter- generational transfers may become negligible and then turn positive. In Brazil and Mexico, this process has already advanced substantially. The age pattern of their transfers is closer to the industrial countries than to the other developing countries considered here, although the level of transfers is still relatively low by the standards of the industrial nations. Net Externalities to Childbearing Summing the subtotals of externalities arising from collective wealth, public goods, proportional taxes, and intergenerational transfers, we find an estimate of the net externality to childbearing. The bottom row of the table shows the ratio of this value to the level of per capita GDP for each country. For the 16. For India, where the stable population growth rate would be about 2 percent per year, the real rate of return on long-term Indian government bonds issued over the past decade was almost exactly 2 percent. We have not checked the comparable number for the other countries. Regarding the discount rates, as noted elsewhere, higher discount rates largely reflect inflation and productivity growth, and both of these generate a corresponding rate of increase in costs and taxes and therefore leave the result of the calculation unchanged. Therefore it is unlikely that higher discount rates, if handled appropriately, would lead to very different results. 290 Population Growth, Externalities to Childbearing, and Fertility Policy developing countries, these ratios are either close to zero or negative. (The ranges given in the table derive from specific items and are not intended to indicate the overall degree of uncertainty, which is far greater.) It is particularly striking that the ratios for Kenya and Bangladesh-both widely viewed as hav- ing serious population problems-are close to zero. It appears that only coun- tries with high values of mineral reserves or other natural resources per head (Saudi Arabia, Brazil, and possibly India) have strongly negative childbearing externalities. If we take the plausible view that Indian coal reserves have a low value, then India would join Kenya, Bangladesh, and Mexico with net exter- nalities close to zero. In contrast to these developing countries, the United States has positive externalities arising from three sources: substantial national debt; heavy military expenditures; and major public transfers to the elderly. The effects of these may be considerably reduced by the distortions of proportional taxes. IV. EXTENSION TO A TWO-CLASS AGRARIAN SOCIETY As explained, the previous analysis was based on the assumption of a homo- geneous population, in which all members shared the same initial wealth and tastes. One might well wonder to what extent the results would hold up if we dropped this assumption. To explore these largely uncharted waters, we con- sider an agrarian society with a class of landless laborers and a class of land- owners who also contribute labor. The two classes have identical preferences. Each behaves according to the appropriately amended baseline model. For land- less labor, increased population reduces wages without any offsetting benefit from rising rents; perhaps, then, laborers could do better with lower fertility than under laissez-faire, and perhaps they could even do sufficiently better to compensate the landowning class for the difference between reduced rents and increased wages. Analysis of the model (see appendix 1) confirms that under laissez-faire, the class of laborers chooses fertility that is too high to be optimal, in the sense that as a class they could benefit from reduced fertility, thereby withholding some of the next generation's labor. This is true even after taking appropriate account of the satisfaction and economic services they expect to receive from their children. The class-specific pseudo-externality to a child may be very large-perhaps as much as ten times the annual agrarian wage. However, the high fertility of laborers generates a gain for the landowner- parents, by raising rents. In the neighborhood of the laissez-faire equilibrium, the utility gains of laborers from a fertility reduction would be exactly offset by the utility losses of landowners, so that no compensation or bribe would be possible (see appendix 1). The laissez-faire outcome is therefore Pareto-optimal, and a planner could not improve on it. For a society that cared about the equality of income distribution, a fertility control program targeted on the landless laborers would be attractive. Not only Leeand Miller 291 would it equalize the distribution of welfare in the current generation but for the next generation it would both equalize the distribution of income and raise per capita income (it would increase the welfare of the laborers primarily through raising their anticipations for their children's future, as it would reduce the anticipations of the landlords). Therefore, a case could be made for government intervention based on concern for income growth over time as well as for equita- ble income distribution within each generation. This case would not rest on externalities in the sense of market failure, however.17 V. RESERVATIONS At this point, let us consider some problems with the method and the data. We discuss three that we believe are particularly important. Estimates Reflect Transitory and Suboptimal Policy Decisions in Other Spheres First, the estimates are based heavily on current government allocation deci- sions and therefore reflect nonoptimal and transitory policies in other spheres. For example, in the mid-1980s, military spending in the United States was 6.6 percent of GDP. This public good outlay alone generates an externality of 3.3 times per capita GDP, or roughly half of the total net externality found. But now, with the relaxation of East-West tensions, military expenditures may be sub- stantially reduced, leading to a reduction in calculated externalities. Each child born has a permanent effect on the population size, ceteris paribus, and calcu- lated externalities should not depend on such changeable items. Nor is it clear that population policy should depend on a level of military expenditures that many would perceive as inappropriately high. Similar comments could be made about the effects of current public transfers to the elderly. These were far less in the 1960s, and calculated externalities surely would have looked very different then. In developing countries, such expenditures are likely to become increas- ingly important as both fertility and mortality decline and as development pro- ceeds. The calculations might be better if they were based either on some con- cept of optimal public expenditures (which would be operationally impossible) or perhaps on an appropriate cross-national average of public expenditure allocation. Such an approach would reduce the influence of military spending in the U.S. calculation and thereby reduce the positive externality. For other public goods, and for collective wealth, we believe it would make little difference. For population age structure (intergenerational transfers), it would probably neu- tralize the currently mildly negative externalities to fertility in all developing countries here examined. 17. In practice, there is only very weak evidence of an association of population growth rates with income distribution (Lam 1987) or of an association between population growth or density and per capita levels or growth rates of income (National Research Council 1986). 292 Population Growth, Externalities to Childbearing, and Fertility Policy Estimates Ignore Environmental Consequences A second serious problem is that potentially enormous environmental exter- nalities have been excluded from the collective wealth category, for want of appropriate data. Many valued aspects of the environment are highly congest- ible, such as the ability of airsheds to absorb emissions of carbon and other pollutants without serious degradation, manifested as acid rain, global warm- ing, and deterioration of the ozone layer. Proper consideration of such effects might swamp the externalities considered here. At the same time, it must be noted that most of these externalities are global, not national, in scope, and therefore best addressed through international rather than national channels. In addition, environmental concerns may well provide compelling new reasons for public involvement in private reproductive decisions, but they have not been the basis for most population policies to date, and therefore it is still instructive to examine the logic of the more familiar and conventional bases for intervention. Estimates Assume Population Homogeneity The third major difficulty with these estimates is that they are based on the assumption that the populations are homogeneous with respect to wealth and tastes. But if one subgroup has stronger preferences for children than another, might not its fertility inflict externalities on the other (see Greene 1985)? Or might not the fertility of poorer subgroups impose costs on the public sector that would not be fully repaid by the children as adults? These are difficult questions on which very little work has been done. We suggested in the previous section that in a society of landowners and landless laborers, the fertility of laborers would be at a socially efficient (that is, Pareto-optimal) level under laissez-faire, even though laborers as a class could raise their utility by collectively reducing their fertility. Thus a policy aimed at reducing fertility of the landless laborers would raise the utility of their current generation, redistribute income toward them in the next generation, and reduce the utility of landowners in the current and subsequent generation. For laborers as a class, the laissez-faire fertility level is not optimal, even taking their satisfaction from childbearing into account, while their laissez-faire fertility is not too high from a societal point of view, if distributional issues are ignored. It is our guess that aside from environmental collective wealth, the various difficulties and adjustments just discussed do not seriously alter the main results. We suspect that net negative externalities to childbearing are not sizable in most developing countries, except when there are unusually valuable mineral reserves, and that even then this source of externality will appear distant from the concerns of policymakers in most countries. We therefore do not believe that the reason why parents in many developing countries choose high fertility is because they are able to pass important net costs of childbearing onto society, although common sense tells us that fertility is often far too high. The reasons for high fertility, and the justifications for governmental intervention, typically must be sought elsewhere. Lee and Miller 293 VI. POLICY IMPLICATIONS Implications of Externalities The policy implication of an identified net externality is in principle straight- forward. As measured here, the externality is the difference between the parental perception of the net cost of a child and the social perception, which incorpo- rates the parents' but also adds any spillovers. The indicated policy is either to subsidize or to tax each birth so as to equate the parental perception of costs with the full social cost. Public funds required or raised in this manner are then taken from tax revenues or used to offset general taxes, as the case may be. A finding of significant negative externalities per child therefore would provide explicit support for a policy of financial disincentives for childbearing and would indeed identify a specific amount for the disincentive. As it happens, the research reported here did not find significant net externalities for some coun- tries of prime interest, such as Bangladesh and Kenya. If reproductive externalities do not appear to justify policies going beyond family planning in some important high-fertility settings, where does that leave fertility policy? Must these countries reconcile themselves to the rapid popula- tion growth that would continue even if all couples attained their current repro- ductive goals-goals that may be very high, as they are in Kenya, for example? Recent research disagrees about the scope for reducing fertility and population growth rates through elimination of unwanted fertility (see Bongaarts 1990; Westoff 1988), so such a conclusion might well cause concern.18 In fact, how- ever, such a conclusion would be premature, for several reasons. Effects of Less Costly Contraception Sociologists and family planning workers have long argued that the ability to control fertility can itself lead to a reduction in the desired family size. Becker (1981) showed that because of the peculiar interaction of the quantity and quality of children in the household budget constraint, a reduction in cost of contraception has a kind of multiplier effect on fertility, leading to a simul- taneous decline in desired family size and increase in desired investment per child. For this reason, family planning programs can be expected to do more than help couples achieve their current family size goals; they also can induce a reduction in those goals. Inaccurate Parental Expectations The estimates of externalities assumed that parents had complete information about the conditions to be faced by their grown children and about their proba- 18. Westoff (1988, p. 232), after analyzing data from recent Demographic and Health Surveys for Brazil, Peru, the Dominican Republic, and Liberia, concludes that "the overwhelming majority of women who want no more children or who want to postpone fertility, at least in the four countries discussed here, are behaving in a manner consistent with that goal." Bongaarts (1990) asserts that conventional estimates are biased and develops a new measure that suggests that on average 22 percent of births in forty-six developing-country populations are unwanted. 294 Population Growth, Externalities to Childbearing, and Fertility Policy bility of survival. But there may be systematic distortions in the information on which parents base their decisions. On the one hand, parents probably under- estimate the extent of secular economic growth in per capita incomes and there- fore underestimate their children's future incomes. This would lead them to reduce their fertility more than would be appropriate. On the other hand, par- ents probably underestimate the pace and certainty of mortality decline and therefore have higher fertility than necessary to achieve a high probability of the desired number of surviving adult children. This leads them to raise their fertility above the appropriate level. Likewise, parents may misjudge the future return to education and other investments in child quality, leading them to choose a greater number of children and invest less in each. The point here is that a policy of disseminating information on these matters may aid couples in choosing an individually appropriate level of fertility, and this level may well be lower than otherwise. Individual Concern for Community Outcomes The analysis of externalities was based on the assumption that individuals care only about their own well-being and that of their children. But it is plausible to believe that individuals care also about the distribution of income in their communities and about the improvement over time in average living conditions. Such considerations add a new dimension to the analysis and would surely lead to important negative externalities. Absence of Institutional Substitutes for Children An absence of negative externalities suggests that individual fertility motives are consistent with maximization of individually based social welfare. But their absence does not imply that current fertility levels are either individually or socially efficient in a broader sense. Fertility decisions are taken in a particular institutional context, and the calculation assesses their efficiency only relative to that context. In most high-fertility settings, surveys and anthropological studies indicate that children are valued for a number of instrumental purposes: to provide old age support for the parents; to provide insurance against risks to income, health, and physical security; and to provide labor services at home and in production (for example, see Bulatao 1979). Such services that children sup- ply are viewed as more important than direct psychic satisfactions, at least for higher-parity children. As developing markets and the public sector come to provide more cost-effective substitutes for many of these services, the individu- ally optimal level of fertility falls, and both individual and social welfare rise. High fertility is not absolutely optimal; rather, it is optimal in a suboptimal institutional context. VII. CONCLUSIONS Evidence on adverse macroeconomic consequences of population growth is weak and mixed, but, in any event, it could not in itself provide grounds for Leeand Miller 295 governmental intervention in reproductive decisionmaking. After all, if parents prefer to have more children at the cost of lower per capita household income, why should governments dispute that decision? Of course, if couples' fertility exceeds their desired family size, an excellent case can be made for governmental support for family planning programs to enable couples better to regulate their fertility. But policies designed to alter a couple's family size choice, or to coerce a certain outcome, are a different matter. From the point of view of welfare theory, justification requires that societal goals differ from individual ones, that individuals are poorly informed, or that the setting deviates in significant ways from a competitive market economy. Here we have considered one important deviation from full competitive markets-external costs and benefits of child- bearing. If consequences of childbearing are neither borne entirely by the parents nor pass through markets but rather spill over to society at large, then society has a legitimate interest in influencing childbearing decisions. Belief that such childbearing externalities exist and are negative in most developing countries is the most common justification for policies going beyond family planning. We have identified four broad categories of externality: dilution of the per capita value of collective wealth; dilution of costs of collective projects with public good aspects; incentive reduction due to proportional taxes; and the effect of population age distribution on the tax rate necessary to support public sector activities such as health, education, pensions, social infrastructure, and other services. We attempted a rough evaluation of many kinds of externalities in each of these four categories for a variety of countries. For some of the countries widely viewed as having serious population problems, the net total of these quantifiable externalities was close to zero. For others, the value of collec- tively held mineral rights dominated the calculation, leading to a large negative externality, but one that may seem unconvincing to many as a basis for fertility policy. Countries in which most agricultural land is collectively owned, such as China, have an additional large source of negative childbearing externalities. For the United States, there is evidence of positive externalities to childbearing, a result that may prove typical for industrial nations when environmental costs are excluded. We conclude that externalities to childbearing, although apparently some- what negative in most developing nations, do not typically provide a strong rationale for fertility policies going beyond pure family planning (the point estimates of the negative externalities are often large enough to warrant inter- ventionist policies. The difficulty lies in the wide band of uncertainty surround- ing them). This finding, which might well be altered if environmental problems were incorporated in the analysis, is tentative for reasons discussed earlier. In any event, a policy of vigorous family planning, combined with dissemination of information about mortality declines and the gains from parental investment in education, would be entirely consistent with this finding. Such policies in them- selves might lead to reductions in desired family size, particularly by promoting investment in child quality. We further suggest that the apparent absence of sizable negative externalities in many developing countries does not mean that 296 Population Growth, Externalities to Childbearing, and Fertility Policy high fertility is socially or individually desirable in any general sense, because it is rational only in settings where superior institutional substitutes for many of the services of children do not exist. Efforts to provide alternative sources of risk spreading, provision for old-age support, physical security, and health care might well lead to fertility declines and might raise both individual and social welfare. At the same time, coercive measures to reduce fertility without such institutions in place might reduce both individual and social welfare. Stronger fertility policies might be justified on grounds not considered here: society may care more for the welfare of future generations or for that of some current family members than do family decisionmakers; or society may care about the distribution of income rather than just Pareto-efficiency. The main finding here, however, is that given the current state of research, externalities to childbearing do not themselves provide a convincing rationale for fertility poli- cies involving financial incentives or various forms of coercion. The analysis here-restricted to the case of pure or technical externalities to childbearing- does not suggest that in most developing countries fertility policies should go beyond assisting well-informed parents to attain the family size goals they choose. APPENDIX 1. THE MODEL, WITH PUBLIC GOODS, COLLECTIVE WEALTH, AND PROPORTIONAL TAXATION Baseline Case Following Nerlove, Razin, and Sadka (1987), consider a population of N couples homogeneous with respect to tastes and wealth. Utility depends on a family's current consumption, c1, their number of children, n, and the consump- tion they anticipate for their children as adults in the second period, c2. Parental satisfaction arises from the average consumption by their children as adults, not from the total. Each family owns a farm yielding output that depends on labor inputs according to the homogeneous production function f(.), with land sup- pressed as an argument. They may leave a positive or negative bequest of b to the next generation. Storage of b is possible without gain or loss. Assume with Nerlove, Razin, and Sadka (1987) that all families, besides possessing identical tastes and endowments, also have perfect foresight about the fertility of other families, so that second-period wages, w2, and rents, r2, are known to first- period parents. The individual couple seeks to maximize its utility as follows: (A-1) max U(1, C2, n) (over c1, c2, n) Subject to: w1 + r1 - b = cl nA(w2) + A(r2) + b = nA(C2) A(w2) = W2 A(r2) = r2 A(C2) = C2 Lee and Miller 297 The terms wi and ri refer to wages and rents (to the fixed family holding) in each period, and A(.) denotes the second-period value anticipated in the first period. The last three equations express the assumption of perfect foresight. The five constraints can be combined to get (A-2) w1 + r1 + nw2 + r2 = c1 + nc2 The representative couple's first-order conditions are (A-3) Un = c2-w2 and U,2 =nU,. U" The assumption of perfect foresight assures that the wages and rents expected to prevail in the second period will indeed do so. An omniscient planner recognizes that second-period wages and rents depend on the choice of first-period fertility, n, so that (A-4) w2 =J'(n) and r2 =f (n) - nf'(n) Otherwise, the planner's problem is identical to the individual's. The planner's first-order condition for fertility differs from the individual's by a term: U, (dr2/ dn + ndw2/dn). But the term in parentheses is readily shown to equal (A-5) [-nf'(n) + nf"(n)] = 0 Therefore the form of the planner's first-order condition for fertility is no differ- ent from the individual's. The other first-order condition is readily seen to be the same as well. Therefore the planner, in seeking to maximize the utility of the first-generation parents, would make exactly the same choice of fertility as they, and no externality occurs. Public Goods and Collective Wealth with a Household Tax Assume the parental utility function now additionally incorporates satisfac- tion from consumption of a public good in period 1, P1, and from their chil- dren's consumption of a public good in period 2, P2. In each period, the govern- ment independently chooses a level of the public good (which may or may not be optimal) and levies an equal tax, ti, on all households to pay for it. (The assumption that the public good is financed by a household tax rather than a dynastic tax is critical, because Nerlove, Razin, and Sadka showed that with a dynastic tax no externality occurs.) In addition to revenues from taxes, the government can sell the publicly held national wealth, which has value R. The government has a balanced budget, and the net tax may be positive or negative. Note that environmental collective wealth under current institutional arrange- ments is not owned by the government and cannot be sold; furthermore, it can be enjoyed and used by all, regardless of whether they benefit from tax reductions. The problem for the representative couple is as follows: (A-6) max U(cl, C2, n, P1, P2) (over cl, c2, n; Pis are given) 298 Population Growth, Externalities to Childbearing, and Fertility Policy Subject to: f(l) + f(n) = cl + n 2 + t1 + nt2 The couple's first-order condition for fertility is (A-7) = C2 + t2 -f'(n) The planner recognizes that a larger population can share the potential reve- nues from selling the national wealth, R, and likewise shares the costs of provid- ing P2: (A-8) t = (PN-R) Optimizing with this additional constraint, we get the planner's first-order condition Un (A-9) u- = C2 The term t2 in a couple's first-order condition, equation A-7, vanishes in the planner's, equation A-9, because the planner sees that no matter how many children a couple has, their children's combined next period total tax obligation (for next period's public goods) will remain the same, provided all other couples have the same number of children. Consequently, an amount t2 = (P2 - R) I(nN) should be subtracted from the couple's perceived marginal cost of children (where n is taken equal to the planner's optimal value of children per couple) in order to internalize these two externalities. If P2 > R, there is a net positive externality. If P2 < R, there is a net negative externality. Proportional Faxes So far, we have assumed that all taxes are head taxes. However, more real- istically, taxes are levied at a rate t of income or consumption expenditure. Let the utility function be as before, including public goods, but now suppose that all first- and second-period income is taxed at the rates t1 and t2, respectively, so that the budget constraint becomes (A-10) (1 - t)Jf (1) + (1 - t2)f(n) = c1 + nc2 The parents' first-order condition for fertility under laissez-faire now becomes (A-11) U = c2 -(1-t2)f(n) TUh- The planner knows that the second-period tax will be set to raise revenues equal Lee and Miller 299 to P2 -R, so that (A-12) t2 (P2 -R) Nf (n) Optimizing with this additional constraint we get the planner's first-order condition: (A-13) Un = C2-f'(n) ucl The ractor of 1 - t2 in the private first-order condition vanishes in the planner's, because the planner sees that no matter how many children a couple has, their children's next period total tax obligation (for next period's public goods) will remain the same, provided all other couples have the same number of children. Consequently, an amount (A-14) f'(n)t2 = J'(n)(P2 - R) Nf (n) should be added to the couple's perceived marginal cost of children (where n is taken equal to the planner's optimal value) in order to internalize these two externalities. This is a negative externality per child. Comparison of this figure with the earlier result for a fixed head tax of (P2 - R)/(nN) (see the discussion following equation A-9) indicates that the positive externality in the case of a household tax must be reduced by [r21f (n)] [(P2 - R)I(nN)]. The first quantity in brackets is the proportional share of nonlabor inputs (land) in output; the second is the previous head tax. In other words, there is an additional negative externality to childbearing equal to this expression, to be applied when taxation is proportional. If R > P2, and R is expected to be used to offset future taxes, then this becomes a positive externality-or, put differently, the negative externality resulting from R is smaller by this amount when the tax is proportional rather than fixed. Now suppose that the government taxes the population proportionately to supply them with a private good, such as education or health, in per capita amount z, taken as given by the parents. With the usual utility function and budget constraint, their first-order condition becomes (A-1S) U = (2 - t2)f'(n) The planner, however, knows that z will have to be provided for each child, and also that the tax t2 must be set so as to raise sufficient revenue to do this, so that (A-16) t2f(n) = nz 300 Population Growth, Externalities to Childbearing, and Fertility Policy The planner's first-order condition is then (A-17) Un C2 + z (n) Consequently there is a negative externality per child equal to [t2 f'(n) - z]. Substituting for t2 from equation A-16 and r2 from equation A-4 gives this negative externality as [r2/f (n)]z (where r2/f (n) is the share of nonlabor inputs in output), or -E ,,,,z where E,,n is the elasticity of wages with respect to the numbers of workers. More generally, as explained in the paper, this negative externality will equal the difference between the average product of labor and the marginal product of labor, times the proportional tax rate. Note that the correction is the same for public and private goods, relative to the head tax or household tax. (No externality occurs in the case of a head tax for a private good supplied by the public sector.) Therefore, all that is necessary to make the correction is that the proportional tax rate be known, and the difference between the average and marginal product of an additional member of the population over the life cycle be known. Collective Landownership as a Special Case Suppose all land is collectively owned, and consequently alter the baseline model so that a couple expects each of its children to receive a second-period income equal to the population average income, regardless of the number of children born. The planner's first-order conditions are unchanged; for fertility: (A-18) un f(n) Under laissez-faire, the couple's first-order condition will be (A-19) U - Uc' where Y2 is average second-period income from nonparental sources, given by the implicit wage, f'(n), plus the per capita share of implicit rents on the collec- tively owned land. To make the first-order conditions equal in the neighborhood of the planner's chosen fertility, a term equal to the per capita share of implicit rent at the planner's second-period population size must be added to the laissez- faire first-order condition. If the initial population size is at equilibrium, then this simply equals the initial per capita value of rent; if the initial population is above the optimum, then the appropriate implicit rent share will be somewhat larger than that prevailing in the base period, and conversely. Heterogeneity: Landlords and Laborers Landowners and laborers are assumed to have identical utility functions as described above. Suppose there are M landlord couples in the first period, and the representative couple chooses fertility m. Similarly there are N laborer cou- Lee and Miller 301 ples with the representative couple choosing fertility n. Consumption levels in period i, for laborers and landlords, respectively, are c, and cm.. Landlords supply labor as well as land, and all face the same wages in each period. Both have accurate expectations of second-period outcomes. The budget constraints are (A-20) cm, = r1 + r2 + wI + m(w2 - Cm2) and cnl = w1 + n(W2 - C"2) where r2 refers to second-period rents on the family land as held in period 1. The relevant first-order conditions are readily shown to be (A-21) Ur = cm -w2 and Un = C'2-W2 U'~ Ul" Evidently the implicit price of children is higher for landowners, so long as c2 is a normal good (that is, landowners will choose higher consumption for their children, which makes them more expensive). The planner takes into account that both r2 and w2 depend on the size of the labor force in the second period, L2, equal to mM + nN. Clearly (A-22) div2 = N(dw2/dL2) and dW2 = M(dw2/dL2) dn ~~~~dm and similarly for dr2/dn or dr2/dm. For a constant returns to scale production function, we must also have that the change in total rents is equal to (A-23) M(dr2/dL) = -L(dw2/dL) Is the laissez-faire outcome Pareto-optimal? Consider the effect of a change in the fertility of laborers, n, in the neighborhood of the laissez-faire equilibrium. Evidently higher fertility would raise rents and depress wages, benefiting land- lords and harming workers. But would it be possible for either group to bribe or compensate the other, and still come out ahead? Calculation shows that it would not; the money value of the utility changes, aggregated for each class, are equal and of opposite sign. (Use the budget constraint to substitute for cn and c n2 in the utility function for each class, differentiate with respect to n, use first-order conditions to simplify, divide by the marginal utility of consumption in the first period to express utility changes in terms of the consumption good, and multiply by the number of couples in each class, to establish this result. The result is probably transparent to the clever.) Therefore the laissez-faire outcome is Pareto-optimal, and the planner could do no better. Nonetheless, there is a considerable benefit to workers as a class from restrict- ing their fertility. At the laissez-faire equilibrium, if all change their fertility together, then (A-24) (dUn/dn)/(dU, /d,~,) = nN(dw2/dL2) = (nN/L2)w2E,W,L where EWL is the elasticity of wages with respect to labor, which equals the share of nonlabor inputs in production, or perhaps 0.4 or so. If there are four 302 Population Growth, Externalities to Childbearing, and Fertility Policy times as many laborers as landowners, then this expression has a value of roughly one third of the wage level. Thus the gain to the workers as a class from reducing their fertility by one child would be equal to one third of the lifetime wage earnings of a worker, or perhaps ten times annual wage earnings. APPENDIX 2. PARTIAL LISTING OF DATA SOURCES FOR TABLE 1. General Sources Coale, Ansley J., and Paul Demeny. 1983. Regional Model Lifetables and Stable Populations. New York: Academic Press. International Labour Office. 1985. The Cost of Social Security. Geneva. International Monetary Fund. 1989. Government Finance Statistics Yearbook. Washington, D.C. United Nations. 1988. Demographic Yearbook, 1986. New York. U.S. Department of the Interior, Bureau of Mines. 1989. Minerals Yearbook, 1987. Washington, D.C.: U.S. Government Printing Office. Country-Specific Sources Bangiadesh Bureau of Statistics. 1985. 1982 Statistical Yearbook of Bangladesh. Dhaka; and various other publications of the Bangladesh Bureau of Statistics. Boskin, Michael J., Marc S. Robinson, Terrance O'Reilly, and Praveen Kumar. 1985. "New Estimates of the Value of Federal Mineral Rights and Land." American Economic Review 75, no. 5 (December): 923-36. Chopra, K., G. K. Kadekodi, and M. N. Murty. 1988. Participatory Develop- ment: An Approach to the Management of Common Property Resources. Institute of Economic Growth, monograph. India, Ministry of Planning, Department of Statistics. 1984. Statistical Abstract of India, 1984. Delhi; India Department of Statistics, and various other gov- ernment publications and surveys. Instituto Brasileiro de Geografia e Estatistica. 1985. Anuario Estatistico do Brasil, 1984. Rio de Janeiro; and various other statistical publications of this agency. Instituto Nacional de Estadistica, Geografia e Informatica. 1989. Agenda Estadistica, 1988, Mexico. Mexico City; and various other statistical publica- tions of this agency. Kenya, Central Bureau of Statistics. 1988. Statistical Abstract, 1987. Nairobi; and various other statistical publications of this agency. Peters, Charles M., Alwyn H. Gentry, and Robert 0. Mendelsohn. 1989. "Val- uation of the Amazonian Rain Forest." Nature 339 (June 29): 655-56. Saudi Arabia, Ministry of Finance and National Central Department of Statis- tics. 1988. Statistical Yearbook, 1987. Riyadh; and various other statistical publications of this agency. U.S. Bureau of the Census. 1988. StatisticalAbstract of the United States, 1988. Lee and Miller 303 Washington, D.C.; and various other statistical publications of the Bureau of the Census. REFERENCES Becker, Gary. 1981. A Treatise on the Family. Cambridge, Mass.: Harvard University Press. Blandy, R. 1974. "The Welfare Analysis of Fertility Reduction." Economic Journal 84: 109-29. Bongaarts, John. 1990. "The Measurement of Wanted Fertility." Working Paper no. 10 of the Population Council's Research Division. New York: Population Council. Boserup, Ester. 1981. Population and Technological Change: A Study of Long-Term Trends. Chicago: University of Chicago Press. Bulatao, Rodolfo. 1979. "Further Evidence of the Transition in the Value of Children." Papers of the East-West Population Institute no. 60-B, November. Davis, Kingsley. 1967. "Population Policy: Will Current Programs Succeed?" Science 158: 730-39. Edgeworth, F. Y. 1925. Papers Related to Political Economy, Volume 3. London: Macmillan. Greene, Geoffrey. 1985. "An Analysis of Intergenerational Externalities: Parental Altru- ism, Heterogeneous Family Size Preferences, and Institutions Governing Factor Income Distributions." Background Notes for the Population Growth and Economic Development Study, National Research Council Committee on Population (May). Washington, D.C. Hardin, Garrett. 1968. "The Tragedy of the Commons." Science 162: 1243-48. Kelley, Allan. 1988. "Economic Consequences of Population Change in the Third World." Journal of Economic Literature 26, no. 4 (December): 1685-1728. Lam, David. 1987. "Distribution Issues in the Relationship Between Population Growth and Economic Development." In D. Gale Johnson and Ronald D. Lee, eds., Popula- tion Growth and Economic Development: Issues and Evidence. Madison: University of Wisconsin Press. Lee, Ronald D. Forthcoming, a. "Evaluating Externalities to Childbearing in Developing Countries: The Case of India." In collected papers from the United Nations-INED Conference on Population and Development, New York, August 1988. -. Forthcoming, b. "The Second Tragedy of the Commons." In special issue of Population and Development Review on Population and the Environment. -. 1990. "Population Policy and Externalities to Childbearing." In Samuel Preston, ed., World Population: Approaching the Year 2000. Special Issue of Annals of the American Academy of Political and Social Science 510 (July): 145-54. Miller, Merton H., and Charles W. Upton. 1985. "A Test of the Hotelling Valuation Principle." Journal of Political Economy 93, no. 1 (February): 1-25. National Research Council. 1986. Working Group on Population and Economic Devel- opment, Population Growth and Economic Development: Policy Questions. Washing- ton, D.C.: National Academy Press. Nerlove, Marc, Assaf Razin, and Efraim Sadka. 1987. Household and Economy: Wel- fare Economics of Endogenous Fertility. New York: Academic Press. 304 Population Growth, Externalities to Childbearing, and Fertility Policy Ng, Yew-Kwang. 1986. "The Welfare Economics of Population Control." Population and Development Review 12, no. 2 (June): 247-66. Peters, Charles M., Alwyn H. Gentry, and Robert 0. Mendelsohn. 1989. "Valuation of the Amazonian Rain Forest." Nature 339 (June 29): 655-56. Phelps, E. S. 1968. "Population Increase." Canadian Journal of Economics 1, no. 3 (August): 497-518. Pitchford, J. D. 1974. Population in Economic Growth. Amsterdam: North Holland. . 1985. "External Effects of Population Growth." Oxford Economic Papers 37, no. 2 (June): 264-81. Qaddafi, Muammar. 1987. "Muammar Qaddafi on Population and Power." Population and Development Review 13, no. 3 (September): 563-65. Simon, J. 1977. The Economics of Population Growth. Princeton, N.J.: Princeton Uni- versity Press. .1986. Theory of Population and Economic Growth. Oxford: Basil Blackwell. United Nations, Department of International Economic and Social Affairs. 1989. World Population Policies, Vol. Il: Population Studies, no. 102/Add. 1. New York. Westoff, Charles E. 1988. "Is the KAP-Gap Real?" Population and Development Review 14, no. 2 (June): 225-32. Willis, Robert. 1987. "Externalities and Population." In D. Gale Johnson and Ronald D. Lee, eds., Population Growth and Economic Development: Issues and Evidence. Madison: University of Wisconsin Press. World Bank. 1984. World Development Report 1984. New York: Oxford University Press. PRO CE E DIN GS OF THE W O RLD BANK ANN UAL CO NF ERE NC E ON DE VE LOPMENT ECONOMICS 1 990 COMMENT ON "POPULATION GROWTH, EXTERNALITIES TO CHILDBEARING, AND FERTILITY POLICY IN DEVELOPING COUNTRIES," BY LEE AND MILLER Martha Ainsworth The National Academy of Sciences' study, Population Growth and Economic Development (1986), concluded that "on balance . . . slower population growth would be beneficial to economic development for most developing countries." But the study also noted that quantifying the impact of rapid population growth is fraught with difficulties, because population growth and economic growth are linked in a very complex way. Hence, it made a plea for more research on, among other things, the "nature and extent of externalities to childbearing." The paper by Ronald Lee and Timothy Miller responds to this plea. The divergence between the net costs of children to parents and to society is often cited as a justification for policies to lower fertility, but there have been very few attempts to measure the externalities to childbearing. In this paper the authors estimated the magnitude of three childbearing externalities-those due to collectively owned natural resources, to public goods, and to intergenera- tional transfers. Externalities due to environmental degradation were not evalu- ated but may be important. The authors sensibly do not attempt to value the impact of population density on technological progress or on returns to scale. They calculated selected externalities for seven countries and found that the net total externality for most countries is only slightly negative; that negative net externalities are substantial only when there are important natural resources; and that public expenditures on health, education, and pensions lead only to small externalities in developing countries. I would like to underscore the point that this paper is exclusively about externalities to childbearing and their magnitude. Although population growth is mentioned in the title, the paper does not deal with the optimal rate of population growth or the impact of rapid population growth on development. As the authors note, the presence of externalities is only one of several possible reasons for social planners to want to affect fertility. They may care about raising per capita incomes and income distribution, for example. Even if there are no childbearing externalities, very rapid population growth may exacerbate Martha Ainsworth is an economist in the Population, Health, and Nutrition Division of the World Bank's Africa Technical Department. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 305 306 Comment problems caused by other market imperfections, leading to adoption of policies to lower fertility. The discussion of the nature of externalities and the modeling are very clear. The model is a simple one, designed to demonstrate certain key points. It has some strong and unrealistic assumptions, but the sensitivity of the model to some of the assumptions is dealt with. My comments, then, address two issues: do the results make sense? And, how does this paper inform population policy? On the first issue-whether the results make sense-I would suggest that this paper is an honest and careful attempt to quantify a phenomenon that cannot be satisfactorily quantified. I have never tried to quantify childbearing externalities myself, so I had an open mind about the feasibility of these calculations when I began reading the paper. The characteristics of the externalities to childbearing are well brought out and discussed in the context of a simple two-generation model. Yet, confronted with the numerous assumptions and qualifications to all of these estimates as well as the fact that many externalities simply could not be estimated at all, I had serious doubts about the numbers by the time I reached the conclusion. I fully expected the paper to conclude that these externalities simply cannot be measured and that the numbers presented are of limited use in informing policy. Instead, the authors concluded that "externalities to childbear- ing do not typically provide a strong rationale for fertility policies going beyond family planning." Do the results make sense? The results on net externalities in table 1 lead us to completely counterintuitive policy prescriptions. Theory states that countries with negative net externalities to childbearing should raise taxes or lower sub- sidies on children, whereas those with positive externalities should lower taxes or raise subsidies on children. Consider the results for two countries, Saudi Arabia and Kenya. According to the calculations in table 1, Saudi Arabia's mineral wealth leads to very high negative externalities, so it should be promot- ing policies to lower fertility. Yet we know that Saudi Arabia is among the wealthiest developing countries, rich in resources and capital, but that it faces a labor shortage, prompting the influx of migrant workers. Thus, even though the population growth rate is greater than 4 percent per year, and the total fertility rate (the number of children a woman would have in her lifetime if she were to bear children at the current age-specific fertility rate) is more than seven, in actual fact, authorities would like to raise fertility. Now consider Kenya, which has very low per capita income of only $300 annually, a very high population growth rate of more than 4 percent, stagnant growth in per capita income, and a population-doubling time of only seventeen years. The pressure on arable land in Kenya is severe. Relative to Saudi Arabia, Kenya is rich in labor but poor in resources. These factors have led to an increase in the commitment of leaders to lower fertility in recent years. Nev- ertheless, the calculations in table 1 imply that Kenya should encourage fertility. Even if the paper had found more intuitively satisfying results, its methodol- ogy still presents many aspects that make one uncomfortable. First, most of the Ainsworth 307 individual externalities are dependent on current or past government spending decisions, such as government debt, foreign aid, public goods spending (in particular, defense spending), and intergenerational transfers. In this sense, the level of externality is being endogenously determined by the social planner. Further, the level of spending on these items is not necessarily the optimal level. Poor countries have small negative spending externalities on social services sim- ply because they can afford to spend very little per capita on social services. The result, according to table 1, is that they have less to lose from high fertility than a wealthy country; yet high fertility will prevent them from achieving their objec- tive of raising per capita spending. A second objection, related to the first, is that the levels of externalities in table 1 also reflect a host of existing subsidies, taxes, and other policies else- where in the economy that may or may not be optimal. These distortions are not taken into account in the calculations. Can we be sure that the numbers in table 1 are attributable purely to childbearing externalities and not to other underly- ing market distortions? This issue has an important bearing on the conclusions of the paper. The authors note that "childbearing externalities are assessed in the context of an institutional environment that may itself be highly inefficient." In such circumstances, the absence of externalities no longer guarantees a socially desirable level of fertility. Bearing this in mind, I don't think that we can say that Bangladesh has a socially desirable level of fertility, even though this paper implies that it has zero net externalities. Third, the concept of externalities here is static. There is no dynamic process in which the rate of population growth affects externalities. As for the specifics, I have a few questions about the way the calculations were made: • How plausible is the assumption that the population growth rate equals the discount rate for the countries other than India? - How were these resources valued-by their financial costs or their shadow prices? * The interpretation of the signs on the net externalities is clear; how should the magnitudes be interpreted? (Is this the level of the incentive that should be charged to each household so that fertility decisions correspond to socially optimal outcomes?) In the paper it is suggested that these net externalities are small. Compared with what? How large are other externalities? A negative externality equivalent to 15 to 40 percent of per capita income sounds large to me. On the second major issue I raised-whether the paper informs population policy-I would ask the following question: if the levels of net externalities are so small, fluctuate from year to year, and are highly dependent on planners' decisions and reflect other market distortions, how useful can they be in inform- ing population policy? For all of the reasons cited above, these results should not be used to guide 308 Comment population policy. Actual population policies are often at variance with the' results in table 1, suggesting that (even if the numbers are correct) pure exter- nalities are not the most important criteria guiding fertility policy. In fact, the type of externalities examined here are not the most widely cited reasons for fertility policy. Government planners are usually concerned with raising per capita income faster than otherwise would be the case. Welfare economists are concerned that rapid population growth will result in greater income inequality; that high fertility in this generation will depress wages in the next; and that the next generation will have higher per capita incomes if fertility is limited. Rapid population growth lowers the returns to labor and raises the returns to capital and fixed resources, worsening income distribution. Both of these concerns-the impact of rapid population growth on per capita income growth and on the distribution of income-arise from pecuniary exter- nalities to population growth. The market adjusts, and a Pareto-efficient result prevails. However, there will be clear winners and losers, and this is the concern of policymakers. I am not convinced that one can estimate the magnitude of externalities, but if one were to try, it would be far more useful for population policy to assess the magnitude of the two pecuniary externalities almost always cited for fertility policy. Those are my main observations. The authors raise many of the same criti- cisms in their paper. The difference is that we disagree on the conclusions and policy relevance. As a parting observation, I'd like to note that the paper does seem to equate "beyond family planning" policies with incentives, disincentives, and coercion. Yet there is an important category of "beyond family planning" policies that act to change peoples' tastes for large families while at the same time promoting development. I am speaking of policies to educate girls, to implement reforms in land rights, and to improve credit for the poor, to name three. They do not fit well into the theory of externalities, which takes tastes as given, but they nevertheless promote lower fertility and are ethically more acceptable than disincentives. REFERENCE National Academy of Sciences. 1986. Population Growth and Economic Development. Washington, D.C.: National Academy Press. PROCE E D INGS OF TH E WO R LD BANK ANN U AL CONFE RE NCE ON D E VE L O PME NT E C ON OM ICS 1 9 9 0 Incentives for Small Families: Concepts and Issues Kenneth M. Chomitz and Nancy Birdsall This paper examines the benefits and costs of client-targeted family planning incentives-public policies providing specified rewards or penalties for specified fertility-related behavior. Two rationales for incentives are discussed: first, where mar- kets for contraceptives or contraceptive information fail, incentives can increase fami- lies' welfare by reducing barriers to the use of contraception; and, second, where childbearing imposes external costs not borne by parents, incentives can align private and social costs, improving social welfare. These distinct rationales require distinct types of incentives. Incentive costs include economic costs of program administration; the fiscal (but not economic) cost of the incentive payments; welfare losses incurred if ill-informed or myopic individuals make choices they later regret. Little has been done to assess the costs and benefits of existing or proposed incentive schemes. A framework for doing so is presented. Preliminary evidence from an experiment in the state of Tamil Nadu in India suggests that incentives for short-term trial of contraceptives or for acquiring information may offer substantial benefits at low cost. The use of incentives to encourage small family size has evoked strong emo- tional reactions. In this paper we take an analytic view of this value-laden subject. In doing so, we use the simple analytics of welfare economics, in which public policy must be justified as enhancing overall social welfare; and in which trade-offs between the welfare of different individuals, and between individuals and society as a whole, are carefully evaluated. We define incentives as public schemes or policies that provide specified rewards (or penalties; that is, disincentives) for specified fertility-related behav- ior. Although most family planning programs do not employ such policies, incentives have been used on a massive scale by many governments over the last Kenneth M. Chomitz is assistant professor of economics at Boston University. Nancy Birdsall is chief of the Environment Division of the Latin America and the Caribbean Technical Department at the World Bank. The authors are grateful to Janice and Carl Stevens for valuable discussions; to Jere R. Behrman for useful comments on an earlier draft; to seminar participants at Harvard University and the World Bank; to Christine Evans for research assistance; and to Professor R. Subramanian, head of the Department of Applied Research, and his colleagues at the Gandhigram Rural Institute for making available data from their Ammanpettai Survey. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 309 310 Incentivesfor Small Families three decades as a means to reduce fertility. I The best known, most widely used, and most controversial incentives are lump-sum payments to persons accepting sterilization, used in Bangladesh, India, and Sri Lanka. In India, 1984 federal- level compensation payments to clients and canvassers (principally for steriliza- tion acceptors) were 920 million rupees (more than US$75 million), constituting 20 percent of the family welfare budget (Rao, Khan, and Prasad 1986). These payments were supplemented by substantial state, local, and private-sector expenditures. Other incentives have included small monthly payments for non- pregnancy, tested for a period in India; and taxes (disincentives) on children or other penalties for large family size, used in Singapore and China in combination with such positive incentives as extra educational and maternity benefits.2 Concern with incentives arises for at least two reasons. First, the most wide- spread use of incentives has been tied to sterilization-an irreversible procedure-in societies where average incomes are low. Do such payments induce behavior that is not really voluntary, entrapping the poor into decisions they cannot reverse and may later regret? Second, incentives link financial rewards to a fundamental or inherent right to bear children. Do they pose in some societies a choice whose very framing violates cultural norms? We return to these issues below. Although we cannot and do not resolve them directly in this paper, our analytic approach at least partly takes these issues into account. Our theme in this paper is that the size, type, and design of particular incen- tives should be a function of the rationale for the incentive in the first place-for example, the market failure the incentive is meant to address. The costs and benefits of particular incentives are best analyzed given the objective, and given the alternative nonincentive policies that might also address that objective. The approach we use leads us to consider incentives different from those most com- monly used; we conclude that certain of these rarely used incentives are simple, of low cost, and pose virtually no ethical dilemma nor risk of welfare loss. The cost-effectiveness of these incentives is probably high, but not really known, given that few schemes of this type exist and that fewer still have been analyzed systematically. Section I briefly reviews the overall justification for public policies of any sort to affect fertility and indicates the potential role of incentives as one form of public policy. Section II examines particular market failures and their implica- tions for incentive design. Section III discusses the costs of incentives, broadly construed, and presents a cost-benefit framework for assessing incentive schemes. Ethical issues in incentive design are also discussed. Section IV illus- trates the proposed analytic methodology in the context of an interesting new incentive scheme. Section V summarizes policy conclusions. 1. Incentives to increase fertility have been widespread in Europe for an even longer period. Incentives to increase fertility include, for example, generous family allowances. 2. In South Asia, fieldworkers have received extra compensation based on the number of clients they refer to family planning clinics for sterilizations and in some cases IUD insertions. We do not discuss these third-party incentives in this paper; their analysis would present entirely different conceptual issues. Chomitz and Birdsall 311 I. FERTILITY POI.ICY AND MARKET FAILURES In developing countries, public policies to affect fertility choices of individuals are generally justified on two grounds. One is the presence of externalities associated with childbearing, so that people have higher fertility than is socially optimal, because they do not incur the full costs of children themselves.3 The second is failure in the markets for contraceptive information and services, so that people have higher fertility than they would choose were they better- informed about contraceptive options or better provided with services (see, for example, National Research Council 1986; and Birdsall 1989). The single most common public intervention to affect fertility in developing countries is the subsidization of voluntary family planning services.4 Such sub- sidies are widely viewed as simultaneously addressing both types of market failures. First, to the extent that lower costs of contraception lead couples to voluntarily reduce their own fertility, aggregate fertility falls, and any negative externalities associated with childbearing are reduced. Second, by filling in for missing private markets, public family planning programs reduce the financial, health, and psychic costs of contraception. This directly increases couples' wel- fare by providing them with a wider range of choice about family size and expenditure.5 Thus, to the extent that there are any negative externalities to childbearing, contraceptive information and services reduce those externalities without impos- ing any costs on individuals. In fact, they yield direct benefits for individuals. Increases in education, reductions in infant mortality, improvements in the status of women, and other changes in the environment that affect individuals' demand for children are also widely discussed as interventions that governments may use in the presence of externalities to alter individual fertility behavior (see World Bank 1984). For example, women's education is hypothesized to reduce fertility by raising the opportunity cost of children to women. Like public sup- 3. Externalities imply a cost or benefit that markets do not reflect. For instance, your decision to burn gasoline may not take account of the cost imposed on everyone else by the resultant smog. But if that cost can be quantified and then directly imposed on you via a gasoline tax, your consumption will be reduced to the socially optimal level. 4. These may take the form of direct provision by the public sector of contraception, often at zero price to consumers, or of public subsidies to nonprofit and for-profit providers. Subsidizing private sector provision is referred to as contraceptive social marketing (see Behrman 1989). 5. Put another way, with lower costs of using contraception, individuals are better off, either because they avoid so-called unwanted fertility or because they achieve their desired fertility at lower cost. Becker and Lewis (1973) and Easterlin and Crimmins (1985) also point out that lower contraceptive costs may actually change the underlying demand for children, reducing the level of fertility that a couple views as optimal. In the Becker and Lewis model, a change in the price of contraception reduces the optimal number of children by affecting the trade-off between quality and quantity in the budget constraint; the higher price of quantity implied by lower contraceptive costs reduces the relative price of quality of children, raising demand for quality relative to quantity. In the Easterlin and Crimmins model, the optimal number of children is itself a function of the disutility of using contraception, at least for couples whose potential fecundity exceeds their (initial) desired number of children. 312 Incentives for Small Families port for family planning, these environmental changes are viewed as benign policies because they reduce fertility at the same time as they provide direct benefits to individuals. Incentives to reduce fertility, like public subsidies for family planning, can be justified on the grounds that they correct for market failures in the provision of contraceptive information and services. Moreover, like policies to increase women's education, they can be justified on the grounds that they reduce the demand for children by raising the cost of children, correcting for social exter- nalities of high fertility. In the next section we discuss two broad categories of incentives corresponding to these two major classes of market failure. II. MARKET FAILURES AND INCENTIVE DESIGNS Table 1 summarizes our typology of incentives based on the market failure or other problem incentives can address; the implications of each market failure for the design of incentives; and the implications for policy alternatives other than incentives. Barriers to the Use of Contraception A number of market failures and related problems create barriers to the optimal use of contraception from an individual's or a couple's point of view. Table 1. Market Failures and Incentive Design Implication for Failure incentive design Alternatives to incentives Barriers to the use of contraception Poor information; high cost Small payments for limited Expand services; of information time; verification of fieldworkers; community- contraceptive use not based distribution; needed information, education, communication (IEC) High costs of early adoption Payments where prevalence Same as above low; verification not necessary Procrastination; myopia Payment tied to perceived Same as above; upgrade clinic adoption costs; payments services for trial adoption Lack of credit market Payment tied to adoption costs More widespread services to reduce financial costs to individuals Lack of insurance market Payment contingent on high Improve credit and insurance losses markets Gap between private and social costs and benefits Externalities Deferred payments Change environment Possibly larger payments (education, health, and Verification critical so on) More family planning Tax children-but see text Quotas-but see text Chomitz and Birdsall 313 Failure in the market for information. Because information is to some extent a public good, private providers are not likely to supply it in adequate amounts. This is particularly the case when information (for example, about the possi- bility of controlling fertility) cannot easily be tied to a specific marketable prod- uct. Thus, information about rhythm and withdrawal has no private market, nor does information about the pill in rural areas where there is no real market for private medical services. Yet the welfare gains to individuals from accurate information of this kind may exceed the costs of providing that information. Demand for information may also be suboptimal. The collection and evalua- tion of information is costly, particularly for less-educated consumers and for more complex products. Consumers may therefore rationally screen out infor- mation whose expected worth is less than the costs of evaluation. They will therefore sometimes fail to acquire information that is actually worthwhile. This problem of suboptirnal demand for information may be particularly severe in the context of family planning where the product is a complex set of ideas and procedures whose benefits are not immediately apparent but whose perceived risks may be high. In addition, much information-for example, regarding appropriate method and medical contraindications-is client-specific and may be too sensitive and complex for the mass media to convey. To penetrate this information screen, family planning programs sponsor domiciliary visits by fieldworkers as an adjunct to mass media advertising. But in some settings, it may be cheaper to pay for a potential client's attention-for example, to pay him or her to visit a clinic and attend an educational presentation-than to disseminate information via fieldworkers. This device, familiar to developed-country consumers of automobiles and real estate, has also been used with some success in the Gambia to teach mothers how to prepare and administer oral rehydration therapy. The Mass Media and Health Practices Project (see Stanford University and Applied Communication Technol- ogy 1985) used a lottery as a device to encourage women to learn the procedure, which was explained through radio broadcasts and a pictorial flier. In the lot- tery, villages were randomly selected each week to be the site of a health fair. At the fair, women who demonstrated knowledge of oral rehydration therapy were rewarded with a small prize (such as a bar of soap) and a chance at winning a large prize. Information incentives have fewer apparent disadvantages than others we discuss. They need not be tied to use of any particular method of contraception, nor even to use of contraception at all. Thus, they entail no monitoring costs. The payments need not be large. A program of information incentives can be limited to a short time in a particular community. We know of only one example of information incentives specifically for family planning (see section IV below). High costs of early adoption of contraception. Poor information about con- traception raises the costs of early adoption for particular individuals or couples. For the same reason, early adopters of contraception lower the costs of subse- 314 Incentives for Small Families quent adoption for others by providing reliable and specific information about the risks and benefits of contraception. Because early adopters do not take into account the benefits they provide to others, the adoption rate is slower than would be socially optimal. As has been and is the case with agricultural innovations, as long as there is some unevaluated chance of a disastrous outcome, individuals may be justifiably skeptical of official assurances of low risk and high reward. They may put much more credence in the experience of their neighbors. To the extent that family planning is regarded as particularly risky, the demonstration effect-once pioneers verify the innovation's efficacy and safety-could be particularly important. Similar dynamics might prevail where social pressure or conformity weigh heavily on the individual's freedom of choice. in many societies, social norms strongly shape the costs and benefits of individual behavior. It is easy to imagine a situation in which each family would find fertility limitation advantageous, were it not for the fear of violating social norms and incurring social sanctions (Crook 1978). In such a situation, coordinated action could change norms, Although initial innovators would encounter social resistance, we would expect a "tipping" effect (see Schelling 1978). Once a critical percentage of families had adopted family planning, the old social norm would be undercut, and the other families would follow suit.6 Early adopter incentives, like information incentives, have certain advan- tages, particularly if limited to reversible methods of contraception. Incentives for early adoption could be offered only to groups or in communities where contraceptive prevalence is nearly zero, and only to a predetermined proportion of a particular group or target community (for example, the first 10 percent to apply). As with information incentives, verification of actual contraceptive use is not, strictly speaking, necessary; it is the appearance or public acknowledgment of use that is critical.7 Finally, the incentive need not be monetary; where social pressure is a factor, a prestige-related award might be more appropriate.8 6. Some evidence for the importance of demonstration or social-pressure effects is provided by the Taichung experiment (Freedman and Takeshita 1969). Neighborhoods in Taichung were randomly assigned to experimental or control groups. Family planning fieldworkers visited households in the experimental group. Freedman and Takeshita found that, holding neighborhood status (experimental vs. control) constant, adoption rates were higher when adjacent neighborhoods were in the experimental group. It might be hypothesized that within-neighborhood demonstration effects are stronger. 7. To discourage purely opportunistic claimants, the award could be given in two installments: the first given immediately, the second after two years, contingent on no births in the interim. The restricted number of awards would facilitate the more complicated record-keeping and verification needed for a deferred award. 8. Incentive schemes in which the community as a whole receives payments linked to contraceptive prevalence can be interpreted as an attempt to reverse the polarity of social pressures for high fertility. For example, a community-based experiment in Thailand (Weeden and others 1986) linked grants to a village loan fund with the village's contraceptive prevalence level. Although the study reported that experimental villages experienced a greater rise in contraceptive prevalence and a greater decline in "unmet need," the Chomitzand Birdsall 315 Failure of credit and insurance markets. In many countries, significant pro- portions of women say that they would like to limit their fertility yet do not do so. This condition of "unmet need" for fertility limitation is met by one-quarter of married fecund women in Bangladesh, one-fifth in Nepal, and one-eighth in Egypt (see Boulier 1985).9 An additional, and large, proportion of women report unmet need for birth spacing. If the benefits to the couple of a smaller family-reduced expenditure on food, more time to devote to work or leisure, increased maternal attention per child, reduced health risks-substantially outweigh the costs of adoption of contracep- tion (and the pleasures associated with a larger but poorer family), why doesn't the couple itself finance the adoption costs? There are two possible reasons. First, many poor couples, lacking collateral, are not creditworthy. Second, cou- ples may be able to finance the average cost of adopting contraceptives but could not afford-and, in the absence of insurance markets, cannot insure against- the small possibility of complications entailing a long period of unemployment. In these conditions of credit or insurance market failure, subsidizing adoption costs might boost acceptors' welfare by far more than the subsidy. How large, in practice, are acceptance costs? They obviously vary widely, depending on the distance of potential clients to the nearest service point and depending on the type of contraception. Sterilization and insertion of an intra- uterine device (IUD) are the only ones likely to impose substantial costs in lost work; the pill could (rarely) involve medical complications. A 1984 study of acceptors of sterilization in Bangladesh (see Cleland and Mauldin 1987) indicated that for tubectomy and vasectomy acceptors, out-of- pocket transportation and food costs averaged almost twice the average daily wage for male workers. Sixteen percent of acceptors spent more than three times the average daily male wage. More than one-third of acceptors eventually faced additional medical complications, and 11 percent eventually spent a total of almost ten times the average daily male wage. Recuperation time averaged four days for men and ten for women, but was also subject to considerable variation. Some 8 percent of men and 37 percent of women needed more than fifteen days to recuperate. Thus, although mean acceptance costs are relatively low, there is a substantial risk of relatively high costs. Are these costs beyond the ability of acceptors to finance? In the survey of Bangladesh tubectomy acceptors, 42 percent said that they would have been unable to undergo the operation in the absence of the payment offered (of almost six times the average daily male wage; Cleland and Mauldin 1987). The sample was not representative, however, and it is possible that responses were authors conclude that the incentive scheme did not succeed because it directly galvanized community pressure. Instead, the increase in contraceptive prevalence was linked to enthusiastic promotional activity by the loan committee members, who were motivated by intervillage rivalry and by the prospects of administering a large loan fund. 9. Demographers use the term unmet need to signify would-be demand for contraception were the price of contraception (including psychic as well as monetary costs) zero (see Boulier 1985). 316 Incentives for Small Families biased by a desire to please the interviewer. A stronger test is provided by looking at behavioral data and focusing on the adoption of nonpermanent con- traceptives, where acceptance costs are relatively low. Evidence on the effect of distance to a health center (a proxy for acceptance costs) on contraception is provided by Entwisle and others 1984, based on a survey in rural Thailand.10 No effect of distance was found for the group aged 15-24 years old (the least likely to be using contraception), but proximity to a health center significantly boosted the odds of using contraception for both the age 25-34 and 35-44 groups. For the former, the odds of using contraception were 67 percent to 80 percent greater (for a modal woman, with low education, who desires no more children) when a health center was located within four kilometers of her resi- dence.11 This evidence suggests that financing problems are a constraint, despite relatively low costs. One-time payments for contraceptive adoption are often used for sterilization (Satia and Maru 1986; Thapa and others 1987), occasionally for the IUD, but infrequently for other methods. This may reflect a combination of more serious financing constraints for sterilization and the IUD, and practical difficulties in monitoring adoption of other methods. The size of the sterilization incentives has varied widely; it has been as high as one month's per capita income (in Sri Lanka in 1983).12 If financing problems are a constraint to adoption, a first and best policy response would be to supply the missing credit and insurance services. But if markets fail to provide these services, such services may also prove adminis- tratively difficult for government agencies to provide. The most common alter- native is greatly improved access to family planning services, to lower financial costs to potential users. In some settings, an incentive for contraceptive adop- tion, with the size of the incentive tied, ideally, to method-specific and individual-specific adoption costs, could be less costly. The insurance function is more difficult to fulfill, although it might be possible to tie extra payments, after the fact, to medical complications. 10. In this study distances are measured objectively, rather than by respondents' estimates; and the health centers predate the family planning program, reducing the possibility of a biased relationship due to the targeting of clinics at areas with low demand for family planning. The study controls for respon- dents' education and stated desire for no more children, providing a strong test of whether distance matters by minimizing the possibility that distance from a health station is a proxy for traditional attitudes rather than difficult access. 11. Desire for no more children, an explanatory variable, is actually endogenous; thus, the pure effects of distance are probably understated. Distant women are more likely to be traditional, so a distant woman who nonetheless reports wanting no more children is probably more motivated to use contracep- tion than her less-isolated counterpart. Hence, the fall-off of contraceptive use with distance may appear more gradual than it actually is. 12. In India (Satia and Maru 1986), payment levels designated as compensation for lost wages rose from RslO in 1964 to RslOO (US$9.00) in 1983-the latter approximately equivalent to ten to twelve days' wages. An additional Rs40 (men) or Rs7O (women) is designated as compensation for drugs, food, and transport, and many states offer supplementary amounts. In 1983, Sri Lanka offered RsSOO (US$27.50), approximately equivalent to one month's per capita income. Chomitzand Birdsall 317 Against the advantage of overcoming financing problems, these incentives have two major disadvantages. First, they may be difficult to target; that is, many of the claimants may not in fact face financing difficulties. Second, these payments may loom very large for the poorest, and may constitute entrapment. The problem of entrapment is greatest for sterilization, because the procedure is irreversible and because the net financial benefits of acceptance could be rela- tively large for individuals with low acceptance costs (these issues are discussed at greater length in section III). Procrastination or myopia. A frequently cited ethical problem associated with incentives is that the prospect of short-term gains may tempt myopic indi- viduals to act contrary to their long-term interests. The converse problem, rarely mentioned, may be more pervasive: myopic individuals may be deterred by short-term costs from actions that would yield them long-term benefits (in other words, it is their inaction rather than their action that induces regret). This latter category does not fall into any usual classification of market failure by economists. In effect, it represents a challenge to the conventional notion that actual behavior reflects underlying preferences (that is, that there is a well- defined utility function). Schelling (1984) and Loewenstein and Thaler (1989) discuss a related class of problems, which they find both pervasive and difficult to reconcile with received economic theory. In a number of circumstances, people act as if they had two selves with conflicting time preferences: one "self' focused on the moment, the other concerned with the long term. Some common examples include dieting; abstention from alcohol or tobacco; saving; and pro- crastination in general. Schelling points out that a great many people will volun- tarily arrange to put themselves in situations that compel them to behave in accord with the long-term set of interest. For example, people join forced-saving Christmas clubs, even though voluntary savings yield higher interest rates; they pay to be subjected to an enforced diet; they give their car keys to friends before drinking, with instructions to ignore drunken pleas for the keys' return. Along these lines, we may hypothesize that the immediate psychic costs of adopting contraception are for some people large enough to induce severe pro- crastination, despite long-term benefits of a smaller family or more evenly spaced births. This would particularly be the case if there is uncertainty about health effects, fear of relatives' criticism, or concerns about the attitudes of health personnel toward poor or uneducated persons (see Bogue 1983, who cites the first two of these anxieties as major psychic costs deterring contraceptive use). Is procrastination a cause for public policy intervention? Many policy analysts would be reluctant to interfere with an individual's preferences. But this reluc- tance is founded on the assumption of consumer sovereignty that is questioned here. If a case can be made that procrastination in contraceptive adoption leads to a long-run reduction in the procrastinator's welfare, then intervention may be justified. (There is a rough analogy with seat belt laws, which claim to overrule 318 Incentives for Small Families individuals' short-term preferences to promote their long-term welfare.) A possi- ble policy instrument would be an incentive for contraceptive adoption which counterbalances the psychic costs leading to procrastination, allowing the indi- vidual to place greater emphasis on the future consequences of the contraceptive decision. There may be more direct ways of overcoming psychic costs leading to pro- crastination, however. If potential acceptors are deterred by dirty clinics or insensitive staff, it may make more sense to upgrade facilities and retrain pro- viders than to offer acceptor incentives. Alternatives to incentives. As table 1 shows, incentives are seldom the sole alternative to correction of the particular problems discussed above. In particu- lar settings, the cost-effectiveness of incentives should be evaluated relative to that of increasing the number of family planning service points, greater use of fieldworkers, more emphasis on information and communication through the mass media or in the context of clinic services, introduction of community-based distribution systems using village workers, and so on. External Costs of Childbearing Incentives also may be a mechanism by which society as a whole assures that individual couples internalize more fully the costs to society of their decision to have an additional child. We review first the possible sources of externalities and then treat the set of policies, including incentives, that can be used to correct for any externalities. Sources of externalities. The concern about population growth at the national or social level is often based, implicitly or explicitly, on the idea that there are negative externalities to childbearing: my decision to have a child imposes costs on society that I do not take account of (see Simon 1977 for a discussion of both positive and negative externalities to childbearing). Despite the popularity of this assumption, there has been relatively little theoretical work (but see Willis 1987; Nerlove, Razin, and Sadka 1987), and almost no empirical work (except Lee and Miller, this volume), describing the nature and magnitude of these externalities. Lee (1988) describes a number of externalities to childbearing related to pub- lic goods or common property. First, when there is public national wealth (for example, state-owned lands or mineral deposits) the birth of a new citizen dilutes everyone else's claim on this jointly owned property. Similarly, if there are nonprivatizable, congestible common property resources such as fisheries and airsheds, an additional birth reduces per capita consumption of the resource. If the resource is not properly managed, as in the classic "tragedy of the commons" scenario, in which access to the resource is unlimited, then popu- lation growth exacerbates the losses due to mismanagement. A formally similar situation occurs when citizens are entitled to public transfer payments (social security) or subsidized services (education, health), consume Chomitzand Birdsall 319 public goods, and are obligated to pay taxes. In such situations, the birth of an additional citizen adds both to public revenue and expenditure, resulting in positive and negative externalities to other taxpayers. Depending on the amounts and life-cycle timing of tax payments and receipt of benefits, net exter- nalities could be positive or negative (Lee and Miller, this volume). An important pecuniary externality to childbearing occurs in the labor mar- ket. As population, especially rural population, increases, the wages of labor go down, and rents increase. For individual landless or land-poor laborers it is rational to have many children as a strategy for maximizing old-age income. When all pursue this strategy, however, wages are depressed, and both the parents' living standards and those of their children are reduced below their expectations. Hence, one poor family's decision to have children imposes costs on its peers; if all join in a labor-supply cartel and agree to have fewer children, the group's income would increase. As many authors (Ng 1986; Willis 1987; Lee and Miller, this volume) have pointed out, this is not a "true" externality, because it operates through the market. Although wages are depressed by population growth, rents are boosted; high fertility is potentially Pareto-superior to low fertility, assuming that land- lords redistribute some of their rents. In practice, however, the landlords are unlikely to redistribute their rents. From a policy viewpoint, the demonstration that there are no true externalities to childbearing may be of limited interest. Where income inequality and abso- lute poverty are great, and equity is a policy goal, a labor-supply cartel may be one of the most feasible means of redistributing wealth. It therefore makes sense to think of externalities to childbearing as operating within the group of laborers.13 A final source of childbearing externalities occurs when couples care about their relative-rather than absolute-family size. A family's voice in the village council, prestige, physical security, or claim to the use of common property may depend on its size relative to the group (Crook 1978). When a couple has more children, they slightly increase the average family size in their group, making others relatively worse off. An attempt by each couple to have more children than the average family size is mutually frustrating: average family size increases, and no couple achieves the sought-after advantage, although all now have larger families to support. Policy instruments: taxes. If childbearing imposes external costs, the stan- dard economic remedy would be to impose a tax on children equivalent to the external costs. Then, assuming that couples have perfect control over their 13. Evenson (1988) provides estimates of the effect of population on wages and rents in North India. He estimates that a 10 percent decline in population would boost overall per capita income by 8 percent, boost per capita income of rural landless households by 15 percent, and reduce land rents by 25 percent, ignoring scale economies and induced investments associated with population density. If scale economies are taken into account, a reduction in population actually boosts land rents. 320 Incentivesfor Small Families fertility, a child would be born only if its value to its parents exceeded its total cost to society. The response to the tax depends on a couple's income; their taste for children (especially high-order children) relative to goods; and whether their children are net assets. If we make the standard assumption of diminishing marginal utility-so that the fifth child, for example, yields somewhat less satis- faction to its parents than the first-we would expect the tax to reduce family size, but probably not to zero. We would also expect the tax to have a larger impact on the fertility of poorer couples. The welfare-economic approach to population policy implicit in this argu- ment is very different from the standard outlook. Population policies commonly set a quantitative target for a reduction in the birth rate; how that reduction is allocated among families is treated as a secondary consideration, subordinated to logistical and managerial constraints in meeting the quantity goal. In con- trast, the welfare-economic approach is principally concerned with the distribu- tion of births, seeking to ensure that each birth is highly valued both by parents and by society. The sheer number of averted births is irrelevant. (Indeed, if there were external benefits to childbearing, fertility would be subsidized.) These points are illustrated in figures 1 and 2. Figure 1 shows the demand, D, for children (n) of two different types of families. For simplicity, we assume that the price of children C is the same for all family types. The effect of a child tax T depends on the elasticity of demand at the current price. Panel A shows a wealthy family, whose fertility is relatively insensitive to price. A tax equivalent to the external social cost of children reduces the family's demand from n to n . In the absence of the tax, there is a social welfare loss, depicted by the shaded area. Panel B shows a poor rural family, with elastic demand for higher-parity children, but a very inelastic demand for the first two or three children. Figure 2 aggregates these private demands into an aggregate demand curve D. The pri- vate cost curve C shows the private costs per child, assumed constant for sim- plicity. The social cost curve C' includes both private and external costs. The number of children born in a laissez-faire regime would be N, but the optimal number of children is given by N*, and would be achieved through the imposi- tion of a tax T equal to the difference between C' and C at that point. The shaded triangle depicts the total welfare loss: these potential children impose costs on society greater than the enjoyment they bring to their parents. (We sidestep here the philosophical issue of how much weight to attach to the contin- gent future welfare of potential but as-yet-unconceived people; see Dasgupta 1987.) Clearly the magnitude of this welfare loss depends not only on the size of the externality but also on the elasticity of demand. The incidence and effect of the tax depend on individuals' demand elasticities. Family B reduces fertility much more than family A in response to the tax, but still has more children and pays a higher tax. Such taxes have been used or proposed in several countries. China introduced explicit taxes on unauthorized second or third children in 1979 (Banister 1987). Provisions differed between provinces; in Sichuan, as an example, the penalty Chomitz and Birdsall 321 Figure 1. Demandfor Children by lndividual Families Price per child D C+T (A) Wealthy family n* n Children D Price per child (B) Poor rural family C+T C * n Children 322 Incentives for Small Families Figure 2. Social Costs and Benefits of Children: Aggregate Demand for and Supply of Children Price per child D C' C N* ~~N Children was a 10 to 20 percent income tax for seven years (Hardee-Cleaveland and Banister 1988). However, the effects of this economic penalty are difficult to disentangle from the intense political and social pressures that are applied to couples. Singapore, during the 1970s, applied much more modest disincentives, whose effect may have been mostly symbolic (Chen and Fawcett 1979; Saw 1980). In any case the far-reaching economic and social changes in Singapore during this period make it difficult to isolate the effect of these disincentives. Despite their theoretical advantages, child taxes are likely to hurt the poor. For the poor much more than for the rich, children are likely to be an important source of old-age income and security. Yet in no society-developed or developing-is it easy to borrow against the future earnings of one's children. Because of this failure of capital markets, it is easy to imagine a tax forcing a peasant to forgo the birth of a child whose financial security value alone is much greater than the tax. Even if capital markets posed no problem, an adminis- tratively simple tax would probably be highly regressive, and it might be difficult Chomitz and Birdsall 323 to guarantee that the tax revenues would be progressively distributed. 14 There is a particular danger that the tax will penalize the children born, by accident or design, into large poor families. Policy instruments: incentives. Rather than charging a family for the costs inflicted by its child on others, society might choose to bribe the family not to have a child, making the following sort of offer: "We, as a group, will save $C for each child you choose not to have. We propose to share this dividend with you, as follows. When you are fifty, if you have no children, we will pay you $6C; if you have one child, we will pay you $5C, and so on." Once again, assuming perfect control of fertility, the only children born will be those whose social benefit is at least as great as their social cost. Assuming that the incentives are financed via progressive taxation, this version should be more favorable to the poor. This scheme is equivalent to a lump-sum transfer combined with a child tax. Diagrammatically, its impact is similar to that depicted in figures 1 and 2, except that the demand curves simultaneously shift upward as a result of the transfer component. Incentives for small families have been tried on a small scale in Taiwan (Fin- nigan and Sun 1972; Wang and Chen 1973) and in the No-birth Bonus scheme in South India (Ridker 1980). The latter program was instituted in 1971 at three tea estates with large female work forces. For each month during which a woman was not pregnant, the estate would pay 5 rupees (then about a day's wage) into a savings account. The savings account was redeemable, with 5 percent compound interest, at the end of the woman's childbearing years. How- ever, 50 to 100 rupees were to be forfeited if the woman had a third child, 250 in the event of a fourth child, and the entire account in the event of a fifth child. Unfortunately, the experiment was marred by poor administration. Within a few years, the estates ceased to distribute and update the savings passbooks that constituted the participants' principal positive reinforcement for participation. As a result, by 1975, only 8 percent of participating women were able to name any of the forfeiture conditions. The collapse of the No-birth Bonus scheme highlights a serious disadvantage of this type of incentive program-its reliance on a competent and enduring administrative structure. To succeed, such an institution has to monitor a large number of people over a very long period of time, and to convince participants that an inflation-protected payment will be made dependably decades in the future. (There may be a paradox here. If high fertility is a strategy for coping with a politically disorganized society [Cain 1983; Organski and others 1984], then nations that are capable of inspiring the requisite confidence in small-family 14. Ideally, the tax would be child-specific; that is, it would compensate society for the external costs of that child. This might imply higher taxes for children born into poor households, if those children receive from society relatively more benefits (schooling, health, and so on) net of taxes. 324 Incentives for Small Families bonds may not need them). In addition, the fiscal outlay may be very large, a point we discuss in section III. It is sometimes proposed to use entitlements rather than money as a deferred incentive. For example, several Indian states give sterilization acceptors and their families preferred access to health services, education, and loans (Satia and Maru 1986). Where these services are not expanded to meet the extra demand, however, this becomes a disincentive. Such schemes have therefore been crit- icized because they can potentially penalize a child for the parents' action (Edgar and Greenawalt 1977). Policy instruments: quotas. Quotas on childbearing are an alternative to taxes or incentives, but quotas are both less efficient and less equitable. They are inefficient because they do not balance social costs and social benefits in the manner shown in figure 2. That is, they permit children to be born to some parents who value those children only slightly, while denying, say, third and fourth children to others who value them greatly. Quotas are superficially equitable if the same quota applies to everyone. However, quotas deprive the poor of an important resource without compensa- tion (except the benefits of slower population growth, which may accrue to their permitted children, not to them); the rich are much less economically reliant on their children. A law forbidding both rich and poor to have a third child is no more equitable than the proverbial law forbidding both rich and poor to sleep under bridges. III. A SIMPLE BENEFIT-COST FRAMEWORK Section II dealt with the potential welfare benefits that could result from incentives appropriately designed to counteract specific failures of markets and of individual decisionmaking. Potential welfare gains for families include reduc- tions in the financial, emotional, and health burdens of unwanted births; allevia- tion of poverty; and increased shares of ownership of common property. We also alluded to the demographic impact-the number of births averted or deferred. However, it should be clear from our discussion that we believe that welfare benefits should be policymakers' ultimate concern. Demographic impact is only a crude proxy for those benefits, with the advantage of being more easily estimated. Moreover, it is possible to have welfare gains even in the absence of major demographic effects. For example, increased provision of sterilization to women past their prime childbearing years may relieve anxieties about health hazards or social stigmatization associated with late pregnancies. Against the potential benefits of incentives, we can distinguish three kinds of costs. The fiscal cost comprises all public expenditures on the operation of the incentive program, including the incentives themselves. But the economic costs are simply the costs of administering the program, because the incentives are transfer payments that consume no resources. Welfare losses are the third cost. As noted earlier, individuals may suffer welfare losses from incentives if the tax to finance the incentives falls heavily on Chomitz and Birdsall 325 them. But the potential welfare loss of most serious concern is that incentives will entrap individuals into actions against their long-term interest. Monitoring, Fraud, and Administrative Costs An incentive program has two difficult responsibilities. First, it must disburse or collect money, typically in large numbers of small transactions. Although this invites corruption, the success of the Bangladesh sterilization program in main- taining a strict audit system shows that the problem is solvable. Second, the program administration must verify the incentive claimants' com- pliance with the incentive conditions. The cost of doing so (or the trade-off between verification and losses due to fraudulent claims) depends of course on the nature of the conditions. Sterilization incentives are popular, in part, because verification costs are relatively low; -attempts at undergoing multiple sterilizations are easily detected. (Other conditions imposed by sterilization pro- grams, including maximum age and consent of spouse, are not so easily verifi- able, however.) Verification of the use of implantable contraceptives is also relatively easy. Providing incentives for IUD use, however, is a strategy that runs the risk of inducing women repeatedly to remove the IUDS. At the other extreme, verification of the use of pills or condoms is virtually impossible. No-birth bonuses, discussed in section II, avoid method-specific bias by mon- itoring either nonpregnancy or birth but have formidable administrative require- ments. A program that relies on birth monitoring presupposes a well-developed, evasion-proof vital registration system, one that is probably more sophisticated than most developing countries possess. A program that monitors nonpregnancy faces a trade-off between frequency of monitoring and losses due to noncom- pliance. In either case, positive identification and nationwide coordination are needed to reduce multiple claims by the same individual. In sum, it is administratively difficult to devise incentives for long-term con- traception that are not biased toward sterilization (or perhaps implantables). Payments to early adopters or incentives for information acquisition may be more feasible, however. Fiscal Costs and Targeting For many incentive schemes, the incentive payments themselves are likely to constitute the bulk of program expenditures. Because incentive payments are transfers and do not consume real resources, these payments are fiscal but not economic costs. Fiscal costs are important, however, because governments have a limited ability to raise revenues-which is to say that the people who pay the transfers care about their magnitude. The fiscal costs of an incentive program are directly tied to the number of incentive claimants: people who meet the award criteria and receive incentive payments. But not all those who qualify for, and claim, the incentive actually change their behavior (for example, use contraceptives) as a result. As we noted in the preceding section, some claimants are frauds who do not actually meet the award criteria. Others are windfall claimants who would have met the award 326 Incentives for Small Families conditions even in the absence of the incentive. Still others do change their behavior but would have done so for more modest incentive payments. The rents collected by these latter two groups represent fiscal costs without social benefits. Figure 3 illustrates the role of rents in incentive payouts, using a no-birth bonus scheme as an illustration. The two panels show the demand, D, for children of two types of couples. C is the private cost of children; C' is the social cost of children. Offering a flat incentive payment of I = C' - C for each birth less than some maximum, m, results in payouts shown by the shaded rectangles; couple A receives a particularly large payment because their realized fertility, n, is low. (For expositional purposes, we make the unrealistic assumption that the demand curve does not shift upward as a result of the income-transfer compo- nent of the incentive.) Suppose, however, it were possible to pay each couple the minimum necessary to convince them to reduce their fertility to the socially optimal level, n *. Couple A is nearly at that level, so only a small payment, abc, would be necessary; couple B requires a much larger payment, abc, in panel B. The difference between the flat-rate incentive payments and the minimum neces- sary payments represents rents received under the flat-rate system. Clearly the fiscal benefit-cost ratio of an incentive scheme can be increased by decreasing the amount of rents paid. This can be done by targeting incentive eligibility away from windfall claimants (such as couple A) and by customizing incentive levels to match individuals' "reservation prices" for behavioral change. For example, payments and eligibility might be related to age, sex, parity, and geographical region. However, it is easy to imagine situations in which targeting would be politically or socially unattractive: where, for example, fiscal effi- ciency dictates paying higher incentives to wealthier couples or concentrating incentives on a particular ethnic group. Entrapment, Myopia, and Welfare Costs Perhaps the greatest fear concerning client incentives is that they will tempt a myopic or ill-informed individual to sacrifice his or her long-term welfare for a short-term gain (we have examined the converse problem in section II). The problem arises mainly with regard to sterilization, because it is irreversible. Suppose that the poor have short time horizons or high discount rates, limited understanding of their options, and urgent needs.15 Then they will find a cash payment for sterilization irresistible, yet later they may regret their choice. 15. The possibility of limited understanding of their options by some clients is easiest to address. It is not difficult, in principle, to introduce screeners or counselors whose job is to ensure that sterilization adopters understand the risks and consequences of the operation. In practice, an independent survey of Bangladesh sterilization acceptors found that knowledge of sterilization's irreversibility, and of alterna- tive, nonpermanent family planning methods, was virtually universal (Cleland and Mauldin 1987). In addition, 86 percent of male and 76 percent of female respondents had discussed the operation with someone who had been sterilized. However, Thapa and others (1987) reported that 14 percent of low- income acceptors in a Sri Lankan sample did not know that vasectomies were irreversible. Chomitz and Birdsall 327 Figure 3. Rents and Incentive Payments in a No-Birth Bonus Scheme (a) Price per child D C, __ b_= n* n m Children (b) Price per child D Cn b C~m e * hide 328 Incentives for Small Families Myopic acceptors potentially could be screened out by using a deferred incen- tive (King 1974). For example, sterilization acceptors could be invited to return for a checkup a week or two after the operation, at which time the incentive would be paid. This has the added advantage of encouraging follow-up care. It has the disadvantage that it may undercut the value of the incentive as a device for helping people finance sterilization (see section II). This may not matter much if the incentive is more important as insurance against the cost of compli- cations than as a means for financing average costs. Empirically assessing myopia is difficult, but Cleland and Mauldin (1987) explore several possible avenues in assessing voluntariness among Bangladesh sterilization acceptors. The most direct approach is to ask acceptors if they regret their decision, or if they would like to reverse the operation, if that were possible. Although positivists may question the validity of the response to such questions (especially in a fatalistic culture), the results of the Bangladesh survey suggest that the responses were meaningful. The incidence of regret among tubectomy acceptors decreased with family size and increased if the woman had experienced the death of a child since her operation. A second, indirect approach is to evaluate the decision process. Among the Bangladesh acceptors, 98 percent of men and 93 percent of women said they had thought about the decision for more than a month; 98 percent of men and 82 percent of women said they had waited more than a day between making the decision and undergo- ing the operation. Again, a skeptic might question the meaningfulness of the responses. A third approach is to examine the behavior of people suffering temporary income shortfalls. If their discount rates are very high, these people may be more likely to accept sterilization and to do so without deliberation. (Alternatively, one might argue that current privation prompts them to consider more carefully the long-term impact of a larger family size.) Cleland and Mauldin (1987) report that among vasectomy acceptors, those who reported inadequate food con- sumption during the two weeks before acceptance were more likely to have taken less than a month to decide to be sterilized. A behavioral extension of this approach might examine whether there are seasonal peaks in acceptance, for example, in the lean times just before harvest. Such an analysis would have to control carefully for seasonal labor market conditions, though, because the opportunity cost of recuperation also may be low in slack times before harvest. A Framework for Balancing Costs and Benefits Computing the costs and benefits associated with a particular incentive requires a variety of information. We need to know the structure of the incen- tive, which is defined by a set of rewards or penalties (for example, a cash payment); a set of behaviors or characteristics on which the rewards are condi- tioned (for example, adoption of a contraceptive); and a mechanism for mon- Chomitz and Birdsall 329 itoring compliance with the incentive conditions and administering the award or penalty. Then we need to answer three questions: 1. Who claims the incentive? 2. Who changes his or her behavior as a result of the incentive? 3. What are the net welfare gains realized as a result of behavioral change? In principle, given natural or experimental variation in the incentive structure, we could quantify the answers to the first two questions. That is, we could statistically estimate the probability that an individual claims the incentive, and changes his or her contraceptive behavior, as a function both of the incentive structure and the individual's characteristics. This would allow us to compute the fiscal cost (from the answer to question 1) and the effect on fertility, contra- ceptive prevalence, or contraceptive knowledge (question 2). Translating contraceptive prevalence or fertility impacts into welfare impacts (question 3) is far more difficult. It involves quantifying the benefits of smaller families to individuals and to society, and assessing the hazards of entrapment and regret. We believe, though, that an attempt at putting even order-of- magnitude bounds on these effects would be a useful exercise and would offer more guidance than do arbitrary population targets. Ideally, the answers to these questions could be expressed as functions of the incentive structure. Even crude estimates of these functions would give some insight into the relation of costs and benefits to the magnitude of the incentive program. Consider, for example, the costs and benefits of the South Asian sterilization incentive programs. If financial barriers to adoption are relatively small, it is possible that most of the benefits associated with such incentives are exhausted at relatively low payment levels, whereas fiscal costs and possibly the incidence of regret might be expected to increase more than proportionately to payment levels. Unfortunately, despite the controversy surrounding, and considerable expen- ditures on, the South Asian incentive schemes, we understand very little about the claim, behavior, or welfare functions involved (Cleland and Mauldin 1987 represents a heroic and impressive effort to answer these questions without variation in the incentive structure). There have been no experiments and little analysis of natural variation in the structure of these programs. It does appear that the claim rate is strongly related, at least in the short run, to the payment level. In Bangladesh, an October 1983 boost in the sterilization payment from about Taka 100 to Taka 175 was followed by a 143 percent increase in annual vasectomies. But the Sri Lankan family planning program serves as a neater natural experiment because of oscillations in the incentive offer (Thapa and others 1987). Between 1980 and 1983, the incentive payment jumped from 100 to 500 rupees, fell to 200, and bounced back to 500. The number of acceptances closely tracked the size of the incentive, with male acceptances exhibiting a greater incentive elasticity. The post-1983 male acceptance rate, however, was 330 Incentives for Small Families substantially below the 1980 peak response, suggesting that a significant frac- tion of the incentive-sensitive men accepted within a few months of the initial 500-rupee offer. For sterilization, the behavior and claim functions are almost identical, differ- ing only because of a minor amount of fraud. It is the welfare function that is crucial for policy analysis, and here we have even less information. We cannot even answer the demographic-impact question: do higher payment levels tend to attract older, less fertile acceptors? In principle, demographic impact should be quantifiable, but we do not have analyses linking payment levels to births averted. Still less is known about effects on health, per capita income, and measures of psychological well-being. Some interesting data from Bangladesh, however, suggest that the incidence of regret after sterilization is 2.5 times higher among women with two children than among women with three or more (Cleland and Mauldin 1987). Findings of this sort could be useful in designing the eligibility requirements for incentives. Are Incentives Immoral? Consider an incentive for small families or for contraceptive use with safe- guards against the possibility of entrapment, so that acceptors understand the consequences of their actions and are not tempted by immediate payoffs. Is such an incentive immoral? The question cannot be answered by economic theory. A philosophical treat- ment is beyond the scope of this paper, but see Bayles (1980) and Edgar and Greenawalt (1977) for comprehensive discussions. We confine ourselves to com- ments, from an economic viewpoint, on two aspects of the question. First, are incentives coercive? Setting aside the case of myopia, consider a very poor but rational couple who opt after careful consideration for an incentive payment rather than a large family. Some argue that this constitutes coercion because the choice is so stark that the couple's decision is virtually predeter- mined. Yet it is for precisely this reason that an economic analysis claims that their welfare has been improved. It may be quite true that only poor people would opt for the incentive, but depriving them of the option would only make them worse off (see Schelling 1984). There is an important qualification, how- ever: the incentive offer could be considered exploitative and immoral if the offer really should be an unconditional entitlement (Edgar and Greenawalt 1977). For example, we would consider it immoral for a government to offer disaster victims a small-family incentive while withholding emergency relief. Second, many moral codes prohibit certain actions or transactions that might otherwise increase an individual's utility (for example, prostitution, or selling one's own kidney, or purchasing beer on Sunday). Economic analysis can point out losses in output or economic welfare from these restrictions, but it cannot claim that these losses outweigh maintaining the underlying value system. For some cultures a bald choice between money and children may be unac- Chomitz and Birdsall 331 ceptable. Yet a culturally sensitive framing of the choice may serve to legitimize it; after all, implicit trade-offs between children and goods exist for most fami- lies. Thus, noncash incentives may be acceptable where cash incentives are not. In particular, incentives that support family norms-such as educational benefits-may be perceived as legitimate. (We note in passing that incentives that give educational priority to children in small families are equivalent to disincentives that penalize children born to large families, unless additional educational facilities are provided.) Societies where fertility is believed to be too high are faced with what Cal- abresi and Bobbitt (1978) call a "tragic choice." This is a choice that must be made (in this case, a determination of whose fertility should be reduced) but whose framing violates cultural norms. As a result, the choices are made in a hidden or implicit manner. For example, many find it morally unacceptable to trade off the travel convenience of higher speed limits against the resultant increase in traffic fatalities; a cultural norm holds that life is infinitely valuable. However, the logical consequence of the norm-that automotive travel should be entirely prohibited-is also unacceptable. Setting a speed limit thus becomes a difficult exercise. Similarly, if the current level of fertility is deemed too high, then some scheme must be used to allocate the desired fertility reductions. As Calabresi and Bobbitt point out, societies use a variety of implicit economic and social mechanisms to allocate children among families, although generally avoiding recognition that these allocations embody value judgments about whose fertility should be reduced. Incentive schemes make this value judgment explicit and are therefore controversial. IV. NEW DIRECTIONS IN INCENTIVE DESIGN AND ANALYSIS: THE AMMANPETTAI EXPERIMENT An ongoing program in Tamil Nadu, India, rewards clients in a manner that can be interpreted as providing an incentive to acquire information and to overcome procrastination.16 It is of particular interest here for two reasons. First, it avoids the risk of entrapment. Second, recently collected data make it possible to illustrate the applicability of the analytic methodology set out in section III. Background The Ammanpettai Family Welfare Program provided an incentive for trial adoption of pills, IUDS, and condoms in a rural area where knowledge of tempo- rary methods of contraception was low (about 30 percent of the population) and 16. The program was designed by Janice and Carl Stevens and implemented through the Stella Maris Charity Foundation. The following account of the program's history draws on Stevens and Stevens (1987, 1988, 1990). 332 Incentives for Small Families use negligible (about 3 percent). The motivating assumptions for program design were the following: * That there is considerable latent demand for these methods by women who want to space births or who wish to stop childbearing but fear sterilization. (The government program has emphasized sterilization; in this area, female acceptors of sterilization greatly outnumber male acceptors.) * That clinic-based education, combined with supervised trial use, is the most effective means of learning about the advantages and disadvantages of contraception. A limited-term incentive was designed to induce these women to try tempo- rary methods, on the theory that after the trial period, the now better-informed and satisfied users would continue to use contraception. The project was designed as a controlled experiment. Beginning in 1987, women in a control group of villages were offered contraceptive supplies and information through fieldworkers and at a centrally located private clinic, where health services and a range of contraceptives were made available. Women in experimental villages (located the same distance from the clinic) were offered the same services and, in addition, an incentive to visit the clinic. Contingent on nonpregnancy, women were paid 20 rupees upon each of their first five clinic visits. (In the survey described below, the median reported monthly family income was 350 rupees; only 7 percent reported receiving less than 275 rupees.) Women could receive free supplies and services, but no payment, in subsequent visits. Response to the incentive exceeded the capacity of the program, so experimental-village women were enrolled in two waves of 500, in May and November 1987. Although only fourteen control-village women applied to the program in the absence of incentives, subsequent recruitment by fieldworkers resulted in approximately equal numbers of reported contraceptive users in the two groups. To evaluate the program's impact, the Gandhigram Rural Institute conducted baseline (March 1987) and follow-up (August 1989) random population-based surveys of married women, ages 18-45, in both groups of villages. The follow- ing analysis employs data from the 1989 survey (Gandhigram Rural Institute 1989). This survey took place about fifteen months after the last of the five payments was made to participants in the second wave of enrollees in the experi- mental villages. Characteristics of Acceptors The incentive offer did not overcome traditional Muslim reluctance to visit clinics; among ninety-seven Muslim women in the sample, only one participated Chomitz and Birdsall 333 Table 2. Determinants of Incentive Program Participation Standard error of Asymp- Variable coeffi- totic Variable standard Variable Coefficient cient T-ratio mean deviation Constant -3.77598 5.18828 -0.728 1.0000 0.00000 Wife'sage .001918 0.314447 -0.006 27.556 4.6692 Wife's age squared/1,000 .171542 5.53278 0.031 0.78095 0.26831 Ln monthly family income (Rs) .397004 0.434148 0.914 5.8695 0.33851 Wifeilliterate(D) -.180891 0.253227 -0.714 0.51111 0.50174 Wife completed secondary school (D) -.739927 0.551785 -1.341 0.08889 0.28564 Husband completed secondary school (D) -.364679 0.349083 -1.045 0.18519 0.38989 Family owns noland (D) .711211 0.391774 1.815 0.85185 0.35657 Christian (D) .453755 0.294310 1.542 0.22222 0.41729 Scheduled caste .484715 0.273713 1.771 0.25926 0.43986 Number of pre-1987 pregnancies .366090 .229905 1.592 2.7852 1.8219 Number of pre-1987 pregnancies squared -.043105 .030190 - 1.428 11.052 16.508 Note: Dependent variable: participation in incentive program (dummy) (sample restricted to experi- mental villages). Number of observations, 135; mean of dependent variable, 0.429630. Probit estimates: log-likelihood, -83.881; restricted (slopes = 0) log-L, -92.233; chi-squared (11), 16.704; significance level, 0.11693. D indicates a dummy variable. Subsample: non-Muslim, nonsterilized, age under 38, at least one pre-1987 child, most recent birth after May 1983. Data available from authors. in the incentive program. Table 2 presents a probit analysis of program partici- pation among the remainder of the population.17 Several results stand out clearly: among non-Muslims, the program appeals to Christians; members of scheduled castes; the landless; and to women with more than three prior pregnancies. The program appears to have no differential attraction for the illiterate or the better-educated. Interestingly, the cash offer does not tend to attract those with low current incomes differentially; if any- thing, women from higher-income families have a slightly higher probability of responding, although the effect is not statistically significant. In general, these results suggest that the program attracts those with plausible motives for wanting to control fertility (landless, high-parity women) rather than opportunists (the poor) or windfall claimants (the educated, who are more likely already to be familiar with temporary methods of contraception). It is successful in reaching one potentially disadvantaged or underserved group (the 17. In this and subsequent analyses, the sample was restricted to women who were less than thirty- eight years old, not sterilized, and with husbands who were not sterilized at the time of the survey, and who had at least one pregnancy at the time of the experiment's inception, but whose most recent birth had taken place after May 1983. These restrictions approximate the criteria for incentive eligibility, but they may introduce sample selection bias. Adjusting for personal and household characteristics, women in the experimental villages had slightly higher pre-experiment fertility and a lower post-experiment incidence of sterilization. 334 Incentives for Small Families scheduled castes), but it does not attract Muslims, who may also be under- served. Impact of the Experiment Actually, two experiments were carried out. First, in both the control and experimental villages, temporary contraceptives were actively promoted for the first time by fieldworkers. Although we lack full controls for this experiment, temporary-method prevalence in the control villages increased from 3.6 percent to 8.0 percent over the two-year period (Stevens and Stevens 1990). Heightened emphasis on temporary methods by the government family welfare program accounted for some of this increase, which shows the impact of a wider menu of contraceptive choice even in the absence of incentives. In the experimental villages, however, the increase in temporary-method prev- alence was far steeper, jumping from 3.2 percent to 1 8.3 percent of respondents, or nearly a quarter of nonsterilized respondents (Stevens and Stevens 1990). Virtually all of these women adopted pills or IUDS. To make sure that the differential in aggregate contraceptive prevalence is not caused by differences in the composition of the two groups of villages, we use microdata to adjust these results for personal and household characteristics. Although it is tempting to compare the program participants (incentive claimants) to nonparticipants, it Table 3. Determinants of Current Temporary-Method Contraceptive Use Standard error of Asymp- Variable coeffi- totic Variable standard Variable Coefficient cient T-ratio mean deviation Constant -7.96947 4.02351 -1.981 1.0000 0.00000 Wife's age .225284 .264183 .853 27.613 4.5200 Wife's age squared/1,000 -3.86923 4.65236 - .832 .78282 .25844 Ln monthly family income (Rs) .478646 .274797 1.742 5.9168 .41807 Wife illiterate (D) -.318610 .220263 -1.446 .50000 .50109 Wife completed secondary school (D) -.285042 .39708S - .718 .10000 .30065 Husband completed secondary school (D) .263850 .268298 .983 .21304 .41035 Family owns no land (D) .261330 .285460 .915 .81304 .39073 Wife employed outside home (D) .176545 .222473 .794 .39130 .48911 Christian (D) .173866 .295124 .589 .13478 .34224 Scheduled caste (D) .638108 .226185 2.821 .27826 .44912 Number of pre-1987 pregnancies .421092 .235620 1.787 2.6043 1.6441 Number of pre-1987 pregnancies squared - .053536 .0348256 - 1.537 9.4739 13.712 Live in experimental villages (D) .550549 .209385 2.629 .58696 .49345 Note: Dependent variable: currently using modern temporary contraception (dummy). Number of observations, 230; mean of dependent variable, 0.273913. Probit estimates: log-likelihood, -120.84; restricted (slopes = 0) log-L, -135.04; chi-squared (13), 28.384; significance level, 0.0079957. D indicates dummy variables; In indicates natural logarithm. Subsample: non-Muslim, nonsterilized, age under 38, at least one pre-1987 child, most recent birth after May 1983. Data available from authors. Chomitz and Birdsall 335 then becomes difficult to disentangle the causal effects of the program from the determinants of the decision to participate. Of course, as always in nonran- domized social experiments, we have to admit the possibility that our adjust- ments are insufficient-that differences between the two groups are caused by some nonexperimental but unmeasured factors. Table 3 presents a probit estimate for current contraceptive use, restricted to the non-Muslim population. Holding constant residence in a control or experi- mental village, use is much higher among two disparate groups: higher-income households, and the scheduled castes. Holding constant age, income, birth his- tory, education, and religious and social background, residents of the experi- mental villages are much more likely to be using contraception. At the sample mean values of the independent variables, residence in the experimental villages is predicted to boost the likelihood of contraception from 0.16 to 0.32; the effect is significant at the 0.01 level. Is this differential due to the incentive, or could it result from more active promotion by public or private fieldworkers in the experimental villages? Table 4 shows a breakdown of current users in the subsample by their initial source of information about the method. In gross terms, the excess of incentive-village users is accounted for by women who learned about the method at the clinic; all but one of these were recipients of the incentive. Implications Women opted to continue contraceptive use after incentive payments lapsed. So the program appears to yield net welfare improvements while avoiding the entrapment risks associated with irreversible procedures. The relatively low cost (both economic and fiscal) of the program, and its appeal to relatively young, low- to medium-parity women, suggests that the program is a cost-effective Table 4. Number of Current Users of Temporary Contraception Methods by Initial Source of Information Experimental (incentive-offer) Source Control villages villages Stella Maris Clinic (incentive source) 6 29 Government clinic 1 3 Government fieldworker 5 3 Relatives 0 1 Friends and neighbors 3 5 Radio and television 3 2 Past users 0 1 Ammanpettai Family Welfare Program 0 1 contact persons (fieldworkers) Total users 18 45 Note: Subsample: non-Muslim, nonsterilized, age under 38, at least one pre-1987 child, most recent birth after May 1983, currently (August 1989) using modern temporary methods. Data available from authors. 336 Incentives for Small Families Table 5. Summary of Benefits, Disadvantages, and Alternatives to Incentive Schemes Objective (or benefit) Incentive type Disadvantages Alternatives Finance adoption Immediate one-time Possibility of entrap- More clinics (to costs and insure payment or deferred ment for one-time reduce travel costs) against complications one-time payment payment; possibly a (principally for steril- large proportion of ization, where costs windfall claimants are high) Increase diffusion of Payment for attend- Possibly a large pro- More fieldworkers; accurate information ing informational portion of windfall IEC about contraception; presentations; claimants overcome procras- Ammanpettai-type tination limited-period non- pregnancy incentives Overcome costs One-time payments Untried IEc to change norms (uncertainty, social or prestigious awards disapproval) for early for early adopters adopters in low con- only; Ammanpettai- traceptive prevalence type scheme with lim- areas ited enrollment Equilibrate social Child taxes or small- Taxes: apt to be Changes in socio- costs and benefits of family bonds regressive; small- economic environ- childbearing family bonds; require ment favoring small sophisticated admin- families istrative apparatus alternative to sterilization incentives and may also be able to substitute, to some extent, for fieldworkers.'8 (It is likely that the choice is not one of fieldworkers versus incentives but rather of the proper mix.) V. POLICY AND RESEARCH RECOMMENDATIONS Table 5 summarizes the potential rationales or objectives for incentives; the associated incentive design; potential weaknesses of the incentives; and alterna- tive means of achieving the objective. Our principal points are as follows: * Incentive schemes are not all-purpose tools for population policy but are possible responses to particular problems or market failures. Policy analysis should start by identifying a problem and then choosing among available policy responses, including appropriately designed incentives. * There is a basic distinction between incentives that remove barriers to con- traceptive adoption and incentives that compensate for externalities to child- bearing. The latter are more difficult to justify (just as externalities to childbear- ing are difficult to demonstrate); the appropriate amounts are not obvious; and 18. Stevens and Stevens (1988), in a preliminary analysis, concluded that the program was cost- effective. They estimated an economic cost of Rs76 per program participant, compared to a very rough estimate of Rs6SO per acceptor for the government program. The estimates in table 2, along with data on administrative costs, would now permit a more definitive analysis. Chomitzand Birdsall 337 because of monitoring costs they are difficult to implement in a way that is not biased toward sterilization. Among the former, incentives to encourage clients to acquire information have several advantages; in contexts where information about contraceptives is poor, this kind of incentive may be a cost-effective substitute for fieldworkers, without the danger of entrapment and at relatively modest administrative cost. * To evaluate the costs relative to benefits of incentives and their cost- effectiveness relative to other programs requires better information on who responds to incentives, and with what effect, particularly with regard to the large sterilization incentive programs of South Asia. Key issues are assessing the characteristics of acceptors who are prone to later regret; targeting incentives away from windfall claimants; and estimating the response rate as a function of incentive level and conditions. Ideally, one would like an experiment (designed or natural) in which incentive levels varied between districts; data on individual responses would be necessary to analyze the impact of incentives. * The Ammanpettai program of incentives for information acquisition repre- sents a potentially fruitful direction for incentive design. It has the advantages of information incentives noted above and appears to have been effective in increasing women's contraceptive options, use, and welfare. REFERENCES Banister, Judith. 1987. China's Changing Population. Stanford, Calif.: Stanford Univer- sity Press. Bayles, Michael D. 1980. Morality and Population Policy. Tuscaloosa: University of Alabama Press. Becker, Gary, and H. Gregg Lewis. 1973. "On the Interaction Between the Quantity and Quality of Children." Journal of Political Economy 81 (no. 2, part 2): 279-88. Behrman, Jere R. 1989. "The Simple Analytics of Contraceptive Social Marketing." World Development 17, no. 10: 1499-1521. Birdsall, Nancy. 1989. "Economic Analyses of Rapid Population Growth." World Bank Research Observer 4, no. 1: 23-50. Bogue, Donald J. 1983. "Normative and Psychic Costs of Contraception." In R. A. Bulatao and Ronald D. Lee, eds., Determinants of Fertility in Developing Countries. New York: Academic Press. Boulier, Bryan L. 1985. Evaluating Unmet Need for Contraception: Estimates for Thirty-Six Developing Countries. World Bank Staff Working Paper 678. Washington, D. C. Cain, Mead. 1983. "Fertility as an Adjustment to Risk." Population and Development Review 9, no. 4: 688-702. Calabresi, Guido, and Philip Bobbitt. 1978. Tragic Choices. New York: Norton. Chen, Peter S. J., and James T. Fawcett, eds. 1979. Public Policy and Population Change in Singapore. New York: Population Council. Chomitz, Kenneth M. 1984. "A Relativistic Model of Labor Supply." Ph.D. dissertation, University of California, Irvine. 338 Incentives for Small Families Cleland, John, and W. Parker Mauldin. 1987. Study of Compensation Payments and Family Planning in Bangladesh. Dacca: National Institute of Population and Training. Crook, Nigel R. 1978. "On Social Norms and Fertility Decline." In G. Hawthorn, ed., Population and Development. London: Frank Cass. Dasgupta, Partha. 1987. "The Ethical Foundations of Population Policy." In D. Gale Johnson and Ronald D. Lee, eds., Population Growth and Economic Development: Issues and Evidence. Madison: University of Wisconsin Press. Easterlin, Richard, and Eileen Crimmins. 1985. The Fertility Revolution: A Supply- Demand Analysis. Chicago: University of Chicago Press. Edgar, Harold, and Kent Greenawalt. 1977. "The Legal Tradition." In Robert M. Veatch, ed., Population Policy and Ethics: The American Experience. New York: Irvington. Entwisle, B., A. I. Hermalin, P. Hamnuasilpa, and A. Chamratrithirong. 1984. "A Multilevel Model of Family Planning Availability and Contraceptive Use in Rural Thailand." Demography 21, no. 4: 559-74. Evenson, Robert E. 1988. "Population Growth, Infrastructure, and Real Incomes in North India." In Ronald D. Lee and others, eds., Population, Food, and Rural Devel- opment. New York: Oxford University Press. Finnigan, Oliver D., and T. H. Sun. 1972. "Planning, Starting, and Operating an Educational Incentives Project." Studies in Family Planning 3, no. 1: 1-7. Freedman, Ronald, and John Y. Takeshita. 1969. Family Planning in Taiwan: An Experiment in Social Change. Princeton, N. J.: Princeton University Press. Gandhigram Rural Institute, Faculty of Development, Department of Applied Research. 1989. "Population Based Survey Research: A Follow-up Survey of the Introductory Cash Incentive Programme on Family Welfare and Child Spacing in Ammanpettai, Thanjavur." Gandhigram Rural Institute Research Study, Gandhigram, India. Pro- cessed. Hardee-Cleaveland, Karen, and Judith Banister. 1988. "Fertility Policy in China, 1986- 88." Population and Development Review 14, no. 2: 245-86. King, Timothy. 1974. Population Policies and Economic Development. Baltimore, Md.: Johns Hopkins University Press. Loewenstein, George, and Richard H. Thaler. 1989. "Anomalies: Intertemporal Choice." Journal of Economic Perspectives 3, no. 4: 181-93. National Research Council. 1986. Population Growth and Economic Development: Policy Questions. Washington, D. C.: National Academy Press. Nerlove, Marc, Assaf Razin, and Efraim Sadka. 1987. Household and Economy: Wel- fare Economics of Endogenous Fertility. New York: Academic Press. National Research Council. 1986. Population Growth and Economic Development: Policy Questions. Washington, D. C.: National Academy Press. Nerlove, Marc, Assaf Razin, and Efraim Sadka. 1987. Household and Economy: Wel- fare Economics of Endogenous Fertility. New York: Academic Press. Ng, Yew-Kwang. 1986. "The Welfare Economics of Population Control." Population and Development Review 12, no. 2: 247-66. Organski, M. F. K., Jacek Kugler, J. Timothy Johnson, and Youssef Cohen. 1984. Birth, Deaths and Taxes: The Demographic and Political Transitions. Chicago: Uni- versity of Chicago Press. Chomitzand Birdsall 339 Rao, N. E., M. E. Khan, and C. V. S. Prasad. 1986. Health Sector Expenditure Differentials in India: A State and National Level Study. Baroda, India: Operational Research Group (sponsored by the Ministry of Health and Family Welfare). Ridker, R. 1980. "The No-birth Bonus Scheme: The Use of Savings Accounts for Family Planning in South India." Population and Development Review 6, no. 1: 31-47. Satia, J. K., and R. Maru. 1986. "Incentives and Disincentives in the Indian Family Planning Program." Studies in Family Planning 17, no. 3: 136-45. Saw, Swee-Hock. 1980. Population Controlfor Zero Growth in Singapore. Singapore: Oxford University Press. Schelling, Thomas C. 1978. Micromotives and Macrobehavior. New York: Norton. .1984. Choice and Consequence. Cambridge, Mass.: Harvard University Press. Simon, Julian L. 1977. The Economics of Population Growth. Princeton: Princeton University Press. Stanford University and Applied Communication Technology. 1985. The Mass Media and Health Practices Evaluation in the Gambia: A Report of the Major Findings. Washington, D. C.: Academy for Educational Development. Stevens, Janice, and Carl Stevens. 1987. "The Ammanpettai Introductory Family Wel- fare, Child-Spacing Program and Child Survival Program, Tamil Nadu, India: Prog- ress Report." Bethesda, Md.: National Institute of Mental Health and Reed College. Processed. 1 1988. "The Ammanpettai Introductory Family Welfare, Child-Spacing Program and Child Survival Program, Tamil Nadu, India: Progress Report." Bethesda, Md.: National Institute of Mental Health and Reed College. Processed. -1 990. "The Ammanpettai Introductory Family Welfare, Child-Spacing Program and Child Survival Program, Tamil Nadu, India: Progress Report." Bethesda, Md.: National Institute of Mental Health and Reed College. Processed. Thapa, Shyam, Daya Abeywickrema, and Lynne R. Wilkens. 1987. "Effects of Compen- satory Payment on Vasectomy Acceptance in Urban Sri Lanka: A Comparison of Two Economic Groups." Studies in Family Planning 18, no. 6: 352-60. Wang, C. M., and S. Y. Chen. 1973. "Evaluation of the First Year of the Educational Savings Program in Taiwan." Studies in Family Planning 4, no. 7: 157-61. Weeden, D., A. Bennett, D. Lauro, and M. Viravaidya. 1986. "An Incentives Program to Increase Contraceptive Prevalence in Rural Thailand." International Family Plan- ning Perspectives 12, no. 1: 11-16. Willis, Robert J. 1987. "Externalities and Population." In D. Gale Johnson and Ronald D. Lee, eds., Population Growth and Economic Development: Issues and Evidence. Madison: University of Wisconsin Press. World Bank. 1984. World Development Report. New York: Oxford University Press. PRO CE ED fNG S OF THE W O RLD B A NK ANN UAL CON FE RE NC E ON D EVE LO PM ENT EC ONO MI CS 1 9 9 0 COMMFNT ON "INCENTIVES FOR SMALL FAMILIES," BY CHOMITZ AND BIRDSALL Paulina Makinwa-Adebusoye Most governments of developing countries have seen their rapid rates of popula- tion growth-resulting from a sharp decline of mortality since World War II and from persistent high fertility rates-as creating a "population problem." Of course, population growth per se is not the villain; rather, it is unduly rapid population growth that retards governments' efforts to promote social and eco- nomic development. Governments' efforts to address the population problem by influencing fertil- ity levels usually consist of applying one or several of the following policy prescriptions: * Information, education, and communication (IEC) campaigns; *Propaganda to change pronatalist attitudes; * Provision of family planning services; * Manipulation of incentives and disincentives to effect desired fertility behavior; * Pursuit of general or selective development to change the socioeconomic deter- minants of fertility such as education, infant and child mortality, and status of women; - Coercion through state pressure and direct sanctions. Intervention to influence fertility choices of individuals is based on demo- graphic or socioeconomic considerations, health considerations, concern for human rights, and concern for justice and equity. Many of the countries that perceive rapid population growth as a source of concern have favored benign family planning programs consisting of IEC cam- paigns, or the manipulation of public and private access to modern contracep- tives on the one hand, and application of the Bucharest formula-"take care of economic development and population growth will take care of itself"-on the other. However, these have generally failed or are destined to fail to have the desired effect on fertility or to improve overall welfare. It therefore becomes important, as Chomitz and Birdsall's paper strongly implies, that countries con- Paulina Makinwa-Adebusoye is professor of population geography and head of the Population Research Unit, Nigerian Institute of Social and Economic Research, Ibadan. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 341 342 Comment cerned about the negative effects of rapid population growth on public welfare should engage in an intensive search for viable policies that would complement existing programs. Chomitz and Birdsall make a case for client-targeted family planning incentives as one such policy. According to Chomitz and Birdsall, incentives for family planning are particu- larly useful, first, where markets for contraceptives or contraceptive information have failed and, second, where failure of individuals' decisionmaking has resulted in less-than-optimal childbearing that imposes externalities on society. Where people have higher fertility than is optimal, the resulting rapid growth of population is seen as a major contributor to-if not the main cause of- crowded cities, inadequate housing, shortage of infrastructural amenities, and continued illiteracy. Incentive schemes are also justifiable where unwanted births arising from contraceptive failures can jeopardize the health of mothers and children; impair human rights (that is, the basic right of individuals and couples to decide freely on the number and spacing of their children); and impinge on justice and equity. The authors present a typology of incentives according to size and design to counteract each of several problems arising from the two specific market failures. Incentives to correct failure of the market for contraceptive information and services and thereby to remove perceived barriers to contraceptive usage are more readily justifiable on the grounds that they represent payments to meet costs by various categories of acceptors: those who volunteer to receive contra- ceptive information; early adopters; and those who require compensation for costs of transport to service points, for salary loss, or for time lost in case of complications. Incentives that compensate for failure of individuals' decisionmaking and resulting externalities are not so readily justifiable. Disincentives such as explicit taxes on unauthorized children, as tried in China, or the less stringent variety tried in Singapore in the 1970s, are discouraged on the grounds that such taxes are usually regressive and more likely to penalize the poor and children born into large, poor families. Incentives, a form of bribe to dissuade couples from having extra births, are preferred. On the cost side of any incentive scheme are fiscal costs-all public expendi- tures including the incentive payment, economic costs of administering the pro- gram, and welfare losses incurred when myopic acceptors preferring immediate gains later regret their actions. Chomitz and Birdsall underscore the administrative difficulties and economic costs of incentive schemes-particularly the opportunities for large-scale fraud engendered by small payments to millions of acceptors and the near impos- sibility of verification of claims. However, they consider these difficulties serious only in the case of incentives for long-term contraception and for methods other than sterilization or perhaps implantables. The authors consider incentives in the form of payments for information Makinwa-Adebusoye 343 acquisition and early adoption as feasible, and they cite the Ammanpettai pro- gram as an example of the successful application of such incentives. To reduce the heavy fiscal costs involved in incentive schemes, the authors suggest eliminating payments that yield no welfare gains. One way of doing this is by avoiding windfall claimants-that is, those who meet the award conditions but would have small families even in the absence of the incentive. Another is by ensuring that incentive levels match individuals' "reservation prices" for behav- ioral change-that is, payment of the minimum necessary incentive, which may be less than the amount being offered. The third cost, welfare losses incurred by myopic acceptors (especially accept- ors of irreversible methods such as sterilization), is more difficult to prevent, because there is no satisfactory way of empirically recognizing and screening out myopic acceptors. For making incentive choices with reasonable efficiency, the authors suggest a framework for prior analysis of alternatives by balancing costs and benefits. However, the knowledge requirements of the framework are quite formidable. Essential elements of the framework are an understanding of the structure of various incentives and the claim (who claims the incentives), behavior (who changes behavior as a result of the incentive), and welfare functions of incen- tives. To these should be added an understanding of the direct or indirect impact of population processes on the determinants of welfare. There is also the need to compare incentives with other fertility-influencing programs and alternative programs. In theory, knowledge of these relationships would facilitate a ranking of all possible programs, including the laissez-faire option and the selection of the best incentive policy package. Lack of requisite data is mainly responsible for trial-and-error approaches in government policies. Incentive programs are particularly susceptible to pitfalls and inefficiencies because there is no well of past experience from which to draw. Little is known about claim, behavior, or welfare functions of various programs in the South Asian countries that have a long tradition of incentives. Chomitz and Birdsall's paper raises important ethical, administrative, and political questions. On the ethical plane, one finds it easy to agree with the authors' views that incentives are neither coercive nor immoral. If one is to avoid paternalism, one must agree that individuals are best able to recognize their own interest. Therefore, they can be depended on to act on it and thus to ensure that they are as well off as can be after allowing for constraints that may curtail available alternatives. If this is so, incentives can be no more coercive than accepted cultural beliefs and practices (not government policies) that regard the unmarried or childless state (in a woman considered old enough) as deviant behavior. Although there is an ethical dilemma created in trying to decide between the right of individuals to exercise their freedom to choose the size of their families and governments' role as promoter of societal welfare, incentive schemes that permit individuals free choice between various options may avoid this dilemma. Furthermore, incentive or disincentive schemes should include 344 Comment safeguards for the protection of innocent third parties-usually children-who might be affected. On the administrative level, the problems are less easily surmountable. A prominent impediment to the success of incentive schemes is their susceptibility to large-scale corruption and fraud. Even when incentives are for short periods and limited to those willing to acquire information and to early adopters of modern methods of contraception, avenues for corruption exist mainly because of the large numbers of people involved. For example, one Indian program involved more than ten cadres of officials ranging from teachers and public health workers to district superintendents of police. It is impossible to see how incentives on a large scale may not degenerate into a mindless and corruption- laden racket. Still on the administrative level, the ability of governments in developing countries-particularly in Sub-Saharan Africa-to meet the cost of incentives is in doubt, especially under present conditions of austerity. As the authors of the paper have suggested, costs can be reduced in theory if windfall claimants are screened out and if incentives are not larger than individuals' "reservation prices." But no objective method exists of a priori identification of those two categories of potential clients. Costs for incentive schemes in Sub-Saharan Africa may be much higher than in South Asia. The high wage structure and culture of high reward expectations in present-day Nigeria, for example, may necessitate higher levels of incentive payments to motivate behavioral change. Furthermore, governments that con- template or have effected drastic cuts in their subventions for vital sectors whose welfare gains the public readily recognizes (such as health and education) may meet opposition to any move to reallocate incentive payments to undertakings the public views as of uncertain or only longer-term benefit (such as lower fertility). Other administrative problems will arise from the level of confidence in gov- ernments' ability to fulfill promises. If Nigeria is any example, governments of Sub-Saharan African countries do not have a good record of redeeming prom- ises. And the question of credibility is a complex one with many determinants. For example, the success of any incentive scheme will also depend on how well the problem of inflation is tackled. As it becomes evident that existing programs are not enough to achieve desir- able targets of fertility reduction, Sub-Saharan African countries may be expected in the near future to emulate the example of South Asian countries- notably China, India, and Indonesia-and adopt "hard" population programs in the form of incentives and disincentives. A feasible African model of such a policy is indicated by current practices in Ghana and Tanzania, where there are limitations on tax deductions for children, and on paid maternity leave beyond a certain number of children or based on certain minimal birth intervals. Perhaps the attraction of these disincentives lies in their general acceptability, their near- costlessness, and the lessened chance for corruption. Although limited to urban Makinwa-Adebusoye 345 wage earners, these disincentives, if vigorously applied, may reinforce tenden- cies toward smaller family size. Before looking "beyond family planning," however, there is a need to take stock of all available options and invigorate existing programs. A lot of fertility reduction can be achieved in Sub-Saharan African countries by a combination of activities on two fronts. First, the existing cultural roots of African high fertility can be undermined by lifting all or most of the existing pressures on women to bear children. The Nigerian population policy (Nigeria 1988) speaks to this strategy in its sections on the role and status of man in family life and on the role and status of women in development. Second, vigorous pursuit of existing explicit population programs is neces- sary. In this regard, governments' commitments need to be better demonstrated than at present. Strong and up-front political commitment to existing family planning programs is still absent in many Sub-Saharan African countries. Some countries continue to devote very little budgetary allowance to fertility- influencing programs. Similarly, public commitment or the support of public opinion for the stated policy goal-fertility reduction-is nowhere assured. Under present parlous conditions in most countries, intensification of popular motivational campaigns, coupled with governments' vigorous pursuit of the demographic ends of existing programs, may result in faster declines in fertility levels than presently imagined. REFERENCE Nigeria, Federal Republic of. 1988. National Policy on Population for Develop- ment, Unity, Progress, and Self-reliance. Lagos: Federal Republic of Nigeria. I PR OCE ED ING S OF T HE W O RLD BANK ANNU AL CON FE R EN C E ON D EVE LO PM ENT EC ONO MI CS 1 9 9 0 FLOOR DISCUSSION OF THE LEE AND CHOMITZ-BIRDSALL PAPERS On incentives for contraception, a participant noted that in his own study of an Indian program similar to the Ammanpettai program reported in the Chomitz and Birdsall paper, the medical officers all reported that incentives were necessary-without them the officers could not fulfill their quotas for steriliza- tion. The women who had been sterilized also agreed that incentives were important, but for a different reason-the incentive payments were necessary to cover post-sterilization medical costs. The dangers of coercion and irreversibility were the greatest for the poor and for landless agricultural laborers, particularly during natural disasters. The participant reported that evidence from Indian studies showed that during famines or economic crises that had caused severe agricultural unemployment, all members of a scheduled caste in some villages had accepted sterilization in return for an incentive payment. Prompted by a remark made by John Caldwell in the morning's conference session, another participant observed that one potential externality not dis- cussed in Lee's paper related to joint family systems. In such arrangements, there is a good probability that one's own child will be brought up largely by some- body else; this externality would imply that parents would have more children, because they are diffusing the responsibility to support them but have the expec- tation of getting support from them later on. Against this is the probability that in a joint family, aged parents might receive support from other than their own children; this would argue for smaller families. If there was a bias-that is, if the probability of being supported in old age by one's own children is higher than the extent to which one has to support the children when they are young-there would be a tendency for larger families. With reference to Lee's paper, a participant noted that concerns about income distribution and poverty were precisely the motivating factors for population interventions, and to relegate them, as the paper did, as special cases did not do justice to the policy importance of these population interventions. Another participant speculated that if Lee had not assumed perfect foresight in his paper, perhaps the values of the externalities he had estimated would have been larger. The foresight assumption seemed counterintuitive in the context of fertility decisions. On incentives, the participant noted that altering the structure of relative prices between different contraceptives also created incentives within This session was chaired by Carmen Mir6, director of the Institute of National Studies at the University of Panama, Panama City. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 347 348 Floor Discussion of Chomitz-Birdsall and Lee Papers which contraceptive prevalence could be maintained at a lower family planning budget. He cited the example of Thailand for such a policy. A participant, commenting on the Ammanpettai program, said that the pro- gram's founders had thought that its main effect would be to "condition" people to accept contraception by providing monetary incentives. However, it had far exceeded their expectations by serving as an excellent marketing and educa- tional device, giving married women less than thirty-five years old an oppor- tunity to hear about and try contraception and to overcome hesitation and fear. Whether such programs would work in areas of Africa where pronatalist tradi- tion is strong is another matter, however. A participant noted that many of the externalities cited in the Lee paper were not typically the ones considered by policymakers in framing population poli- cies. For example, externalities relating to foreign aid did not seem to be very relevant. A World Bank participant raised the issue of the difficulty of evaluating incentives programs because what is measured is almost always the acceptance of contraceptives and not the end impact of the incentives on fertility. There are also myriad stories about the effects of incentives in producing windfall claimants-for example, women accepting sterilization who weren't going to have more children anyway. Thus, acceptance rates can be highly misleading, especially in the presence of incentives. He also made the point that the Lee paper tended to mix externalities and income distribution questions. Parents do not consider the wage-depressing effect on the future labor force of having an additional child. The impact of an additional child here is not an externality, just as growing more wheat and depressing wheat prices is not an externality. So, the debate in the context of population growth is generally not about pure externalities but about redistributing future wealth and income. In response to this participant, Lee clarified that his paper is concerned with only one among a large set of reasons for programs that went beyond family planning. He agreed with the need to keep pure externalities and distributional issues separate. Lee stood by his paper's conclusion, which Ainsworth (discus- sant) had considered too strong, that given the current state of research, exter- nalities to childbearing do not provide a convincing rationale for fertility policies involving financial incentives or forms of coercion. He cautioned readers not to interpret his paper simplistically to mean that countries that were attempting to control population growth must be facing negative externalities associated with population growth, and vice versa. Regarding the comment about perfect fore- sight, Lee agreed that it was not a very good assumption, as he had himself pointed out in the paper; furthermore, the paper noted that the lack of foresight and information created a role for government educational programs. Lee reiterated his paper's conclusion that with our present state of knowledge, it did not appear that high fertility in developing countries came about because of the existence of externalities that did not have to be borne by parents and were passed on to society. What then explained the pervasive sense that fertility Floor Discussion of Lee and Chomitz-Birdsall Papers 349 is too high-inefficiently high? Lee opined that high fertility in developing coun- tries reflected inefficient or missing markets-for insurance, for example-and inadequate public sector institutions. Because of these deficiencies, children rep- resent the only viable form of old age security. Chomitz said that their paper, contrary to Makinwa-Adebusoye's (discussant) comment, did not make a general and strong case for incentives. Rather, the paper's message was that there was a potentially strong case for incentives, but there was an equally strong case against them. The policy analyst's job was to determine what makes sense in a particular context. Chomitz argued that cost- benefit calculations of incentive programs were feasible, important, and very inexpensive compared with the overall program costs, particularly if programs were set up so that they generated variation in the data. Chomitz agreed with the comment that incentive schemes have to evaluated in terms of their effects on fertility as such, and not acceptance rates. He also found the suggestion intriguing that incentives were necessary to cover post- sterilization medical costs-a sort of insurance scheme. Regarding the comment that incentive schemes would be misfiring if they provoked a rise in sterilizations during hard times, Chomitz said that he and Birdsall had suggested in their paper that any blip in payment acceptances during such times would be a cause for concern. Carmen Mir6 (chair) ended the session by noting that both the topics of externalities and incentives were not likely to be resolved in the near future and will continue to be discussed for some time. PRO C E ED ING S OF THE W OR LD BANK ANN UAL CON FE RE NC E ON DE V E LOPM ENT EC ON OM I CS 1 99 0 Project Appraisal and Planning Twenty Years On I. M. D. Little and J. A. Mirrlees This paper discusses the use and usefulness of social cost-benefit analysis. It briefly recalls the main features of the procedures put forward in the book by the authors and addresses some of the issues that have arisen in theoretical work since the early 1 970s, particularly the implications of uncertainty. An analysis of the value of project appraisal under uncertainty is provided. The rise and decline of project appraisal in the World Bank and elsewhere is charted, and reasons are considered. Particular attention is given to the difficulty of using shadow prices and to the institutional environment. It is argued that evidence from analysis of the World Bank's projects is consistent with placing a high value on the process of economic appraisal. The costs of appraisal are noted, as is the need to avoid excessive complexity. The paper argues for the impor- tance of an economic appraisal system and of attending to the incentives that operate in such a system. In the late 1960s there was considerable development of methods for applying social cost-benefit analysis to investment in developing countries. In the 1970s, these methods began to be applied. From 1974 to 1982 there were massive public sector investment booms in many of the developing countries. It appears that much of this investment yielded little or nothing. Were the methods flawed in principle? Were they ignored? And if so, why? Was the possible effect of project appraisal swamped by unforeseen events that rendered potentially valu- able investments useless? We do not provide definitive answers to these ques- tions, but we hope to contribute to their answers by discussing the current state of the theory of project appraisal and the current practice and influence of social cost-benefit analysis, particularly within the World Bank. I. REFLECTIONS ON THE PRINCIPLES OF PROJECT APPRAISAL In our book (Little and Mirrlees 1974) we provided a method for social investment decisions. In brief, we proposed the following techniques: I. M. D. Little is emeritus fellow and J. A. Mirrlees is professor of economics at Nuffield College, Oxford. The title of the paper alludes to that of their book on project appraisal; the paper does not deal with planning directly. The authors wish to acknowledge discussions with economists in the World Bank and with J. D. MacArthur,J. M. Healey, and J. Wilmshurst. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 351 352 ProjectAppraisaland Planning Twenty Years On * Measuring the values of outputs (benefits) and inputs (costs) with shadow prices * Using border prices as shadow prices for traded inputs and outputs, or-in cases where demand or supply is not independent of price-using the marginal revenue or cost to the country in foreign exchange, as a first approximation * Where possible, using costs, themselves measured at shadow prices, as shadow prices for nontraded inputs * Using conversion factors, estimated separately for a number of different broad categories of inputs and outputs, to calculate shadow prices from market prices for most minor inputs and outputs * Using as shadow wage rates (SWRS) market wage rates discounted by a conver- sion factor estimated from an overall study of the economy, and depending in particular on a judgment that public income is more valuable at the margin than private income * In cases where no other rule is applicable, using a standard conversion factor (SCF) to deduce the shadow price from the market price, to be estimated by comparing (domestic) market prices with border prices for traded goods, and others for which a sound estimate of the shadow price is available - Estimating the private and public parts of the net (social) costs and income of the project, so that private income and costs could be discounted relative to public income and costs, and giving a low weight to private profit income • Basing the shadow prices used on forecasts of border prices and market prices, not usually on their current values - Discounting net social profits so calculated by means of an accounting rate of interest (ARI) that is high (or low) enough to be expected to ration investment projects in the whole economy to the funds available-a rate that could well vary over time-but in no case using a discount rate less than the rate avail- able for investment in international capital markets * Allowing for uncertainty only to the extent that the profitability of the project was expected to be correlated with the general state of the economy * In the case of large projects, allowing for changes in prices brought about by their introduction, and estimating the incremental value of outputs and inputs by "surplus" calculations * Converting all external effects of the project to numerical terms by making some estimate of the cost or value in terms of public income and including them directly in the calculation of the present social value of the project. These features of the method proposed had grown out of wide debate and discussion among economists. Other studies of cost-benefit analysis which reflected that debate in their own way differed in some respects, but there was much in common.1 We attempted to provide a systematic procedure that would 1. See, for example, the famous UNIDO Guidelines (UNIDO 1972), written by Dasgupta, Marglin, and Sen; work by international trade theorists, much of it presented in Corden (1974); a notable series of studies by Arnold Harberger, collected in Harberger (1972); other work collected or referred to in the useful collection by Layard (1972); and an earlier formulation of our own, Little and Mirrlees (1969). Little and Mirrlees 353 correct all the important distortions. The method we proposed is not a theoreti- cal ideal. It was intended to be practical, and therefore it simplifies and approxi- mates. Even so, some aspects have been found hard to implement. In later sections we discuss implementation. The present section deals with some aspects of these methods, particularly the more controversial features. An extensive literature in the last twenty years has subjected the methods, or at least some of the principles embodied in them, to critical analysis.2 Arising from that, there are some respects in which expansion, development, or revision of these procedures seems to us to be warranted. But to a large extent we stand by them. Traded Goods Much of the literature on methods of social cost-benefit analysis has concen- trated on the issue of optimal shadow prices under perfect certainty. Squire (1989), in the Handbook of Development Economics, concludes that the litera- ture has shown that the border price principle for traded goods is quite robust. In the case of a small country, for which border prices are essentially indepen- dent of the amounts imported and exported, quite weak assumptions are suffi- cient to justify the border price principle. When price varies with the amount of trade, it is not exactly right to use marginal revenues and costs. This was pointed out in Little and Mirrlees (1974, p. 229), but it has tended to be neglected. A more extensive treatment has been given in Bhagwati and Srinivasan (1978). The point is that the change in exports, say, affects not only the price received by the country in world markets but also the prices received by domestic producers and paid by domestic consumers, for these prices are generally linked-by taxes, or by the workings of price control systems-to the world price. These changes in market prices involve changes in the welfare of consumers and the actions of producers that ought not to be neglected. The calculations required in principle are similar to those required for the correct pricing of nontraded factors of production. The attractive idea that one could look only to the external environ- ment when estimating shadow prices is not justified-or, at least, it is not exactly justified. This is only one example of the widespread effects that should in principle be estimated to calculate shadow prices. Although it is reasonable-in the absence of a good applicable model of the economy-to guess that these hard-to-trace effects would not much modify a first approximation based on marginal foreign revenue alone, that is a guess, warranting at least testing in plausible models. Nontraded Goods A major simplifying idea behind the shadow pricing methods suggested is that given an estimate of the shadow wage rate and an estimate of the accounting rate of interest, it would be possible to calculate shadow prices for many nontraded 2. Two recent surveys (Dreze and Stern 1987; Squire 1989) give accounts of the literature, particularly on shadow pricing. 354 Project Appraisal and Planning Twenty Years On goods on the basis of input-output data, describing costs of production. This is based on a general principle that the shadow price for a commodity produced under competitive conditions and constant returns to scale is equal to its average cost, calculated using the shadow prices of the inputs into its production (includ- ing labor and capital).3 Granted the often reasonable approximations of constant returns and compet- itive conditions, this social cost principle seems common sense, but there is a hidden difficulty-one that has struck economists who consider economies in which the number of traded goods appears to exceed the number of nontraded goods and factors-too many equations chasing too few unknowns. The difficulty can best be brought out by contemplating a strikingly unsat- isfactory, but not unrealistic, state of affairs. Suppose that two private pro- ducers are producing the same commodity, using only labor as an input, and one of them is more efficient than the other. Suppose also that subsidies are provided to the less efficient, so that in fact both firms operate. Costs of production in the two firms are different, when measured using shadow prices. Therefore, the social cost principle does not give an unambiguous rule for calculating the relationship between output and input shadow prices. The analysis of Diamond and Mirrlees (1976) assumed that equilibrium was unique. This is a common assumption, but it is not always warranted. In the situation described, we can only say that the appropriate output shadow price depends on what will actually happen when demand for the product changes. Marginal social cost should be calculated, but what that is depends on the shares of the two firms in the change in aggregate output. There is no good reason to expect marginal cost to be equal to average cost in this case. The main areas in which these considerations are relevant is when import substitution is subsidized or when quantitative controls are used to achieve the same effect. Then the border price principle is correct when trade can be expected to adjust to changed demand. Otherwise, economic theory cannot predict how much domestic production will be allowed to supply. The project appraiser has to guess what will be allowed to happen. The appraiser should also make recommendations about sources of supply and point out the possible impact on the value of the project. The right answer to the project question may not be yes or no, but rather "only if these inputs are imported instead of being produced domestically." The social cost rule is incomplete in the sense that it does not enable appraisers to calculate theoretically satisfactory shadow prices for all goods. Sieper (1981) and others have found relatively simple rules that may sometimes be used for calculating the shadow prices for nontraded goods in some of the more difficult cases (for the main results of Sieper's unpublished paper, see Squire 1989). The last resort is to derive the shadow price from the market price by applying a standard conversion factor based on comparisons of shadow and market prices for commodities (such as traded goods) where the shadow price 3. A general theorem with this consequence was proved formally in Diamond and Mirrlees (1976). Little and Mirrlees 355 has been satisfactorily estimated. Despite the weakness, we believe that our proposals take good care of the major distortions. In the simplest terms, our rules would not permit projects making cars at a greater foreign exchange cost than by direct purchase abroad. Another aspect of the social cost procedure has drawn comment (see, for example, Srinivasan 1982). It is essential for applying the principle that there be some means of making prior estimates of shadow wage rates and discount rates. We proposed basing these on simple macroeconomic models, and calculations of that kind have been done for one or two countries. (In other cases, estimates of conversion factors for other countries have been used, and the ubiquitous 10 percent accounting rate of interest does not derive from any serious model.) There is no economic theorem that says that the strategic shadow prices calcu- lated in this way are, in some sense, best unbiased estimators; nor is it easy to conceive how such a theorem could be formulated. As with other features of practical welfare economics, the recommendation to use simple macroestima- tion is essentially intuitive. Income Differences The presumption that public income is more valuable than private income has been challenged. That presumption forms the basis for arguments that the shadow wage rate should be a substantial proportion of actual wage rates (after adjusting from market prices to shadow prices for consumption goods). The reasons for it were, first, that government could use additional revenues for further socially profitable investment to a greater extent than private income earners; second, that there are administrative costs to raising tax revenue; and, third, that raising tax revenue creates distortions. The first consideration is weak for countries in which public income seems to be spent on, or diverted to, many objects of little value. The second is not very large. The third argument, which is emphasized by Squire (1989), seems to need further analysis. Claims about the cost of public funds in developed countries that are based merely on the existence of distorting taxes are invalid. In any economy there are distorting taxes on labor and commodities. Revenue is rightly raised that way, because discriminating lump-sum taxes are not possible. In many economies there also are elements of the tax system, such as income tax allowances, that approximate a uniform lump-sum subsidy. If that uniform subsidy is at an optimal level, then a public expenditure that brings about equal income increases for a representative sample of people is just as valuable as a project that increases public income by the same amount. That is because one can be trans- formed into the other, at the margin, without distortion. Projects that augment private incomes, however, generally do so differentially, though not necessarily benefiting the rich more than the poor. Taking the usual view of the relationship between income and welfare, private income changes accruing to those above mean income levels are worth less than public income changes; but in the oppo- site case, they are worth more. The above argument may not be very relevant for developing economies, 356 Project Appraisal and Planning Twenty Years On where scarcely any policy tools approximating a uniform subsidy exist. Typ- ically, a reduction in government income or rise in expenditure would be offset by an increase in the "inflation tax," or, after some delay, by taxes on consumer goods, reductions in other public services, or indeed a cutback in investment. Again, a judgment on the relative value of public and private income turns, in part, on a welfare judgment of the relative value of income to people with different incomes. When changes accrue to those of average incomes, a useful guide is the Ramsey model of an economy with identical individuals and all government revenue raised by taxes proportional to sales and purchases of commodities (that is, no tax allowances or general subsidies). For such an economy, plausible specifications of consumer preferences yield a 20 to 40 percent premium for public over private income. (One attempt to estimate this for the U.S. economy is Ballard, Shoven, and Whalley 1985.) When private income accrues to those with incomes much above the income of the average taxpayer (often quite a low income level in developing countries), its value relative to public income is correspondingly lower. Uncertainty It has been argued that careful project appraisal is worth little when uncer- tainty about the project and about the environment within which it will operate is great. Project appraisal is costly, and the benefits of better decisions could fail to justify that cost. But it is by no means clear that the benefits of improved decisions are slight in a very uncertain environment. We have considered this question and find that there can be a reduction in the value of appraisal because of general uncertainty. But in principle it remains considerable. One can think of cost-benefit analysis as the elimination of error-not all error, but some of the error implicit in decisions by hunch or apparent financial profitability. A theory of these matters is sketched in the appendix. It is shown, using a plausible simple model, that in the context of project appraisal, we should expect some reduction in the value of appraisal when general uncertainty is greater. A simple rule of thumb for judging the benefit is developed. It says that the value of a system of project appraisal is (at least) 10 percent of the standard deviation of the errors removed by appraisal, multiplied by the ratio of that standard deviation to the standard deviation of errors not removed. In section III below we discuss some evidence suggesting that the ratio of these standard deviations is around unity. The variation in observed rates of return is very great. It is not implausible to suppose that the standard deviation of errors removed by appraisal is at least a quarter of the mean net value of projects. With good appraisal it ought to be much more. On that basis, one can claim that appraisal is worth at least 2 percent of the mean net value of projects appraised. In our book on project appraisal, we recommended allowing for uncertainty by subtracting from the expected value of net social profitability of the project in any year an amount equal to Acov(P, Y)/E(Y), where cov(P, Y) is the covari- Little and Mirrlees 357 ance between social profit P and national income Y, E( Y) is the mean expected value of national income, and A is a measure of risk aversion known as the coefficient of relative risk aversion, for which we suggested the value 2. Another way to express this is to suggest that the net value of a project in a year be taken as (1) E(P)(1-A * r * cvp - cvy) where r is the correlation coefficient between the project's social profits and national income, cvp is the coefficient of variation (ratio of standard deviation to mean) for social profit, and cvy is the coefficient of variation for national income. The justification for these rules is provided in the appendix. Estimating these correlation coefficients and variances and means is a matter of skill, requiring training, not least in the concepts and judgment of appropriate values in well-specified examples. Frequently there would be very little data to go on. One could use data for similar industries in similar countries where there is more relevant experience. The task is essentially equivalent to the use of "beta coefficients" by financial managers, and it is as hard or easy as that. Some allowance for uncertainty should be made, though it seldom is. The appropriate allowance can be substantial, especially in the later years of a proj- ect. But for most countries, the coefficient of variation for national income is not very large, looking forward for a decade or so, if one may be guided by time series; and therefore in most cases the allowance for uncertainty is not a large adjustment in the value of the project. It should be noted that the allowance could take the form of an increase in the value of the project, if it were negatively correlated with national income. In our book we also pointed out that allowance should sometimes be made for inflexibility, for the cost of irreversible commitment in projects, though we claimed that such an adjustment should seldom be required. We probably understated the importance of this consideration (see Henry 1974 for an analy- sis). In a sense, the issue of flexibility is the desirability of continuing appraisal during the life of the project. At the time of initial investment appraisal, the project can develop in many ways; and some projects have more possibilities of development and adjustment than others. The more flexible projects are worth more for that reason. How much more they are worth depends on how much, and when, flexibility is expected to be exercised. Because of discounting, flexibility that can occur only after decades (as in the choice between projects that are long-lived and ones that are not quite so long- lived) is not likely to be worth a great deal. Flexible response to variations in prices may well be exercised before many years have passed. Granted the great uncertainty in many prices, such flexibility may be of considerable value. The specific rules we listed in our book neither draw attention to this nor provide rules of thumb for evaluating it. In principle, flexibility is already valued in the estimation of expected profit, because the probability distributions of social profit for various more or less flexible projects or programs should have been 358 ProjectAppraisal and Planning Twenty Years On compared, allowing for the response of the project to varying circumstances. But project reports seldom if ever allow for responsive alteration in production, such as closing (temporary or permanent) or expansion. They should. At least they should when it could be important. But this seems to be a difficult and costly task. For a substantial price, consultants will construct a model and do a computer simulation of many alternative futures. It may contain many fairly arbitrary assumptions, and it may ignore shadow pricing, which is likely to be quantitatively even more important. In large projects, where flex- ibility could be substantial in the first two decades, it may be worth attempting such an analysis, but the appraiser needs to keep close track of just what possi- bilities are being simulated. This is an area in which it is too easy to make spurious claims which add to the apparent value of the project. Appraisal Incentives This last line of thought brings us to the question of the environment within which project appraisal itself may operate. Appraisers-like workers, managers, and politicians-have their incentives, and they may be expected to respond to them. Space allows little systematic discussion here. The obvious difficulty is that appraisers will be judged on the number of "good" projects they find rela- tive to the numbers others find. More and more often, as a system of appraisal operates, the standards on numbers of projects found will be based on past performance, and it will not be surprising if arguments and analysis are shaded to continually increase that number. As we discuss below, such a phenomenon appears to have occurred in the World Bank. We have no doubt it has also occurred elsewhere. The difficulty is that the obvious basis for incentives-comparison of appraisals with results-is almost impossible to use, because of the great length of time that elapses between initial recommendation and final results. And if comparison is used, the reappraisal process could be corrupted. There are other possibilities: independent reappraisal of randomly selected projects shortly after the original appraisal, or explicit grading of proiect appraisals by quality, as judged by independent inspectors, paying attention not only to analytical method and conformity to standard procedures but also to checking the empiri- cal basis of the appraisal. If good project appraisal warrants expenditure of resources, as we argue in the appendix and in section IV, so does good appraisal of appraisal. We do not argue for any further levels of appraisal, though that raises an interesting question. II. THE RISE AND FALL OF COST-BENEFIT ANALYSIS IN THE WORLD BANK AND THE STATE OF THE ART ELSEWHERE During the 1970s there was a remarkable development of both the methodol- ogy and the use of shadow pricing in estimating the true national value of projects in economies in which, primarily because of government taxes and Little and Mirrlees 359 controls, the allocation of resources was distorted. This happened first because of an increasing realization of the degree of these distortions, and second because of a desire to apply systematic decision procedures where previously there had been haphazard action and arbitrary plans. We would claim that our development of a method by which a set of consistent shadow prices could be estimated had some part in it (Little and Mirrlees 1969; Dasgupta, Marglin, and Sen were also devising guidelines for project appraisal; see UNIDO [U.N. Indus- trial Development Organization] 1972). Within two or three years most donor agencies and a few developing countries were using some simple version of the methodology. But among donor or lending agencies the methods were most fully used and further developed by the World Bank (the first major demonstration of how to estimate a full set of shadow prices was by Scott; see Scott, MacArthur, and Newbery 1976). The World Bank's own version of the methodology was published in Squire and van der Tak (1974). This methodology incorporated "social" prices whereby consumption was weighted according to the income of the consumer. If such distributional weights were not used, then the shadow prices were (tendentiously) called "efficiency" prices. Efficiency pricing was based on the remarkable and extreme presumption that a dollar's worth of consumption had the same social value to whomsoever it accrued. Efficiency prices were also based on the presumption that all kinds of income and all uses of income were equally valuable. Our book (Little and Mirrlees 1974), which also incorporated distributional weights, was published at about the same time. But it is important to note that Little and Mirrlees (1969), which did not use distributional weights, had not assumed that all uses of income were equally valuable. As in UNIDO (1972), investment was presumed to be worth more than consumption. In Little and Mirrlees (1974) the numeraire was changed to uncommitted government income, which was also presumed to be worth more than average consumption. A battle raged in the World Bank during the 1970s about whether social prices should be used. Formally, the "social price brigade" won, in that guide- lines on the use of distributional weights were actually incorporated in the Operational Manual in 1980. In practice, we believe, they were hardly ever used except in an experimental manner in a few cases. But aside from this battle, there was a good deal of debate, interest, and research. Guidelines flowed from the Central Projects Department under Warren Baum, which also ran a long series of workshops. With or without distributional weights, a fairly full set of sectoral conversion factors and shadow wage rates was estimated by staff mem- bers or consultants for nearly twenty countries. The use of the foreign exchange numeraire became standard practice in the World Bank and remains so. Multiple conversion factors are an important, indeed essential, feature of the Little-Mirrlees and Squire-van der Tak methodologies. We suppose that they were used for the countries for which they were estimated. Shadow pricing in the World Bank reached its apogee around 1981. Yet, inspection of a sample of completion reports of projects approved in the 1970s suggests that the Squire- 360 ProjectAppraisal and Planning Twenty Years On van der Tak system was far from being fully adopted, either intensively or extensively. Many country economists were still unconvinced, either because they were overworked or lazy or because they were genuinely unsure of the usefulness or validity of the system. At the same time, there have always been senior World Bank staff members who have remained suspicious of economic analysis, especially when it conflicts with their hunches about what is good for development-whether it be integrated rural development, afforestation, or steel plants. In the 1980s there was a decline in the use of shadow pricing at the Bank. In terms of interest, attention, guidance, supervision, and research, the decline is steep. The change in actual working practice may be less dramatic. We say this partly because the Bank, even at the apogee, had been less than thoroughly permeated by the Squire-van der Tak methodology, and partly because it is difficult to describe the present state. This difficulty stems from the fact that there is no longer a Central Projects Department, offering common advice and exercising common quality control over project appraisals. We do not know whether regions now diverge significantly, but we guess probably not (or not yet, anyway). There have always been sectoral differences, particularly in the case of power (but also telecommunications and water supply), and these remain. The benefit of power is generally taken to be measured by what it is sold for. This must often yield an underestimate of benefit, where tariffs are controlled at low levels. We were told during interviews we conducted at the World Bank in the process of writing this paper that the World Bank's loans are made conditional on tariffs being raised (though such conditions are not always fulfilled) if this is required to show a 10 percent return. In fact we found one appraisal report of a hydro- electric project in which the benefit was reckoned to be the cost of power from a thermal project (based on imported oil); the benefit was thus grossly overesti- mated, because the World Bank grossly overestimated the future price of oil. We have mentioned this methodological divergence because we return below to the proper appraisal of power and other nontraded goods projects. However difficult they are to make, some general propositions as to present practice are essential, although we have not made sufficient inquiries to be sure they are altogether correct. We believe the following: * Social pricing, using distributional weights, has been abandoned. * No distinction is made between public and private income, or between the uses of income-whether saved or invested. * Sectoral conversion factors are rarely if ever calculated and used. * Shadow wage rates are not systematically used or estimated. * The values of nontraded goods are mostly converted to border values by a single standard conversion factor. To put this in another way, there is seldom if ever any attempt to estimate the actual foreign exchange consequences of using or producing particular nontraded goods. It is also equivalent to saying Little and Mirrlees 361 that the relative prices of nontraded goods are assumed to be undistorted (except perhaps that taxes may be subtracted). If the above propositions are correct, they amount to saying that apart from the cutoff economic rate of return, only one shadow price-a standard conver- sion factor or shadow exchange rate-is used (and it is unclear how this is estimated, or whether it is estimated in the same way for different countries). This is probably not very different from the practice in the 1960s, when shadow exchange rates were used (and even shadow wage rates) when required to make the rate of return good enough. The only important improvement may be some greater use of border prices for tradable goods. It is doubtful whether the World Bank can now be said to have a standard methodology. It is certainly not that of Squire and van der Tak (or of Little and Mirrlees, or even of UNIDO. )4 Before speculating about the reasons for this withering, we briefly survey the situation in the regional banks and other donor agencies, and in developing countries. Here, nothing much seems to have changed in the past ten to fifteen years. There was no buildup of interest and development of the methodology similar to that in the World Bank, nor consequently any subsequent withdrawal. Those adopting in the mid-1970s a limited version of Little-Mirrlees, whether directly or following the World Bank, include the Asian Development Bank (ADB), the Inter-American Development Bank (IDB), the Overseas Development Administration (ODA) in the United Kingdom, and the Kreditanstalt fur Wie- deraufbau in Germany. Japan also claims to follow the World Bank. France and the European Commission still use the "effects method," despite its errors, as pointed out by Balassa (1976). Other countries seem to follow no particular methodology, or hardly use cost-benefit analysis at all. The Canadians and the British agree, however, that most consultants are familiar with the World Bank methodology and use it (and the British add that they are required to do so).5 The IDB is still active in calculating national parameters using input-output methods. Current work apparently includes estimates for Colombia, Panama, and Venezuela. But we do not know what use in practice is made of them. Apart possibly from the IDB, the British ODA seems to be the most persistent upholder of the principles adopted in the mid-1970s. It also appears to be more convinced of the practical value of the methodology than other agencies. A third edition of the ODA's Appraisal of Projects in Developing Countries (ODA 1988) retains guidelines for the calculation of sectoral and other conversion factors, for shadow wages, for the incorporation of different values for saving and invest- ment, and for the use of distributive weights. Although the latter two adjust- 4. In a recent Development Assistance Committee review of Project Appraisal Criteria and Procedures (Development Assistance Committee, OECD 1987 [87J11), the methodology used by the World Bank is described as being that of Squire and van der Tak. Presumably the World Bank supplied this information, but it seems to be more of a travesty than the truth. 5. These observations are based mainly on Development Assistance Committee, OECD (1987); inter- views with the United Kingdom ODA staff; and hearsay (for the ADB and IDB). 362 Project Appraisal and Planning Twenty Years On ments are rarely used, examples are given of the careful calculation and use of multiple conversion factors. Information on the state of the art in developing countries is spotty. Chile, partly inspired by Harberger (1972), has a well-developed system of appraisal operating at all government levels. In India, the project appraisal division of the Planning Commission continues to operate, but it appraises only about one- third of public sector projects. Lal (1980) worked out a full set of national parameters, but practice has degenerated to the use of a single shadow exchange rate which never changes. The Development and Project Planning Centre, Uni- versity of Bradford, has recently on request calculated national parameters for Ethiopia and Jamaica. Work has begun on China and Sri Lanka.6 Squire ( 1989) reports on the basis of information from World Bank staff members that (of twenty-seven countries surveyed) Cote d'Ivoire, Ethiopia, Republic of Korea, Pakistan, Philippines, Sierra Leone, Thailand, and Yugoslavia undertake project appraisals using shadow prices (how rigorously is not clear). Personal informa- tion suggests that Malaysia should be added to the positive list. We here mention some reasons for the rather poor penetration, because they differ from reasons applicable to the World Bank and other lending agencies. Public projects derive from ministries or other agencies with departmental, sec- toral, or regional interests (not to speak of personal, monetary, or political interests, which also sometimes intrude). Only the central government and the ministries of planning or finance can at best be expected to look to the national interest. Some conflict is inevitable and essential, because project proponents do not welcome central control, which constrains the extent to which they can make projects serve their own purposes. Even financial control has proved difficult in many countries, and soft budget- ing has resulted in the subsidization of many decentralized agencies. So a central system of project appraisal must always be unpopular, and therefore it needs strong political backing at the highest level. This may be lacking for many reasons, among them the fact that at least until very recently many planners and politicians in developing countries sensed that the methods proposed would conflict with the import-substituting industrialization policies that were at the heart of their development philosophy. III. SOME REASONS FOR THE DECLINE OF INTEREST IN SHADOW PRICES AND COST-BENEFIT ANALYSIS IN THE WORLD BANK The decline proclaimed in the title of this section assumes that the general propositions of the previous section are not very far from the truth. If so, it is a shattering indictment, because a shadow price is the marginal effect on social welfare of any quantity change. Shadow prices are fundamental to the evalua- tion of policy changes, not merely investment projects. This is well recognized in 6. We are indebted to J. D. MacArthur for this information. Little and Mirrlees 363 the normative theory of taxes and subsidies, but it also applies to the evaluation of quantitative controls-quotas and rationing. Shadow prices and cost-benefit analysis are inseparable. Sometimes actual prices coincide with their shadow values, as if on the equator in the midday sun. Only then is financial analysis also cost-benefit analysis. For public sector proj- ects and enterprises, a decision has to be made about pricing. A good guideline is to make prices coincide as nearly as possible with their shadows, which requires knowledge of the shadows. Sometimes, however, it is impossible or obviously undesirable to charge at all-for example, in the case of most roads. But the shadows are still required to plan the roads. It is sometimes argued that the thrust of policy should be to get the prices right, and it is suggested that this is an argument for forgetting about, or at least deemphasizing, shadow prices. To the extent that activities are private, and there is no price or profit control, this is feasible and may be desirable. But in the public sector, or whenever there is public regulation, getting the prices right implies knowing the shadow prices. Less Room for Project Analysis? The Growth of Nonproject, Health, and Education Loans Despite the importance of our introductory remarks, cost-benefit analysis and shadow pricing are closely associated with investment projects, in the normal narrow sense of the word project. The decline of interest may thus be related partly to the relative decline of project lending. Structural adjustment loans and other nonproject loans-such as those to development finance corporations- account for an increasing proportion of World Bank lending. There may also be no identifiable project in the case of sectoral adjustment loans. In the case of loans for health and education, no economic rate of return (or present value) is estimated; shadow pricing is still relevant for cost effectiveness, but we do not know whether it is used. In a few countries-such as Brazil-World Bank lend- ing now includes almost no projects to which cost-benefit analysis can be applied. Nevertheless, overall, there are still a great many projects, and, further- more, sectoral adjustment loans also lend themselves to analysis that makes use of shadow prices. The growth of forms of lending other than straight project loans has probably caused a diversion of attention out of proportion to their relative importance. Institutional Reasons The World Bank is a relatively unsuitable and unstable institution for time- consuming and dedicated quality control (via technical and economic appraisal at all stages) of the investments it finances. Ironically, in the 1970s, when the World Bank's Central Projects Department was trying to engineer an improvement in economic analysis, its president, Robert McNamara, was presiding over a huge increase in the volume of lending, which put pressure on country and project staff to get enough projects 364 Project Appraisal and Planning Twenty Years On approved. There is no doubt that this created tension. When the pressure is on to get the money out, it is not surprising that demands for more complex analysis are unwelcome. Worse than this, project analysts would never get promoted if they were honestly compelled to report unfavorably on several projects. Promotion would come from writing good reports that would help steer the project through the processes of approval, and not from improving projects or improving project selection. Economic rates of return (errs) on projects became more and more optimistic, without justification. No one we have spoken to denies that the incentives facing the staff worked toward a deterioration in the quality of project work or asserts that this was a trivial matter; it is well known as the "McNamara effect." It parallels what was hap- pening in the commercial banks. The present situation is not very different. The World Bank is again under pressure to lend, because of the debt crisis. The debt crisis has in turn arisen largely because the massive lending from 1974 to 1982 was matched by invest- ments that turned out to have very low or negative returns. Almost all develop- ing countries raised the share of investment in gross national product (GNP) in this period, some by a great deal, and the increase was predominantly in the public sector. The World Bank was not immune from promoting or supporting some very bad investments, for which its own price projections were partly responsible. The reorganizations of the World Bank in the 1980s have clearly contributed to the decline of cost-benefit analysis. There is now no Central Projects Depart- ment to promote common methods and to exercise quality control. Regionaliza- tion need not, of itself, have had this result. But it has. There is now no fully independent watchdog over the quality of project work. We have the impression of widespread deterioration. This is not only in economic analysis. Fewer resources are apparently devoted to project work. There is, we are told, less continuity and expertise in project teams than earlier on. Technical, financial, and institutional analysis may all have suffered. Obviously economic analysis would not be immune. New Concerns New concerns may have complicated the project analyst's task. There was a new emphasis in the 1970s on projects that would benefit the poor directly. Cost-benefit methods were adapted to make the poverty impact quantifiable and to integrate it into the cost-benefit analysis. That this has been rejected reduces the work load, for it is difficult to trace the beneficiaries of a project and assess their wealth. Yet we gather that for some kinds of projects, the analyst is still expected to do this (only, having done so, the analyst must not apply weights to these differential benefits-it could be argued that this is incurring the costs of the analysis without the benefits). Now the development of women has been added as a project concern. Surely this is an irrelevant consideration in most projects. Then there is the environment. All effects of a project-including on the development of women and the environment-that might seriously affect its Little and Mirrlees 365 value should always be considered and quantified if possible. But there is a cost to insisting that in all cases the appraiser should be seen to have spent consider- able resources on trying to identify kinds of consequences that are unlikely to be of any importance. This is not to say that apparently remote environmental damage may not be large and important. In some areas, the specialist's assess- ment of these magnitudes is much to be desired. Sustainability has come to be used in recent years in connection with projects. This is more of a buzzword-probably derived from the environmental lobby- than a genuine concept. It has no merit. Whether a project is sustainable (forever?-or just for a long time?) has nothing to do with whether it is desir- able. If unsustainability were really regarded as a reason for rejecting a project, there would be no mining, and no industry. The world would be a very primitive place. In defense of sustainability, it will be said that the concept has drawn attention to some often disregarded reasons for project failure. But it seems much better to detail what these reasons are.7 We were told in our discussions with World Bank staff that on occasion the same analyst would give a project a high ERR but would say it was unsustain- able. Of course a reason for this could be that sustainability was irrelevant, as with the possibly rapid extraction of an ore body. But this is not what was meant. What was meant was that the same analyst could give a project a high ERR but could also say that it would probably fail (that is, end up with a low ex post return). This could arise because the project analyst, wearing the hat of ERR analyst (rightly), thinks that it is up to him to take account only of limited kinds of risk-for example, that future prices will be worse for the project than pre- dicted, or that agricultural response has been overestimated. The analyst does not, in writing a project report, assess such risks as that the dictator will be shot and the subsequent economic management will be so muddled that a major slump will be unavoidable, or even that the economic administration will deteri- orate to the extent that the project will be starved of funds for recurrent costs. He is also probably in no position to assess whether the country is overborrow- ing or planning more investment than it can possibly handle efficiently. It appears that the "division" of labor is as it should be. The project appraiser should make it clear which risks have been taken into account and which have not. The appraiser or some other person familiar with the political and macro- economic situation of the country (or maybe the administration of a sector) may comment that no project in that country (or sector) is at present likely to suc- ceed, whatever the calculated ERR. Methodological Defects We have been considering some institutional reasons for the decline of cost- benefit analysis. But cost-benefit analysis, or the World Bank's version of it, may have been its own worst enemy. The methodology may have been unsound. We 7. Sustainability is also described as a "central notion" in the extraordinarily vapid document, Princi- plesfor Project Appraisal, Development Assistance Committee, OECD (1988). 366 Project Appraisal and Planning Twenty Years On discussed this in section I of this paper. There was some sniping from theoretical perfectionists over the years. This may have reduced its acceptability, but proba- bly only among those already disposed to attach a high value to comprehensive countrywide planning. Little-Mirrlees methods have stood up to intensive theo- retical discussion remarkably well, and on balance the large outpouring of the- ory consequent on the original publication of Little and Mirrlees (1969) has confirmed the correctness of the authors' original intuitions. The other way in which the methods could have been self-defeating is that they were too complex, so that the simplifications we have described may have been both inevitable and all to the good. This argument is best examined after reviewing the evidence provided by the World Bank's Operations Evaluation Department (OED). Here, we will remark only that simplification and theoretical soundness are not good bedfellows. IV. EVIDENCE FROM WORLD BANK DATA ON THE VALUE OF COST-BENEFIT ANALYSIS The World Bank's annual review of evaluation results for 1988 gives some information on nearly 2,000 projects approved in the period 1968-80 and subsequently evaluated. The World Bank's OED provides original appraisal ERRS and reestimated economic rates of return (RERRs) on more than 1,000 projects approved in the period 1968-80, together with figures for original and com- pleted costs and expected and actual gestation periods. Gerhard Pohl and Dabrarko Mihaljek (1989) have analyzed these OED data, and this section draws on their work. 8 RERRs are taken from project completion reports usually made by those responsible for the project, which could lead to bias. They will normally allow for known changes in capital cost and any known changes in the mix of inputs and outputs. But actual experience of operation either will be limited to a short period or will be nonexistent. Changes in price predictions also will be fed in. The difference between the ERRS and the RERRS is mainly a difference of expec- tation. The methodology of the estimation of the ERR will not normally be changed. Obviously, the RERR is very far from being a postmortem ERR. To the extent that resources permit, the OED conducts an "audit" on top of the project completion report and ranks projects as "satisfactory" or "unsatisfac- tory". In the case of projects with calculated RERRS, there is in most cases a coincidence of an unsatisfactory rating and a RERR of less than 10 percent. But there are exceptions, in which a project with a high RERR is deemed unsatisfac- tory. In such cases it is clear that the OED does not believe the RERR. But it does not have the resources to make an independent estimate, so the incredible RERR is recorded. 8. Note that in many cases, the so-called economic rates of return are really modified financial rates of return. That is true for power, telecommunications, water supply (and possibly some other subsectors). It is important to remember this limitation in the following discussion. Little and Mirrlees 367 Over the years 1968-80, there was a remarkable and continuous rise in appraisal ERRS. They averaged 17 percent in 1968 and 29 percent in 1980. There was no similar trend in average RERRS for projects approved in that period, which, apart from the years 1970 and 1980, were in the range 13 to 17 percent.9 As a result, a large gap between ERRS and RERRS has arisen and has excited some attention. It casts some doubt on the objectivity and stability of appraisal, there being no reason to think that investment possibilities were rapidly improving. The rise in appraisal ERRS was common to all regions and sectors but was most marked in Asia and Africa; and sectorally in agriculture and rural development and in transport and tourism. We do not know to what extent the apparent euphoria was genuinely felt, or to what extent it was due to the McNamara effect. The RERRS of 1968-80 were calculated over the period from 1974 to 1988, with an average lag between approval and reappraisal of nine years. One would expect the RERRS to be importantly affected by the year of reappraisal, particu- larly by the euphoria that may have continued until mid-1982. After that, the reappraisers' expectations of project performance would surely have been scaled down, in view of the difficulties in which most developing countries found themselves, and all sectors would have been affected. Almost no projects approved before 1971 were evaluated in 1983 or later: Percentage of projects Year of approval evaluated after 1 982 1971 16 1972 23 1973 35 1974 66 1975 86 1976 90 1977 92 1978 96 1979 98 1980 98 All other things being equal, one would expect, with scaled down expecta- tions and with the rising proportion of reappraisals done after 1982, falling RERRS, especially for the projects approved from 1973 to 1976. The data do show low RERRS (13 percent) for projects approved between 1973 and 1976, but they then rise (to about 16 percent) for the projects with 1977, 1978, and 1979 vintages. We leave the reader with this puzzle. We do not have figures for appraisal ERRS in the 1980s. 9. We have no explanation for the high average RERR of 21 percent in 1970. We distrust the high figure for 1980 (24 percent), because relatively few projects approved in that year have been evaluated, and because there is a tendency for good projects to have short evaluation lags. We do not know how negative rates of return that could approach infinity are reckoned. From Pohl and Mihaljek (1989) it appears they may all be counted as minus S percent. In view of this arbitrariness, it would be useful to know the median rates. 368 ProjectAppraisal and Planning Twenty Years On The proportion of unsatisfactory projects is of interest, as are average RERRS. As we have seen, an unsatisfactory rating usually coincides with a RERR of less than 10 percent, but not invariably so. The average percentage of unsatisfactory operations among those approved in the period 1968-73 was 15 percent. In the period 1974-79, it was 23 percent. This deterioration occurred in all regions except Europe, the Middle East, and North Africa. It would be of interest to know whether this can be accounted for by a shift to more difficult sectors (agriculture and rural development) and regions (Africa), or whether there was some general deterioration in project identification and appraisal, again possibly resulting from the McNamara effect. We turn to Pohl and Mihaljek's (1989) analysis. They regress RERRS on a number of independent variables, including ERRS. Except for regional and sec- toral dummies, few of these variables are significant. The simplest equation of all, regressing RERR on ERR alone, gives a coefficient of 0.44 for ERR, which is highly significant, explaining 19 percent of the variance. Sectoral dummies for transport and urban projects are significant and raise the yield by 7 to 9 percent- age points compared with agriculture. Regional dummies are also significant. Asian projects are better than Europe, the Middle East, North Africa, and Latin America, whereas Sub-Saharan Africa trails badly. But of course project appraisers know which region and what sector they are in. The coefficient for the appraisal ERR is hardly affected. The essential findings of the analysis, relevant to assessing the benefits of cost- benefit analysis, are that there was an unjustified rise in revealed optimism about project returns in the 1970s; and that the ERR is a. highly significant explanatory variable for RERRS, with a coefficient of 0.44, but it explains only about 20 percent of the variance (R2 = 0.19). The main conclusion drawn by Pohl and Mihaljek (1989, p. 32) is that "cost- benefit and rate of return calculations seem to be most useful in the case of large and capital-intensive investment projects. A sufficiently high minimum rate of return criterion, say 10 percent, will help to screen out large and capital- intensive projects with potentially low rates of return. But further refinements in the methodology, such as allowing for distorted prices resulting from overvalued exchange rates, while intellectually appealing, do not seem to make much differ- ence in practice." This conclusion is far from clear. Taken literally, it seems to suggest that all shadow pricing is futile and that even financial rates of return should be calcu- lated only for large, capital-intensive projects. Worse, there is nothing in the analysis reported above that has any bearing on the conclusions. There was no analysis relevant to the question of whether refinement of method would improve ERRS (that is, make them better predictors of RERRS), or of whether such improvement would justify its cost. Pohl and Mihaljek (1989, p. 32) do say that "this was also brought out by an analysis of industrial projects for which both financial and economic rates of return were available. Appraisal of financial rates of return were just as good a Little and Mirrlees 369 predictor of ex post economic rates of return as ex ante economic rates of return. The adjustments for price distortions seemed to make little difference, at least for projects that had been accepted for World Bank financing." For this a statement is hardly enough: supporting evidence needs to be closely considered. It is hard to believe that one can do as well aiming at the wrong target. If so, it seems to be unnecessary to take thought at all. Pohl and Mihal- jek's final twelve words should be noted. For industrial projects, the main point of shadow pricing is to spot projects in which financial returns are misleading, say, because of protection. If some of these are successfully rejected, there will be a bias toward some coincidence of financial returns and ERRS. Taking the figures for pre- and post-project evaluations at face value, it is important to appreciate what the apparently low incidence of association between them is saying. A linear regression of the RERR on the ERR is not appropriate, or at least it is not to be interpreted in the way one might usually interpret a regression coefficient. To interpret the statistical results reported by Pohl and Mihaliek (1989) we need a properly specified underlying model. A simple plausible model is that the ex ante and ex post estimates of rates of returns are attempts to assess the true value of the project, both subject to additive errors (multiplicative would be better, but Pohl and Mihaljek have used additive, and we are interpreting their results). We can interpret the regression as revealing the value that is common to the two observations. Write C for the common element of the two evaluations. It is subject to errors u and v, indepen- dent of one another, and of C. We then have ERR = C + U; RERR = C + v. The regression coefficient of the RERR on the ERR is COV(RERR, ERR) / var(ERR), where cov is covariance and var is variance. With this additive model, that is equal to var(C)/[var(C) + var(u)]. The regression coefficient therefore shows the proportion of total variability that is due to the common element. At least that would be correct if the error, u, were independent of the common element, C. It is not. The common element is made up of the true value of the project and any source of error common to the two appraisals. Because the worst observations in the first round, having rates of return below 10 percent, are (largely) rejected, there is a negative covariance between the common value and the error u (assuming that the mean return in all projects assessed is above the cutoff value). The regression coefficient is then var(C)/ [var(C) + var(u) + cov(C, u)]. This means that the regression coefficient actually underestimates the ratio of the common variance to the error variance. We are interested in the variance of the true value of the project and the extent to which it seems to be picked up by these observations. We cannot directly tell how much of the common element is the true value and how much is the common sources of error. Because many sources of error (such as utilization estimates and prices) changed greatly between the two evaluations, we believe that the true value is a rather substantial part of the common value. All in all a regression coefficient of 0.44 for ERR, when regressing RERR on ERR alone in Pohl and Mihaljek's simplest equation, suggests that the variance of the true 370 ProjectAppraisal and Planning Twenty Years On value of the project may be nearly as large as the variance of error in an evaluation. The "proportion of the variance explained" in the regression equation is the product of that regression coefficient and the corresponding one for RERRS. Even allowing for the selection bias, it appears that errors in the reevaluation may not be much less than errors in the initial evaluation. It should be appreci- ated that the low proportion of variance explained is in fact a compound of the error magnitude at the two evaluations. It is not surprising that it is as low as 20 percent. That is not in itself evidence of a particularly low common element. These errors in the project appraisals are substantial, but they are not immense. Taken literally, they are quite good news for the users of project appraisals. We show in the appendix on uncertainty that the value of project appraisal can be measured roughly by taking 10 percent of the standard devia- tion of true project values, times the ratio of that standard deviation to the standard deviation of measurement errors. If the two standard deviations are of about the same magnitude, as seems to be implied by this data, the value of project appraisal is truly substantial. Several considerations suggest that the data may considerably understate the precision with which projects can be appraised. First, we should not pay exces- sive attention to the average divergence. Given the World Bank's selection methods, it makes no difference if a project yielding 20 percent was appraised at 40 percent. Excessive optimism is likely to be important only if it pushes a bad project over the 10 percent barrier. One would hope that around that level, appraisals would have been done with greater care. We believe that the associa- tion between ERRS and RERRS is greater around these low appraisal levels. More generally, the simple model we have used to interpret the data gives too much weight to outliers. The data suggest some further observations, comparing the World Bank's experience of evaluations with that of the countries for which they are done. Improving the World Bank's portfolio may not improve that of the country. There is the problem of fungibility between project funds and the general bud- get. The value of project aid has often been called into question over the past forty years. Despite this long history, we do not know of any empirical work devoted to ascertaining first what difference is made to a country's investment by the involvement of donors and lending agencies in project preparation and selection. We do not know how many projects are effectively invented by the World Bank (and whether these were particularly successful or unsuccessful); we do not know what happened to projects the World Bank rejected; and we do not know whether, if carried out, the projects had high or low returns. It has been repeated ad nauseam that economic appraisal at a late stage very rarely stops a project. It must be applied early on to stop work on the project or to effect improvements in the design of a project that will finally go ahead. If one asks whether such improvements are made-and we have asked this question of a few people in the World Bank, in ODA, and in India's Project Appraisal Divi- Littleand Mirrlees 371 sion of the Planning Commission-one is always told that they are. But there is no way of assessing the magnitude of this benefit of the appraisal process, which is already incorporated in reported ERRS and helps the borrower even if fun- gibility were complete. Thus we do not know to what extent the high average RERRS on World Bank projects are achieved because the World Bank improves projects or invents good projects, and to what extent they come simply from skimming the cream from the country's own proposals (nor do we know whether the World Bank helps the country to achieve a higher-yielding portfolio by convincing it not to go ahead with unpromising ventures the World Bank rejects). There is a lot we do not know. There is no doubt, however, that the World Bank's average RERR is high for the projects of the 1970s-about 16 percent. Of course, as we have pointed out, RERRS are not true ex post rates, which may be significantly lower. Even if the true return is several percentage points lower, it would still be above the real foreign borrowing rate (an ERR calculated using border prices and conversion factors for nontraded goods will be a good approx- imation to the foreign exchange yield). This is more than one can say for total investment in many, if not most, developing countries. Because almost all did use their borrowing to support or raise the ratio of investment to GNP, there would be no debt crisis if total investment had yielded even half that of the World Bank-financed investments. India is a country with no debt crisis because it did not borrow commercially until recent years. India engineered no foreign-debt-supported public investment boom, as did many other countries in which the rise in investment was so rapid that it can only have been hastily planned and executed. But, in work in prog- ress, Joshi and Little (1990) have calculated nevertheless that the return on public sector investment in India over the period 1974-75 to 1984-85 was no more than 6 percent.10 The average RERR (both weighted by size of project and unweighted) over roughly the same period for projects in India was very high, about 23 percent (negative returns arbitrarily put at minus 5 percent)." This makes India one of the great stars for World Bank-financed investments, although it is not only Joshi and Little who find the returns to investment in India to be low. It is also notable that the apparent high performance of World Bank investments is not attributable to the World Bank having avoided agricul- 10. Joshi and Little's work is a draft chapter, entitled "Macroeconomic Management, Investment and Growth in India, 1960-61 to 1984-85," of a contribution to the World Bank research project on macroeconomic policy and growth. The calculation is made at market, not shadow, prices. Recalculation at shadow prices would raise the yield insofar as the benefits from public sector investments are not reflected in public sector value added. This may be considerable. An offset, however, is that a good deal of public sector industrial value added would be even less than it is if calculated at shadow prices. It should also be noted that similar calculations for total manufacturing investment (public and private) show returns of about 10 percent. 11. The averages are taken over a total of sixty-eight Indian projects evaluated by the OED for the period 1974-87 for which an RERR was calculated. Lipton and Toye (1990) also found an average RERR of about 23 percent on thirty-three projects for which the OED provided them with evaluation documents. 372 ProjectAppraisal and Planning Twenty Years On ture, which has low returns in other countries. More than half the projects are agricultural, and even in value terms they amount to about a third of World Bank-financed investment in India, with average returns about as high as in other sectors. RERRS may be grossly overestimated, the World Bank may be very good at creaming off projects, or the World Bank is very good at creating or improving projects. It is clearly of great importance for World Bank policy to know how much weight to give to these different possibilities. We cannot tell. But if the World Bank is good at discovering and improving projects (and rejecting poor projects in countries with limited alternative sources of finance), then there is a strong case for the World Bank to do all it can to teach countries to improve their selection of investments, or to make its influence felt over a greater value of investments than it actually helps to finance (this should be possible in the case of sectoral loans). But whatever the value of the World Bank's involvement in projects, including economic appraisal, it remains true that cost-benefit analysis would be much more effective if done in the country. It can then be applied to all large projects, and there is no problem of fungibility. We have very briefly reviewed above what little we know of the use of cost- benefit analysis for public sector investments in developing countries. In most it is nonexistent, and in only a very few is it better than rudimentary. There is, of course, more to planning public sector investment than project appraisal (in Little and Mirrlees 1974 we did try to show how the two can, and should, be integrated). From this wider perspective also it seems that in only a minority of developing countries is there any coherent and rational control over ministries and other public agencies, with the result that far more projects get started than can be financed without interruption. The World Bank does not seem to have played a leading role in the promotion of rational public investment planning. Yet there would appear to be room for the World Bank to have a considerable beneficial influence. It is surely an essential part of continuous (sustainable) structural adjustment that public sec- tor investments be well chosen. This applies both at the most central level, where intersectoral choices are required, and at sectoral levels, whether or not sectoral adjustment loans are being made. For the World Bank to have such an influence it must have-and be under- stood to have-the necessary skills. These include a thorough understanding of cost-benefit analysis and shadow pricing. They also include experience of the control of public expenditure. The declarations of ignorance in this section suggest the need for research on how far the World Bank's activities improve the productivity of investment in the borrowing countries. Because examination of rejected project proposals and what happened to them is essential, and very many projects need to be exam- ined, a deeper analysis than the OED has the resources for is warranted. A pilot study of one cooperative country (if such can be found) could throw much light on the possible value of a wide-ranging study. Littleand Mirrlees 373 V. THE COSTS OF COST-BENEFIT ANALYSIS We have discussed the benefits of cost-benefit analysis in general, and of shadow pricing in particular, at some length. What are the costs? No one suggests that industrial or other "commercial" projects should be approved without even a financial analysis. Nor does anyone suggest that any old road should be approved without a traditional estimate of its benefits in terms of expected vehicle cost savings, time savings, and traffic generation; nor that an irrigation or settlement project be approved without some estimate of the value of the extra crops expected to be produced. What more is required for an economic rate of return? For traded goods it is now generally accepted that border prices are to be used. But this is little or no extra work, because the future international prices of such goods will be needed in any case for a financial analysis. Otherwise present practice merely involves multiplying some of the rows by the standard conversion factor (SCF), adding up, and recalculating the ERR and present value-a few minutes' work at most for the project analyst on a personal computer. The country economist supplies the SCF. This could take some time if conscientiously done, but, knowing how often it has been "calculated" as 0.8, the economist may get away with this and just write a memo saying "SCF = 0.8." The benefits, as we have argued above and argue again in the appendix, are immensely greater than the costs of applying these procedures. They would justify much more extensive work, both on national parameters and on the specifics of the individual project. This cost-benefit analysis at least seems to give a very clear answer. It is less easy to generalize about the value of refining the appraisal (as opposed to the use of simplifications and often justifiable shortcuts) and about the loss from cutting short an analysis when important aspects have not been adequately allowed for. VI. THE TRADE-OFF BETWEEN COMPLEXITY AND RELIABILITY Here, we comment briefly on the value of refining calculations at both the national and the sectoral or project level. National Parameters How sophisticated should be the analysis that goes into the calculation of national parameters, such as discount rates and shadow wage rates? There is no very short answer. Most of the national parameters calculated by the World Bank are out of date. The IDB and the Development and Project Planning Cen- tre, Bradford University, have calculated recent sets for a number of countries, whether for their own use or at the country's request (the introduction of Little and Mirrlees's methods to China with a set of national parameters is still a victim of Tiananmen Square). Where do the priorities lie? Given the World Bank's policy of using a single accounting rate of interest, unvarying by country 374 ProjectAppraisal and Planning Twenty Years On and time, there may seem to be little point in estimating this parameter. (The rate is almost always 10 percent, but 12 percent is used for a few cases.) Yet, in theory, variations should be one of the main responses to changing external or exogenous conditions. There is a case for more serious attention to these issues. The shadow wage rate, or the conversion factor from which it is derived, is a parameter that can vary substantially by region, and for which local information specific to the project should influence the figure used. Here, we consider its estimation at the national level. Employment is one way in which the benefits of a public sector project "leak" into the private sector. This arises when the increased demand for labor raises wages or when the wage paid is above the supply price of labor. Part of the resultant increased private income is a cost to the economy if, as we have insisted, it is worth less than public income. Given this, the calculation of a shadow wage (the real cost of employing someone) becomes very complicated. It is hard to estimate not only the loss of output elsewhere but also the increase in private income. Regarding modern sector urban wages, there is a consensus that the shadow wage is probably not very different from the wage paid, despite the presence of a wage gap between the "organized" and "informal" sectors. Given that modern sector projects are also capital-intensive, it is reasonable to assume that the actual wage is as good an approximation of the shadow wage as can be found. Research in recent years has tended to suggest that rural labor markets, and informal sector markets generally, are active and that the wage paid is probably close to the marginal product. With some large-scale agricultural projects, how- ever, wages in the area may well be raised, with some loss of producers' surplus and some gain in private incomes. Some agricultural projects have foundered as a result of lack of knowledge of, or enquiry about, labor supply conditions. Such effects as those suggested above certainly need to be investigated, whether or not they are formally used to quantify a shadow wage rate (SWR). Sectoral Costs and Benefits Specific conversion factors for the use of important nontraded goods were regarded as an essential part of the Little-Mirrlees or Squire-van der Tak meth- odologies. But we also need to consider the numeraire value of the outputs of projects in these sectors. We take power as an example, mention transport and construction more briefly, and must bypass the important subject of irrigation for lack of space. Public utilities (defined as public enterprises for whose output little or no charge is made) present special problems. Power. Perhaps the most important nontraded good is power. The World Bank has power projects in most countries, and power is an input into all other projects-occasionally a very large input. We have seen that the output of power projects was never normally valued in Little and Mirrlees 375 the Squire-van der Tak manner. This was reasonable. It would be absurd to calculate an ERR for every power station financed. The ideal procedure is to obtain agreement with the country on an appropriate tariff system. Normally this will be based on long-run marginal cost (LRMC) at accounting prices. When the tariff is in place, demand at those prices must be estimated and should be satisfied (obviously by the least-social-cost mode). Sectoral loans can then be made, and there is no need to calculate a power conversion factor for other projects. There is an equilibrium, and the price is already "right." Unfor- tunately, things get more complex when there is serious excess supply or demand at a price equal to LRMC (which still needs to be calculated). There may then be a good case for charging less or more than LRMC, as the accounting price diverges from LRMC for several-even many-years. When there is excess demand, as in India, the value of power is its marginal product, not its marginal cost. However, the stipulation of serious excess supply or demand at a price equal to LRMC at least makes it clear whether any investment in power is justified. Construction and transport. A construction conversion factor may be desir- able on the ground that construction enters importantly into many projects, and that it may be very labor-intensive. If the latter is the case, a lot in turn depends on whether SWRS should be used. This could make a construction conversion factor significantly different from (lower than) the SCF. Construction is of course virtually the only input into road programs, which are a large component of World Bank lending in several countries. Sectoral lending seems to make good sense for roads, where a component of an ongoing program that can be separately and validly assessed may be impossible to iden- tify. As with power, the World Bank may then want to assure itself that road programs are designed and assessed according to its own guidelines (noting that roads are public utilities to the extent that there is little or no cost recovery; see below). It would then and only then be willing to lend on a sectoral, and not a project, basis. Transport costs also enter into almost all projects. Although transport is nontraded, its use may have quite large foreign exchange implications, making a special conversion factor desirable. The extent to which multiple conversion factors should be calculated by the World Bank depends on the country (how distorted are the relative prices of nontraded goods?), on the sectoral composition and size of the World Bank's lending programs, and on the extent to which the country itself is willing to adopt good cost-benefit analysis on a sectoral (even if not a countrywide) basis. If a set of consistent conversion factors calculated by input-output procedures is not available, the project analyst can often make a good guess at one that is particularly relevant, by back-of-the-envelope methods. Public utilities. Public utilities are special. Their costs are (generally) borne by 376 Project Appraisal and Planning Twenty Years On the public sector, whereas a large part of the benefit accrues to the private sector.12 A transfer to the private sector occurs when, as often happens with electricity, gas, and water supply, there is no full cost recovery. We have argued that public income in most developing economies is worth more than private income because of the administrative costs of taxation and the absence of anything even approximating a general lump-sum subsidy, so that taxation is necessarily distorting at the margin in the relevant sense. These considerations are sometimes strengthened when most saving comes through the public budget. The argument for assigning a greater weight to public than private income holds even if no distributional weights are acknowledged. In brief, increasing the public sector deficit is costly, and projects that do increase it should be penalized. As we have already mentioned, differential weighting of public and private income is ignored in cost-benefit analysis done by the World Bank and other agencies. We would urge that the fiscal effect of all projects should be estimated and included in appraisal reports. As far as possible, this should include indirect effects, such as changes in tax payments by recipients of additional income. If that is done, it will be clear how the present value of a project can depend on the way it is financed. Public utility pricing policy and project appraisal results are inseparable. If, as we suspect, public income can have a much greater value than private income, the impact of pricing on the value of the project could be very substantial. Estimation of the administrative and distortional costs of taxation in developing countries is a matter on which more research would be valuable. VII. CONCLUSIONS Social cost-benefit analysis is not as widely, as well, or as effectively practiced as its expected net value might lead one to hope and expect. We have covered, or at least touched on, many aspects of its theory, practice, and effects. We have claimed that much in the rules we collected and prescribed in 1969 and 1974 has survived analytical scrutiny, that these procedures are capable of being used effectively, and that many important aspects of them have been neglected by project evaluators. We have found that the extent to which they are used and have real influence is not great, even in the World Bank. We have examined some of the data appearing to show considerable randomness in the evaluations performed within the World Bank. We went on to consider the bearing of the very considerable uncertainty that has been shown to attend project appraisal upon the value of the appraisal activity itself. We argue that the value is indeed probably diminished by that uncertainty but is nevertheless very large. In aggregate, much of the investment in developing countries has had very low returns. This is evident in the low growth of many of these countries during the 1980s. Project appraisal is an essential part of the business of avoiding these mistakes in the future. Good project appraisal is done by people with their own 12. This section owes much to Squire (1989). Little and Mirrlees 377 incentives, within organizations that wittingly or not set these incentives. Both environments of project appraisal, the intellectual and the political- organizational, are keys to the quality of selection overall. This needs to be most seriously considered by those who manage and create these environments. APPENDIX: UNCERTAINTY AND PROJECT APPRAISAL The Value of Appraisal One can regard project appraisal as a reduction in uncertainty, that is to say, acquisition of information. The value of information has been studied in the literature.13 Here we apply it to the making of decisions. The process of appraisal is selection among several possibilities, each of uncertain value. Sup- pose, for example, that there are two projects. Their true values are x and y. Suppose, for definiteness, that x > y. Appraisal will yield apparent values x + A and y + B, where the random variables A and B are the errors that remain after appraisal. Before appraisal, uncertainty is even greater: their values seem to be x + A + C and y + B + D, where C and D are further error terms. Assume that all four error terms have zero mean; that amounts to saying that there is no identifiable bias in the relative evaluations for the two projects. Without appraisal, the larger of x + A + C and y + B + D determines the choice. The wrong choice (y rather than x) is made if and only if A + C - B - D < y - x: that is, writing M for the random variable (A - B) and N for the random variable (C - D), the wrong choice is made when (A-1) M + N < y-x Similarly, if project appraisal is used, the wrong choice is made when (A-2) M < y - x M and N-like A, B, C, and D-have zero means. For simplicity, assume that M and N are independent random variables, with single-peaked density functions (for example, normal or lognormal random variables). The chance of the wrong choice is less when there is project appraisal because M + N is a more dispersed random variable than M, and y - x is negative-that is, less than the mean of both M and N. We want to estimate the magnitude of the reduced chance of error. Using an expected-utility representation of the value of projects when there is uncertainty, we can write u(x) and u(y) for the utility of the two projects. Then the value of project appraisal is u(x) - u(y) times the reduction in the probability of error. This reduction in the probability of error is (A-3) P(M + N < y-x)-P(M < y-x) 13. See particularly Gould (1974), which studies the influence of greater uncertainty about what is initially unknown on the value of finding out about it. The question of assessing the value of information quantitatively is not considered, nor is the influence of greater ambient uncertainty on the value of information. In the analysis below, we note particularly the importance of this last issue. 378 Project Appraisal and Planning Twenty Years On where P denotes the probability of the event described. This expression can be written using distribution functions F and G for the random variables M and N; so the second term in equation A-3 is F(y - x), and the first is (A-4) P(M y, because we are on the left-hand side of the probability distribution. The expression f' is greatest when x - y is neither close to zero nor very large and when noise-the degree of general uncertainty not removed by appraisal-is small. Reasonably enough, appraisal is not very valuable when the difference between the projects is small, nor is it valuable when the difference is very great (relative to noise). For given projects, the value of the appraisal is also smaller the greater is noise; greater noise means that f increases less steeply to its maximum at y - x = 0. To assess the magnitude of the expected value of doing project appraisal, we can make two more special assumptions: that u(x) is simply x, and that noise is distributed normally with standard deviation T. Furthermore, we can calculate the maximum value of project appraisal (as y - x varies). This is an upper estimate of the value, but the value is fairly close to its maximum over a substan- tial range of values. It is easiest to calculate this maximum for the low a2 approximation, equation A-7. The maximum value of equation A-7 is U2 (A-8) 0.147 Little and Mirrlees 379 where the constant is 1/[.J27e] = 0.14676. This actually overstates the maxi- mum. When u2 is not small, some mathematical manipulation shows that the value of appraisal is (A-9) (x - Y)[F( y )-F(y - x) where r is arIT, the ratio of the two standard deviations. Numerical calculations of this formula yield the following table of maximum appraisal values as c/i varies: Maximum appraisal /rI vzaluela 0.5 0.069 1.0 0.119 1.5 0.149 2.0 0.164 The maximum occurs at a value of y - x in the middle of the likely range of true values. The limited evidence available, discussed in section III, suggested that the ratio of the standard deviations might be about unity for the class of appraisals we are interested in, although with competent analysis it ought to be substan- tially greater than that. This yields a simple yardstick for the value of appraisal as something like 10 percent of its standard deviation-a very substantial amount considering that, even for small investment decisions, the standard deviation for the present value of the project would usually be many millions of dollars. Allowance for Uncertainty In Little and Mirrlees (1974), we provided a simple formula for estimating the impact of uncertainty on the value of a project: (A-10) V = E(X)-A E((Y) where X is the (random) social profit value of the project, Y the (random) level of national income, E( ) denotes the expected value of the indicated variable, cov( ) the covariance of the two variables, and A is the coefficient of relative risk aversion. The derivation (Little and Mirrlees 1974, section 15.8, p. 331) is only indicated. Although something like the formula is well known in decision theory and the theory of asset values, it may be useful to provide a clearer argument. It will be shown that the project should be undertaken if V in equa- tion A-1 0 is positive (to a first approximation). The formula is based on a simplified view of a project and an economy. The value of national income Y to the economy is taken to be E[u(Y)] for a utility function u. A project equivalent to an uncertain change X in national income is worth doing if 380 Project Appraisal and Planning Twenty Years On (A-11) v = E[u(Y + X)] - E[u(Y)] > 0 Assume that X is going to be small relative to Y, whatever happens. Granted that, we can approximate both terms in equation A-11 by a Taylor expansion around the expected value of Y, E( Y). We have u(Y + X) - u[E(Y)] + u'[E(Y)] [Y-E(Y) + X] 1 (A-12) + - u"[E(Y)] [Y - E(Y) + X]2 2 u(Y) - u[E(Y)] [Y - E(Y)] + I u"[E(Y)] [Y- E(y)]2 Taking expectations, and writing ui and a2y for the variances of X and Y, we find that E[u(Y + X)] u[E(Y)] + u'[E(Y)]E(X) (A-13) + - u'[E(Y)]{a2y+ 2cov(Y,X) + d+ [E(X)]2} E[u(Y)] u[E(Y)] + - u"[E(Y)]ay 2 After subtraction, we obtain an expression for the increase in expected utility from introducing the project: (A-14) v = u'[E(Y)]E(X) + 2 u"[E(Y)]tax+ [E(X)]2 + 2cov(Y,X)} These approximations neglect further terms, among which are, for example, a term 1/2u"' E(X)a2y. This could well be larger in magnitude than the terms 1 /2u"{[E(X)]2 + Ui}, included in equation A-13. But all of these will be small relative to the terms in E(X) and cov(Y, X). The main part of the approximation therefore reduces to (A-15) v u'[E(Y)]E(X) + u"[E(Y)]cov(Y,X) The coefficient of relative risk aversion is defined for any expected utility level, y, as (A-16) A(y) - Yu' (y) Here, we define A more particularly as the value of the coefficient at E(Y). Then, equation A-1 6 may be written (A-17) v = u'lE(Y)\[E(X) - Acov (YX) The right-hand side of equation A-17 is u'[E(Y)]V, where V is the expression A-10. Therefore the criterion, do the project if v > 0, is approximately the same as the criterion, do it if V > 0. This shows the validity of the formula given in equation A-10. Little and Mirrlees 381 The correlation coefficient r between X and Y is defined by (A-18) cov(XY) = r(axoy) where ax, ay are, by our earlier definition, the standard deviations of X and Y. Therefore, the expression in equation A-14 is also equivalent to the more intu- itive form given in equation 1 of section 1, (A-19) V x u'[E(Y)] E(X) 1 - Ar 'x 'Y) } E(X) E(Y) The expression in braces is a multiplier, applied to the expected value of net profit, to adjust for uncertainty. REFERENCES Balassa, Bela. 1976. "The 'Effects Method' of Project Evaluation." Oxford Bulletin of Economics and Statistics 38, no. 4: 219-31. Ballard, C. L., John B. Shoven, and John Whalley. 1985. "General Equilibrium Com- putations of the Marginal Welfare Costs of Taxes in the United States." American Economic Review 75: 128-38. Bhagwati, J. N., and T. N. Srinivasan. 1978. "Shadow Prices for Project Selection in the Presence of Distortions: Effective Rates of Protection and Domestic Resource Costs." Journal of Political Economy 86, no. 1: 97-116. Corden, W. Max. 1974. Trade Policy and Economic Welfare. Oxford, England: Claren- don Press. Development Assistance Committee, OECD. 1987. Review of Project Appraisal Criteria and Procedures (87)11. Paris: OECD. Development Assistance Committee, OECD. 1988. Principles for Project Appraisal. Paris: OECD. Diamond, Peter A., and James A. Mirrlees. 1976. "Private Constant Returns and Public Shadow Prices." Review of Economic Studies 43: 41-47. Dreze, Jean, and Nicholas H. Stern. 1987. "The Theory of Cost-Benefit Analysis." In A. Auerbach and M. Feldstein, eds., Handbook of Public Economics. Amsterdam: North-Holland. Gould, J. P. 1974. "Risk, Stochastic Preference, and the Value of Information." Journal of Economic Theory 8: 64-84. Harberger, Arnold C. 1972. Project Evaluation: Collected Papers. London: Macmillan. Henry, C. 1974. "Option Values in the Economics of Irreplaceable Assets." Review of Economic Studies 41(S): 89-104. Joshi, V., and 1. M. D. Little. 1990. "Macroeconomic Management, Investment, and Growth in India 1960-61 to 1984-85." Oxford University. Processed. Lal, Deepak. 1980. Prices for Planning: Towards the Reform of Indian Planning. Lon- don: Heinemann. Layard, Richard, ed. 1972. Cost-Benefit Analysis: Selected Readings. Harmondsworth, England: Penguin. Lipton, M., and J. Toye. 1990. Does Aid Work in India? A Country Study of the Impact of Official Development Assistance. London: Routledge. 382 ProjectAppraisaland Planning Twenty Years On Little, I. M. D., and James A. Mirrlees. 1969. Manual of Industrial Project Analysis, vol. II. Paris: OECD Development Centre. .1974. Project Appraisal and Planning. London: Heinemann. ODA (Overseas Development Administration). 1988. Appraisal of Projects in Developing Countries, 3d ed. London: HMSO. Pohl, Gerhard, and Dabrarko Mihaljek. 1989. "Project Evaluation in Practice: Uncer- tainty at the World Bank." Economic Advisory Staff, World Bank. Washington, D.C. Processed. Scott, M. F. G., J. D. MacArthur, and D. M. G. Newbery. 1976. Project Appraisal in Practice. London: Heinemann. Sieper, E. 1981. "The Structure of General Equilibrium Shadow Pricing Rules for a Tax- Distorted Economy." Unpublished paper, Department of Economics, Australian National University, Canberra. Processed. Squire, Lyn, and H. G. van der Tak. 1975. Economic Analysis of Projects. Baltimore, Md.: Johns Hopkins University Press. Squire, Lyn. 1989. "Project Evaluation in Theory and Practice." In Hollis Chenery and T. N. Srinivasan, eds., Handbook of Development Economics, vol. II. Amsterdam, Netherlands: North-Holland. Srinivasan, T. N. 1982. "General Equilibrium Theory, Project Evaluation, and Eco- nomic Development." In M. Gersovitz and others, eds., The Theory and Experience of Economic Development. London: George Allen and Unwin. UNIDO (U.N. Industrial Development Organization). 1972. Guidelines for Project Eval- uation. Project Formulation and Evaluation Series, no. 2. New York: United Nations. PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1990 COMMENT ON "PROJECT APPRAISAL AND PLANNING TWENTY YEARS ON," BY LITTLE AND MIRRLEES Lyn Squire Ian Little and Jim Mirrlees have done an excellent job in putting on paper what I suspect many World Bank staff have already acknowledged at least privately. That is: although the Little-Mirrlees approach to project analysis remains theo- retically intact, the World Bank's commitment to its application has clearly declined. Some may question the extent to which the World Bank ever adopted the approach in practice; others may claim that we are doing a much better job than Little and Mirrlees give us credit for. But for the purposes of my comments I would like to accept their assessment as broadly accurate and ask what we can do to improve matters. The starting point is the experience of the mid- and late 1970s, when the World Bank was seriously trying to implement the approach. At that time much was made of the idea that it was important not to reduce cost-benefit analysis to a set of cookbook rules. Instead, it was argued, one should set out the broad approach and then rely on the analyst to mold it to the specific circumstances of each project. To be successful this approach to implementation requires two ingredients: enough well-qualified analysts to conduct the appraisals and an environment that encourages such effort. The World Bank had the first but not the second. Given this circumstance, it may make sense to reconsider the use of some simple rules to at least ensure a minimally acceptable standard of appraisal. In my comments I would like to illustrate this general idea with three examples and in doing so touch on some of the issues Little and Mirrlees have raised. APPRAISAL OPTIMISM The first rule concerns appraisal optimism. A few years ago one of the World Bank's senior vice presidents was so upset with his staff's tendency to make overly optimistic projections of disbursements that he decreed that all future disbursement profiles would be based on historical norms. Of course this meant Lyn Squire is chief of the Country Operations Division for Mexico in the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 383 384 Comment that for any given project the disbursement profile was never going to be exactly right, but it did inject a very strong dose of realism into disbursement forecasts. The same principle could be applied to project costs by insisting that cost projections be more closely tied to historical experience. An even simpler rule would be to require that the cost profile and the disbursement profile be consis- tent. This is not unreasonable, because they are basically the same thing looked at in two different ways. Such a rule would go a long way toward eliminating the gap that has arisen between rates of return calculated at the time of appraisal and those calculated at project completion. SHADOW PRICING A second idea is to insist on the use of shadow pricing, but only when it is critical. One simple rule to implement this idea would be to say that staff must present separately the net present value of the stream of tradable costs and benefits and the net present value of the stream of nontradable costs and bene- fits. Tradables would be valued at border prices and nontradables at domestic prices. The objective then would be to test the sensitivity of the net present value of the whole project to the choice of the standard conversion factor. If the project remains profitable for a wide range of standard conversion factors, no further effort would be put into shadow pricing. This reflects the fact that we are not interested in the accuracy of the estimated rate of return; all we need to know is that for plausible assumptions the project will pass the accept- reject decision. This is why regressions of project completion rates of return on project appraisal rates of return are not very interesting-they focus on the accuracy of estimation and not on whether we have been successful in selecting good projects. But if the accept-reject decision does hinge on the choice of the standard conversion factor, then additional shadow pricing would be justified and should be insisted on. It would be appropriate, for example, to require that in such circumstances the bundle of nontradables be disaggregated and individual con- version factors be estimated for the more important items. PROJECT DESIGN The third rule has to do with project design. Because appraisal usually occurs late in the project cycle, most of the key decisions on physical design will have been made before project appraisal. But this is not true for the project's financial design. For example, if a project results in a net cost to the public sector but a net benefit overall, the analyst should be required to examine the scope for direct or indirect cost recovery. And this can be done quite late in the project cycle. This also relates to the question of using different values for private and public income. I am not suggesting that we should estimate such values on a regular basis, but-and this is the third rule-we should insist that the appraisal docu- Squire 385 ment present the impact of the project on the fiscal budget. In the event that this is negative, the analyst then would be required to explain why cost recovery was not pursued. SUMMARY So, to summarize, if we agree with Little and Mirrlees's assessment-and I certainly do-then the real issue becomes what can we do to improve current practice. The three rules I have suggested deal with appraisal optimism, shadow pricing, and project design. I'm sure there are better rules and other issues that need attention, but the general point is that we need to shift from a system in which project evaluation is left to the discretion of project staff to one in which certain procedures have to be followed more or less automatically. PROCE ED ING S OF T HE W OR L D BANK ANN UAL CON FE RE NC E ON DE V EL O PM ENT E CON OM I CS 1 9 90 COMMENT ON "PROJECT APPRAISAL AND PLANNING TWENTY YEARS ON," BY LITTLE AND MIRRLEES Ernesto R. Fontaine It is an honor for me to comment on a paper by Professors Little and Mirrlees, who have had such an enormous influence in a field I consider terribly important for the economic and social development of all nations. It is now accepted that the quality, and not merely the quantity, of investment is crucial for development-a simple notion that was not emphasized at all in the 1950s and 1960s, when mechanistic growth and "gap" models insisted on the "need" to increase saving and investment ratios through government intervention to achieve greater growth rates. It is clearly understood today that the contribution of capital accumulation to growth can be increased either by increasing the amount of investment-which implies reducing consumption-or by increasing its "real" or economic rate of return (ERR), which implies "getting prices right" for the private sector, being more careful in selecting public projects, and clearly defining the scope and extent of government action in the investment field. I believe that the work on social or economic project appraisal has been very influential in the "getting prices right" crusade, because it clearly showed how a distorted price system could lead to investment efforts by the private sector that were highly profitable for private investors but that yielded low or negative economic rates of return to the country as a whole, thus contributing very little to growth. Distortions in such situations implied more than static welfare losses-a la triangles-because they also affected the dynamics of growth. I have read this relevant paper with pleasure-and also with some difficulty, in parts. I really do not have many quarrels with it, although I did find that the authors could have given more credit to others for the "remarkable development of both the methodology and the use of shadow pricing in estimating the true national value of projects." Although Little and Mirrlees do give some credit to Dasgupta, Marglin, and Sen, they should have paid more attention to Arnold C. Harberger for his pioneering work in this field-and certainly more than just a bibliographic reference. I can attest personally to Harberger's influence; he guided my 1958 social evaluation of the sugar beet industry in Chile, a research Ernesto R. Fontaine is professor of economics at the Catholic University of Chile. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 387 388 Comment project that later developed into my doctoral thesis for the University of Chi- cago. I believe that Edmar Bacha, Lance Taylor, and Danny Schydlowsky also merit some credit in the very important development of methodologies for esti- mating a shadow (social) rate of exchange and its significance for correctly appraising the real contribution of investment projects to growth. Bacha and Taylor did significant work on shadow pricing in Chile in the mid-1960s during their stay at ODEPLAN (Oficina de Planificaci6n Nacional). The institution where I teach, Catholic University of Chile, had also done similar work before that time. Because I really do not have many quarrels with the substance of the paper, my comments will try to reinforce some of its conclusions and to suggest ways to make its analysis more comprehensive. First, I want to deal with the issue of methodological complexity. I have devoted the larger part of my professional life to preaching and teaching socioeconomic project appraisal at universities and special training programs for public officials in Latin America, and in helping Latin American countries to implement public "preinvestment systems," concentrating on Chile, a real success story, over the last decade. I have had thousands of students-engineers, agronomists, economists, doctors, military officers, business administrators, lawyers, and even veterinarians-working in public enterprises, ministries, municipalities, and central and regional planning offices and handling projects in all phases of the project cycle, from identifica- tion to implementation. Bright as these students have been, and rigorous as our eighteen-week full-time lecture course may be, I have found that it is indeed difficult to instill in students even the much more straightforward and natural technique of social pricing a la Harberger. Is it not more natural and straightfor- ward to talk about a shadow rate of exchange, and express everything in pesos, than to become entangled in multiple conversion factors a la Little and Mirrlees? I must therefore wholeheartedly agree with the authors that the complexity of the methodologies they proposed must be an important reason for the decline in their use, even at the World Bank and other lending institutions, where the training and quality of staff are far better than of professionals in most ministries and planning offices in the developing world, with possibly very few exceptions in a handful of public enterprises. Because I believe that the better is an enemy of the good, and that what is simple and good enough to pick up the biggest sources of distortions is better than applying more sophistication to arrive at more refined results, I predict that for years to come we will not see countries or even well-equipped lending institu- tions adopting complete models based on methodologies that stress numeraires; distributional weights; different values to pesos devoted to consumption or investment (and for private or public income); and such nonintuitive concepts as standard, construction, or power conversion factors. Also, methodological niceties such as the paper's concern about the "widespread effects that should in principle be estimated to calculate shadow prices" (for example, the authors' case for not using straight border prices when price varies with amount of trade) Fontaine 389 may be suitable for journal articles, but they will not be of much consequence to professional practice nor to promoting economic appraisal of projects any- where. Because I also believe that something is better than nothing, I would not be so hard on efficiency pricing, and I would at least make some reference to the usefulness of applying Harberger's methodology. Second, I cannot but also agree wholeheartedly with the authors on the need to implement economic appraisals at the early stages of the project cycle, most importantly at the prefeasibility level; I say this because little can be done later to stop really bad projects. But, at the prefeasibility level, data are not that precise, so that very refined and costly methodologies for estimating the precise shadow prices for their economic and social evaluation will not be very productive. Once again, shadow pricing based on very sophisticated methodologies will not be needed to eliminate bad projects that do not stand up to sensitivity analyses at the prefeasibility level. Third, I must also agree with the authors in rejecting the conclusions of the Pohl and Mihaljek analysis of appraisal economic rates of return and reesti- mated economic rates of return (RERRS). I have not read their work, but based on the comments by Little and Mirrlees, it appears that Pohl and Mihaljek do not discuss the reasons behind the observed differences between ERRS and RERRS. What were the differences in the corresponding appraisal and reesti- mated financial rates of return? Were shadow prices wrongly estimated, or were market prices wrongly estimated? Were social or private costs off the mark? If, as is common, the percentage differences between private and social costs do not change dramatically in the absence of big policy changes, the main source of the difference between ERRs and RERRS is not to be found in shadow pricing! In this matter, the Little and Mirrlees paper itself is also not very careful in the use of words, even though they claim that "shadow pricing and cost-benefit analysis are inseparable." For example, in referring to the decline of cost-benefit analysis due in part to the McNamara effect, they mention a "deterioration in the quality of project work," and, in discussing the problems after the reorganiz- ation of the 1980s they mention an "impression of widespread deterioration ... [in] economic analysis"; that there was "less continuity and expertise in project teams"; that "technical, financial, and institutional analysis all may have suf- fered"; and that "obviously economic analysis would not be immune." But what is meant by "economic analysis"? Is it only adjusting private values to obtain social or economic values, when private values are estimated by someone other than the economic analyst? Is economic analysis to be blamed for cost overruns in the construction of a dam when these overruns stem from insufficient geologi- cal studies, expropriation delays, suppliers' delays of deliveries, lack of good planning, or problems with disbursements? There is only so much a poor econo- mist can do! This is why project teams, composed of economists and technical experts who interact during the entire project cycle, are vital to proper project preparation and implementation; the usual practice of asking an economist at the very last moment to justify the project once its engineering aspects have been 390 Comment completely defined is fatal for good project work and relevant cost-benefit analysis. Fourth, I would have liked to have seen a history of the rise and fall of input- output "model building-computer consuming" exercises in planning offices and ministries (especially of those imposed on us by the U.S. government's Alliance for Progress under the Kennedy administration). Successful planning-such as in Chile over the last decade-concentrates on putting together a coherent public capital budget based on projects and programs that have been appraised at all stages of the project cycle. Also, I think the paper ought to have included a history of the quite substan- tial improvements accomplished by the profession in the development of specific techniques and models for both financial and economic analyses in certain sec- tors, such as ports, roads, electricity, potable water and sewerage, multipurpose water projects, and, most important, the so-called social projects. In this respect, I missed in the Little and Mirrlees paper a section on basic needs, which I believe deserve proper consideration nowadays, even though the topic was, I seem to recall, not mentioned in the authors' book. In Chile, we have very successfully used "social prices" based on the basic needs approach for the economic appraisal of projects in the social sectors, whose goods and services can easily be focused on the poorer members of our population, thus giving strong economic arguments in favor of devoting public funds to such human- capital-enhancing projects vis-a-vis traditional infrastructure and industrial projects. In this respect, my long and very enriching experience with project appraisals done by reputable consulting firms and accepted by international lending institu- tions has taught me that the greatest mistakes originate in not paying enough attention to the separability of the projects and to the "with" and "without" project situations, which are the basis for identifying the relevant costs and benefits attributable to the project. In particular, the optimization of the sce- nario without the project has been neglected. This is especially true of public investment in sectors where it clearly preempts or displaces private investment. Advances in this area have been outstanding, thus avoiding construction of new infrastructures when optimization of present ones has been possible-as has been the case with ports. I have found relatively few mistakes in measuring and valuing (putting a private or social value on) the relevant costs and benefits so identified. Also, I have often found the mistaken use of ERR and expected present value for decisions regarding projects whose benefits increase mainly as a function of time-roads, ports, and potable water. Here, ERR and EPV obvi- ously lead to overinvestment and are not relevant for the proper decision of establishing the best year to start the project. The right number to use in these cases is the "instantaneous internal rate of return," or the comparison of first- year benefits with "capital costs" of the project. Fifth, I offer a reflection on the redistributional aspects of projects that do not charge users for the goods and services produced-or that do not charge bene- Fontaine 391 ficiaries for the projects' investment costs. These may be of great relevance for understanding the "political economy" of government investment. It is clear that making the beneficiaries pay for the project will help to avoid political pressures to build irrigation dams, roads, or ports, when and where the national interest is not best served, thus constituting an argument that reinforces Little and Mirr- lees' presumption that public income may be worth more than private income. PROCE ED ING S OF T HE W O RLD BANK ANN U AL CON FE R EN C E ON DE VE LO PM ENT E CO NOMICS 1 9 90 FLOOR DISCUSSION OF THE LITTLE-MIRRLEES PAPER A World Bank participant questioned the authors about the rejection in their paper of sustainability as a concept without merit. The participant felt that the Hicksian definition of income also incorporates a notion of sustainability, so in rejecting sustainability, were they also discarding the Hicksian definition of income? Or was is just that the authors were impatient with "sustainability" being used as a buzzword? Another Bank participant drew attention to the use of discount rates that are constant over time and even sometimes across countries. He thought this was a fatal flaw in the implementation of the methodology, at least in the Bank. He also questioned the wisdom of using point estimates of net present value. Would it not be much better-with the availability of microcomputers-to do Monte Carlo simulations on price and technical parameters and get a distribution of net present values? A participant observed that the prevailing cost-benefit analysis methodology was so complex and needed so many arbitrary assumptions that for a compli- cated project it was possible to arrive at any arbitrary (and predetermined) rate of return by manipulating the assumptions. It was almost impossible to trace such manipulation; perhaps this accounted for the historical rise in expected rates of return: analysts were getting better at the manipulation. A Bank participant said that the authors had made the comment that public sector revenues are worth more than private sector income; however, this raises the question of whether the revenues are likely to be used more profitably by the public sector. The participant was also concerned that when major corrections are made in evaluating public investment projects, this may lead to a situation in which the public and private sectors are marching to different price signals. If major corrections need to be made, then one needs to change macroeconomic policy, not make corrections for individual projects through cost-benefit analy- sis. He agreed with Ernesto Fontaine's (discussant) comment that what is most important to consider is the "without projects" scenario. This scenario can be manipulated to obtain whatever rate of return for a project one wants. Perhaps the solution, as Lyn Squire (discussant) pointed out, was that the Bank should ask its staff to routinely perform an optimized scenario without the project. A Bank participant observed that from his experience, both within and out- side the Bank, expending large resources to improve the cost-benefit analysis of individual projects was much less important than addressing the distortions in This session was chaired by Harris Mule, member of the Council of African Advisers of the World Bank and former permanent secretary in the Ministry of Finance and Planning, Republic of Kenya. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 393 394 Floor Discussion of Little and Mirrlees Paper the economy as a whole. The incentive structure of the economy affected both public sector investments and private investments. The Bank had moved toward adjustment lending precisely to deal with such macroeconomic distortions. However, it seemed to the participant that the situation now had come full cycle-good structural adjustment lending was not easy, either. He felt that on balance there should still be a limit to the resources one devoted to better project analysis if the opportunity cost was resolving or understanding the distortions of the economy as a whole. Ian Little welcomed Fontaine's comment that cost-benefit analysis is comple- mentary to getting prices right, far from being competitive with this endeavor. He commented that the move to adjustment lending from project lending per- haps happened because policymakers were taught by the difficulty of doing cost- benefit analysis that it might be a better idea to get prices right instead! Little pointed out, however, that irrespective of how much one gets prices right, there are always infrastructural projects for which there is no market, and benefits have to be identified without actual market prices. Regarding sustainability, Little said his criticism of the concept did not pre- clude spending to maintain an asset. By and large he accepted the Hicksian definition of income, except that the definition does not deal with cases of changes in expectations or windfalls. Little agreed with the comment that one could always manipulate or fudge a complicated project to make it acceptable; this was a genuine difficulty. Answering the question about the balance between better project analysis and better structural adjustment lending, Little reiterated that he could not see why they should be competitive. He felt that views were divided on this even within the Bank, and after all, three-quarters of Bank lending is still for projects. Little recalled the comment from the conference discussion of the previous day, when it had been observed that the concentration on structural adjustment lending has led to neglect of proper project analysis in the Bank. Jim Mirrlees said that he also did not see a tension between better project analysis and sounder adjustment lending. Sound project appraisal should in no way preclude price reform and policy reform. Responding to the comments by Squire, Mirrlees said that the general principle of thinking of a series of levels of complexity in project appraisal accorded well with the philosophy of project appraisal stated in their paper. A Bank participant suggested that the evaluation experience of World Bank projects had not been as dismal as Little and Mirrlees have suggested. More than approximately 70 percent of the completed projects, as audited by the Bank's Operations Evaluation Department, had been successful. Of the failed projects, there appear to be concentrations in Africa and in rural development. It is interesting, Squire remarked, that the focus is always on the marginal cases. Even the analysis by Pohl and Mihaljek quoted by Little and Mirrlees, he noted, shows that the gaps between ex ante and ex post estimated rates of return are Floor Discussion of Little and Mirrlees Paper 395 greatest where the ex ante rates are very high, on the order of 20 or 30 percent or more. Another participant pointed out that perhaps it was time for the Bank's Eco- nomic Development Institute to assess whether it should reestablish itself as a training center for project evaluation and management. A participant returned to the earlier question about the different valuation of public and private incomes in the Little-Mirrlees framework. He said that whereas many would have had no problem with the proposition twenty years ago, the experience of the intervening years would suggest at least that there is no systematic reason for the higher premium on government income, and much would depend on the government one is dealing with. In many instances one would not put a premium on public income, but in some countries one might. Mirrlees responded to the public-versus-private-income questions by noting that in cases where the cost of public funds is high, and the waste in using public funds is also very high, then the real question, for the value of public income, is what changes if an additional project generates profits in the public sector. The tendency is to think that public sector profits will not be spent well; and that public sector losses will be financed through an inflation tax. One had to avoid a tendency to think that something coming out of the public purse is very costly and that something going into it is not worth very much. It is not just a matter of the difference between particular governments. One faces a difficult judginent about the balance between public income and expenditures. Mirrlees agreed that in their paper they had focused only on the cost side of public income and not on public expenditures, since it was easier to deal with the cost side. In concluding the floor discussion, Harris Mule (chair) summarized by noting that engineers, technical people, institutional experts, and management special- ists have to be brought in, so that cost-benefit analysis can be integrated into overall project identification, preparation, and appraisal. Within the public sec- tor itself, Mule said that the resource allocation that is subjected to social cost- benefit analysis is a very small component of the total outlays. In many coun- tries, current budget expenditures and small projects in the public sector amount to more than the larger projects to be appraised in detail. In devoting significant resources to the latter, we may not be using resources most effectively. Mule ended the session by thanking the participants and the authors. P ROCE ED INGS OF THE W O RLD BANK ANN UAL CON FE RENCE ON DE VE LO PM ENT EC ONO MICS 1 9 9 0 Projects versus Policy Reform Ravi Kanbur This paper examines the interrelations between the World Bank's project (investment) loans and its policy reform (adjustment) loans. Since the late 1970s, adjustment loans have come to constitute about a quarter of annual World Bank lending. The paper proposes a framework, based on general equilibrium economics, for comparing and analyzing the two types of lending and develops methods for evaluating policy reform packages according to "project-like" criteria as a framework for disciplining thinking in assessing the net benefits and appropriate balances of the two types of lending. The paper observes that policy reform that reduces distortions in an economy also increases the social profitability of a range of projects; that projects that increase supply respon- siveness in a particular sector strengthen the argument for reducing price distortions in that sector; and that policy reform packages should be considered "projects" with net return profiles and evaluated accordingly. If official documents are anything to go by, the World Bank these days under- takes two main types of lending. The first is for projects, which are the usual roads and railways affairs. The second goes under the general rubric of adjust- ment lending. This comprises structural adjustment loans (SALS) and sector adjustment loans (SECALS). One quite often reads that adjustment lending is in support of "policy reform." Evidently, the growth in this type of lending has been spectacular. Starting from virtually nothing in the late 1970s, policy-based lending made up 25 percent of annual World Bank lending and 10 percent of its portfolio by 1988 (World Bank 1989). Given these two categories-project- and policy-based development lending- one might ask, what are the interactions between them? I leave to one side institutional questions of managing the two types of lending. My focus is on the conceptual, and I want to pose three questions: (1) In what way does policy reform (or lack of it) influence the evaluation of projects? (2) In what way do projects (or lack of them) influence the evaluation of policy reform? And (3) how are projects and policy reform to be evaluated vis-a-vis each other, especially Ravi Kanbur is editor of the World Bank Economic Review and the World Bank Research Observer at the World Bank and professor of economics at the University of Warwick. He is grateful to Charles Blitzer, Sudhir Chitale, Ajay Chhibber, Paul Isenman, Carl Jayrajah, Geoff Lamb, Anandarup Ray, Joanne Salop, Lyn Squire, and Michael Walton for helpful comments. ( 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 397 398 Projects versus Policy Reform because the marginal dollar of lending, by definition, can only support one or the other? Before proceeding any further, we need to define projects and policy reform. On projects, Squire's (1989) survey provides a straightforward analytical defini- tion: "A project may be defined as a combination of inputs and outputs." Squire goes on to restrict attention to public sector projects, as do Dreze and Stem (1987) in their survey of cost-benefit analysis: "We ... represent a small project as an infinitesimal perturbation . . . of [the] public production plan." These analytical definitions do correspond somewhat to project descriptions one might find in actual World Bank loan documents, although, of course, in many cases the perturbations are anything but infinitesimal. Any attempt to get a precise definition of policy reform by looking at adjust- ment loan documents is unlikely to succeed. (For a detailed analysis of loan agreements for one country, Cote d'Ivoire, see Kanbur 1990.) Each loan has attached to it a multitude of policy changes ranging from price changes to institutional reform in the parastatal sector, to retargeting of health expendi- tures through user fees, and so on. In fact, a reform (for example, of inefficient enterprises) could directly affect the pattern of public production, thereby mak- ing the reform more like a project according to the Squire-Dreze-Stern defini- tion. The situation is complicated further by the fact that many project or investment loans carry conditions on policy reform in the sector or subsector in which the investment takes place. Although investment loans and adjustment loans have certain overlaps, they do differ from each other in the nature and speed of disbursement. Investment loans are closely tied to expenditures incurred on the project and are therefore disbursed relatively slowly over time as the project evolves. Although the tranched releases of adjustment loans are conditioned on the implementation of various policy actions, they are disbursed relatively quickly and without sub- stantial restrictions on the use to which the money is to be put. In practice, the World Bank has a spectrum of lending instruments ranging from loans that are specific to a particular investment and have some subsector-specific policy reforms as a condition, to loans that are untied and quick-disbursing but are conditioned on economywide or sectorwide policy reform. In this paper I cut through these complications and stick with the classification of loans of the types at either end of the spectrum. This sharpens the contrast between investment lending and adjustment lending, the better to compare them by. Moreover, for the most part I conduct the discussion in terms of pricing reform. Although I do this simply to focus ideas, it should be noted that pricing reform is perhaps the dominant component of most policy reform packages. To explore the interrelations and implications of project and policy reform lending, I begin in section I by briefly setting up an analytical framework. Then, in section II, I turn to complementarities between projects and policy reform. In section III, I address the major subject of this paper-the evaluation of policy reform packages. Finally, in section IV, I draw some relevant conclusions on the logic and appropriate balance of project and policy lending. Kanbur 399 I. AN ANALYTICAL FRAMEWORK There is, of course, a large literature on project evaluation and on policy reform, and my intention is to use this literature to focus attention on the three questions posed in the introduction. I use the standard analytical framework of general equilibrium economics, and would note that several variants and exposi- tions are available, including those of Squire (1989) and Dreze and Stern (1987). For my purposes, however, I have found the exposition by Boadway (1975) extremely useful, and I rely on it. Boadway, in turn, follows Diamond and Mirrlees (1971), one of the classic references in the area. I want to focus initially on efficiency questions, so let us suppose that there is a representative consumer who takes consumer prices as given and has commodity demands that maximize utility (factor supplies are treated as negative demands). There are two sources of supply in this economy-the private and the public sectors. The private sector takes producer prices as given and determines sup- plies to maximize profits (factor demands are treated as negative supplies). In the simplest case, we assume constant returns to scale in production, so that profits are maximized at zero. (If profits were not maximized at zero we would have to take into account what happens to them; these complications can be handled but are best left as elaborations on the basic story.) The public sector also supplies commodities and demands factors, and a project is thought of as a change in the pattern of production of the public sector. The objective of public sector production, or of changes in it, is to enhance consumer welfare. Apart from its production role, the government also levies taxes on commodi- ties and factors. It is these taxes that drive a wedge between the prices faced by producers and those faced by consumers, and reducing this "distortion" is the primary objective of many pricing reforms in World Bank SALs and SECALS. We suppose that the government does not have access to lump-sum taxation instru- ments; that is, it does not have access to instruments that do not affect the margin of choice for producers and consumers. The equilibrium of this stylized economy is specified-through equality of supply and demand in all commodity and factor markets-for given public production and given taxes. A further refinement can be introduced by distin- guishing between tradable commodities (that is, commodities whose consumer prices are given from outside, with the equilibrium between supply and demand occurring through trade flows) and nontradable commodities (that is, those for which equilibrium is brought about by equality of domestic demand and domes- tic supply, with prices adjusting to clear the market). Much of the project evaluation literature is concerned with this distinction. I will eschew it for the bulk of my discussion, because my focus is somewhat different. I will start off by assuming that the patterns of public production and taxes are what they are and ask, how can they be improved? Improvement in this context is an increase in the value of some objective function. Boadway (1975) started with a small, arbitrary variation in the pattern of public production and the pattern of taxes. He then solved for the new general equilibrium of the 400 Projects versus Policy Reform economy-that is, the new prices and quantities, such that all markets clear and all budget constraints, including that of the government, are met. The question is, does this new state of affairs increase consumer utility? To introduce the key elements in the answer to this question, I must introduce some notation (following Boadway 1975): z =vector of public production; dz =vector of changes in public production (the "project"); q =vector of consumer prices; p =vector of producer prices; t =q - p = taxes; dt =vector of changes in taxes (the "policy reform"); x =x(q) = vector of consumer demand; y =y(p) = vector of private production; W = consumer's utility in terms of the chosen numeraire commodity. With this notation, Boadway (1975, p. 368) develops a general expression for change in utility due to a project-cum-policy reform: (1) dW =(-+ * E-1 aY) dt + (p + t E1 )dz where ax/aq and ay/ap are matrices of demand and supply derivatives with respect to consumer and producer prices, and E = ax/lap - ay/ap is the matrix of price derivatives of excess demands. A project-policy reform combination that is consistent with equilibrium (in particular, with meeting the government's budget constraint) leads to a change in welfare, as given by equation 1. This allows us to take a systematic approach to answering the three questions posed in the introduction on the interaction between projects and policy reform. The next section starts with the first of these questions. II. COMPLEMENTARITIES BETWEEN PROJECTS AND POLICY REFORM The Impact of Policy Reform on the Evaluation of Projects It has long been recognized that policy-induced distortions in market prices should be taken into account in the evaluation of projects. Indeed, as Squire (1989) notes, it was the presence of these distortions that led to the arguments for using "shadow prices" in the evaluation of public sector projects. If there are no distortions at all-that is, t = 0-then from equation 1 we get: (2) dW = pdz In other words, the public sector should behave like the private sector-only undertake a project if it makes a positive profit at market prices, because then and only then will the project lead to an improvement in consumer welfare. Kanbur 401 If there are distortions, that is, t * 0, but attention is focused only on a project, that is, dt = 0, equation 1 gives: (3) dW =(p + t * * E-1 )dz In this situation, we get the well-known result that the public sector should not behave as though it were a private firm. Rather, it should undertake projects that make a profit at shadow prices: (4) ~~~~~~~ax (4) s =p + t * d E-1 aq which take into account both the repercussions of the project on the rest of the (distorted) economy and this indirect effect on welfare. The adjustment to p is in principle quite complicated, but if cross-price effects and income effects are ignored, it can be shown that "the shadow price should be a weighted average of consumer and producer prices, the weights reflecting the proportions of the increase [in public production] coming from forgone demand and increased [private] supply" (Boadway 1975). This well-known rule goes back at least as far as Harberger (1964). It is also well known that equation 3 would hold for arbitrary dz (not just those dz vectors that in the new equilibrium maintained the government's budget constraint) if the government had lump-sum tax instruments to dispose of any surpluses or deficits that accrue to it as a result of the project. But this is an unrealistic assumption, and much has been written about what happens when, after the project and at the new market equilibrium, the government does indeed end up with a surplus or a deficit (see the surveys by Squire 1989 and by Dreze and Stern 1987). Because one of the budget constraints in the economy is not met, this cannot in fact be described as an equilibrium. If the government restores its budget constraint through the use of distorting taxation, this is another element to be weighed in shadow pricing. Squire (1989, p. 1114) showed that a third term is now added to equation 4-essentially the effects captured in the first term on the right-hand side of equation 1: "This term becomes more important as the change in private sector welfare increases and as the distortionary cost of taxation increases." The above results are standard and fairly well known in the literature. Suffice it to note that in the evaluation of projects we already have present not just the issue of distorted policy but of policy reform. If a project leads to surpluses or deficits, the way an equilibrium is modeled as being restored is through a change of distorting tax instruments-in other words, through policy reform. The ques- tion I now want to raise, however, is this: suppose the "policy reform" branch of the government recommends a change in the pattern of taxes to improve wel- fare. What implications does this have for the "projects" branch of the govern- ment? I seek to capture the extent of distortion in the current system in terms of the marginal welfare cost of raising revenue (measured in terms of the numer- 402 Projects versus Policy Reform aire). Presumably, therefore, only those policy reforms would be recommended that lower this cost. What sort of project would now tend to be accepted? I want to think through this issue in the context of the three archetypal projects analyzed by Squire (1989)-industrial projects, infrastructural projects, and utility projects. But before doing so I want to note that for tradable com- modities in the Diamond and Mirrlees economy, shadow prices can easily be shown to be border prices (I will not go into that voluminous literature, emanat- ing from Little and Mirrlees 1969, here). Given this, industrial projects are in many ways the easiest to analyze because, according to Squire, they are "charac- terized by relatively minor interactions with the private sector. All production is within the public sector and the main outputs and inputs are tradable. Purchases of nontradable inputs and factors of production represent the only interaction with the private sector and these may be relatively small" (1989, p. 1127). In these situations the tradable component of the project dominates, and small changes in shadow prices of nontradables due to a reduction in the distorting costs of taxation are unlikely to influence appraisal. More interesting are infrastructural projects, whose benefits accrue primarily to the private sector. If the output produced by infrastructure is traded, its shadow price is its world price. But a correction has to be made for the fact that in order to recover the benefits from the private sector, the government will naturally have to use the current, distorting tax system. To the extent that the marginal welfare cost of the present system is positive, and to the extent that policy reform reduces this distortion (why else might one recommend the reform?), infrastructure projects would tend to be more favored after the reform. The final type of project Squire (1989) considered is a utility project that provides, say, electricity to the private sector, and this in turn leads to an increase in tradable net output of the private sector. If electricity is sold to the private sector at below cost, then shadow pricing should also take this into account. Thus either or both of (1) a reduction in the welfare cost of distorting taxation, and (2) an increase in electricity charges, would make this project more socially profitable. In fact, the latter reform is already present as a condi- tion in some utility projects. We can thus argue that although in general the interactions are too complex to arrive at clear solutions, in a number of stylized cases policy reform that reduces the welfare cost of distorting taxation is likely to increase the social profitability of projects that were previously being rejected, particularly those for which the benefits accrue primarily to the private sector. It should be noted, however, that there are other stylized cases and mechanisms though which the argument can go the other way. Gersovitz (1990) has argued, for example, that where the spatial pattern of pricing is distorted, perhaps because of pan-territorial pricing, transport investment projects look good because they correct in part for this distortion. When the distortion is removed, the case for transport investment is weaker. Kanbur 403 The Impact of Projects on the Efficacy of Policy Reform Returning to the basic equation 1, let us now concentrate on dt and suppose for the moment that dz = 0; that is, we focus attention only on a policy reform, and distortions are present, t * 0. In other words, we are supposing that the deficit of the public sector has to be financed through distortional taxation. Then (S) dW=(-ta E-1 'p)dt If cross-price effects can be ignored, this becomes recognizable as (6) dW ti axi( ,iXi/,3_ayi/api i (6) ~ ~ ~~ ~~aq ax1laqi - ayilapj dt The term for each tax change shows the usual dependence of the welfare gain from tax reform on the supply and demand elasticities. In particular, the larger the elasticity of supply, the more likely it is that the commodity merits a tax reduction-this is the usual variant of the Ramsey rules. What is of interest to us here is the possible impact of a project on the various elasticities in equation 5 or equation 6. The easiest case is that of an infrastruc- ture project, which affects price elasticities of supply. In fact, it is often argued that infrastructure projects, by increasing agricultural supply responsiveness, can help "adjustment." The evidence on the role of roads and other facilities in agricultural supply response has been mounting (see, for example, Binswanger 1989). But an increase in supply elasticity has immediate implications for tax reform via equation 6-it strengthens the case for reducing distorting taxes on the commodity whose supply elasticity has increased. This is because with a higher supply elasticity the impact on domestic consumer price of a given change in distortional tax is greater. However, another caveat needs to be recorded here. Outside of this framework cases may arise in which investment and price reform are substitutes rather than complements. Gersovitz (1990) shows that after an investment that reduces transport costs, the welfare gains from remov- ing spatial distortions in pricing may be smaller. III. THE CHOICE BETWEEN PROJECTS AND POLICY REFORM On the face of it, the choice between projects and policy reform is a nonissue. Recall equation 1. We simply undertake all combinations of dt and dz that make equation 1 positive. What's there to choose? And yet in reality there does indeed seem to be a margin of choice. Development institutions such as the World Bank have shifted their portfolios of lending in favor of programs of policy reform and away from traditional projects. One presumes, therefore, that there are some projects that would previously have been supported that are not now being supported. How should we think about such a shift in priorities? 404 Projects versus Policy Reform Policy Reform as a Project Let me approach this question in a somewhat indirect way, by asking another question: Why should external agencies provide advice on and financial support for programs of policy reform that improve the government's own objective function? The advice function is clear enough, and this is the simplest to handle. For whatever reason, the outside agency may simply have better technical knowledge and information on the economy than the government itself. In this case, simply conveying the appropriate information should be enough. Although this is plausibly an element in many technical assistance packages, would that matters were really so simple! The answer to why governments do not undertake reforms that advance their own objectives, as shown in stylized manner in equation 1, must be that the stylized representation in equation 1 is fundamentally wrong, and it can be wrong in two related ways. The first is that the government and the develop- ment agency do not share the same objective function. The second is that they do, but the model of the economy underlying equation 1 is seriously misleading. Let us start with the first of these. As noted earlier, distributional considera- tions can be easily introduced into the development of the various expressions in section II-the efficacy of different types of reform now depend on distributional weights. If these distributional weights differ, then there will be disagreement about which reforms to undertake. Several types of disagreement are possible. One is that the government's currently distorting pattem of taxation reflects relatively high weights given to the poor, whereas the development agency is concerned only with efficiency. Another is that the government's currently dis- torting pattern of taxation reflects weights given to groups that support it (and these need not be poor-some of them could be politically powerful by virtue of being rich), whereas the development agency is concerned only with efficiency. The most extreme divergence would be cases in which the current pattern of taxation reflects the interest of the rich, whereas the development agency is concerned with outcomes for the poor. In this view, the financial support provided by development agencies, upon condition of particular policy reforms being undertaken, is essentially a device to relieve the tension between two different objective functions. Some would call it a "bribe." A transfer of resources of this type can lead to compatibility between diverging directions of tax reform dictated by competing objective functions. The simplest way to think of this is if a government can compensate its support- ers in a lump-sum manner for losses incurred in a balanced-budget tax reform that improves the welfare of the poor, say. Then the availability of external resources will make it more likely that the government will in fact undertake the reforms. (For a theoretical analysis of politically constrained price reforms, see Braverman and Kanbur 1987 and Kanbur and Myles 1990). Of course, in principle this same argument applies to projects-to the dz part of the expression in equation 1. There may be divergence between projects Kanbur 405 favored by the government and those favored by the development agency because of divergences in objective functions. However, although this issue does exist, as reflected in the requirement that World Bank project documents at least discuss the impact on the poor, it seems to be not as big an issue in projects as it is in policy reform. Why? One reason seems to be that projects are often localized in their impact on individual standards of living, whereas policy reform has an economywide impact. A change in the price of imported food will affect more people than an infrastructure project in an isolated region. Also, even if the infrastructure project has an impact eventually on the price of food, this occurs in the future, whereas policy reform has an immediate impact on people's stan- dard of living. But we are now introducing the time dimension, which is the second of the reasons why-even if objective functions agree-the government and the devel- opment agency may differ on the efficacy of a given tax reform, and why financial support may be necessary. As is well known, time can be introduced into the Diamond and Mirrlees economy of section II simply by also labeling goods with the time at which they are produced. The vectors dt and dz now consist of tax changes and of input and output changes to the public sector pattern of production, not only at one point in time but at all points in time into the future. The prices p and q, and consumption, x, are similarly long vectors of prices and quantities now and at all dates in the future. This device, although providing some basic insights (for example on the social rate of discount), still cannot capture the real reason some governments fight shy of policy reform. The problems arise in terms of the equilibrating mechanisms in the economy and, once again, in terms of distributional questions in the long term and the short term. To discuss these questions I want to specialize the general Diamond-Mirrlees model to the standard case of a 2 x 2 x 2 trade model. Take the case of an economy that is producing two tradable goods using two factors (say, capital and labor). There is a distorting tariff, and the question is whether to remove it so that the domestic relative price ratio between the importable and the export- able becomes equal to the world price ratio-which is the most efficient outcome for this small open economy. Let us follow the consequences of this change for the economy (see also Kanbur 1987 and 1988). The immediate effect of this relative price change is to increase entrepreneurial profits in the exportable sector and reduce them in the importable sector. Assuming that factors are sector-specific in the short run, factor prices will be bid up in the exportable sector and bid down in the importable sector. If there is downward factor price rigidity, then there will be unemployment of resources in this sector in the short run. Output in the exportable sector is limited by factor supply, and output in the importable sector falls, so there will be a loss of national output in the short run. Let us suppose that in the medium run factor prices are flexible, so that over this horizon prices of both factors will fall in the importable sector and increase in the exportable sector. This divergence over the medium term in factor 406 Projects versus Policy Reform rewards across the two sectors may also be of concern to the government. Over the longer term, factors will move in response to reward differentials, and this reallocation is, after all, the objective of the initial reform. In the new long-run equilibrium there will be the usual Stolper-Samuelson effects-the factor used more intensively in the exportable sector will gain. We thus note two features of the dynamics of the adjustment story. First, if there are factor market rigidities, changes in relative prices that lead to an improvement in national income in the long run may well lead to reductions in national income in the short run. (This point can be made in a number of different contexts and models. Neary [1982], for example, makes it in a model in which wages are equated across the two sectors but this common wage is downwardly rigid, whereas capital is sector-specific in the short run.) Second, throughout the adjustment process there will be distributional effects. These points, which are perhaps best made in the context of simple models, can also be made in models with considerably richer structure. One such exer- cise is that by de Janvry, Fargeix, and Sadoulet (1990). Based on the work of Kouwenaar (1988) for Ecuador, they construct a computable general equilib- rium model with five sectors (agriculture; oil; industry; utilities, construction, and services; and administration) and two factors of production. They consider the consequences of a fall in the price of primary exports and a reduction in government foreign borrowing under various adjustment scenarios-one of which is no adjustment. As can be seen from figure 1, real gross domestic product (GDP) without the shock would have continued increasing. The shock puts paid to this rosy prospect. However, notice the short- and long-run differ- ences in real GDP with and without adjustment. The adjustment scenario hurts in the short run. Relative to the no-adjustment scenario, the "net return profile" is negative to start with and positive thereafter. Even more interesting are the short- and long-run distributional conse- quences, as shown in figures 2 and 3, which show the time paths of rural and urban poverty under the adjustment and no-adjustment scenarios. In the short run, rural and urban poverty are higher with adjustment than without adjust- ment. This is to be expected, given the real GDP paths. However, rural poverty starts doing better (relative to the no-adjustment scenario) much earlier than urban poverty. Moreover, in the long run, rural poverty does better (relative to nonadjustment) than does urban poverty. Let me return to the basic two-sector analytical framework and pose again the question, why adjustment lending? Figure 1 provides a clear and perhaps con- ventional reason. We think of policy reform as leading to higher real national income-otherwise there would be no reason to recommend it (assuming an efficiency objective on the part of the development agency). But in many econ- omies policy reform will reduce real national income in the short run (compared with the no-reform scenario). The profile of net returns is thus rather like that of a typical project-negative to start with and then positive. Whether or not the reform is a good thing depends on the present value of the net return profile, Kanbur 407 Figure 1. Alternative Real GDP Scenarios Index of Real GDP 120 110 100 < 90 80 1 2 3 4 5 6 7 Year Figure 2. Alternative Rural Poverty Scenarios Percentage of rural poor in total population 30 20 10 0 . 1 2 3 4 5 6 7 Year Key: Adjustment; -- - no adjustment; - no shock. Source: de Janvr, Fargeix, and Sadoulet (1990), as reported inWobrd Bank (1990). 408 Projects versus Policy Reform Figure 3. Alternative Urban Poverty Scenarios Percentage of urban poor in total population 30 20 10 . .. . ... 0 1 2 3 4 5 6 7 Year Key. ~~ Adjustment; ------ no adjustment; - - no shock. Source: de Janvry, Fargeix, and Sadoulet (1990), as reported in World Bank (1990). calculated at an appropriate rate of interest. Redistribution from the future to the present is generally preferred, so it would be rational for the country to borrow to smooth out the initial dip in real GDP, if the rate of return to adjust- ment was greater than the rate of interest on borrowed funds. A similar argument would present itself if the government had poverty con- cerns or distributional concerns in general. What is interesting is that the distri- butional arguments are present even if there is no short-run "dip" in national income. In the two-sector model, removing the distorting tariff will increase national income in the long run. Suppose that the long-run equilibrium is attained instantaneously-that is, that there are no short-run problems of resource unemployment. Now we know that if the tariff is removed and relative prices change, there will be winners and losers, even in the long-run equilibrium. The reason we are recommending the reform from an efficiency point of view is that the gainers gain more than the losers lose. To use some old-fashioned welfare economics terminology, the reform passes the compensation test-the gainers could compensate the losers and still have something left. The point is, of course, that there do not necessarily exist mechanisms to effect this redistribution. This issue becomes particularly acute if there are political economy con- straints on the government, delineating the maximum that the losers would Kanbur 409 willingly lose without bringing the system down. For simplicity, let us suppose that the group in question will not countenance any such losses. What is the government to do except resist the advice of the development agency to under- take the reform? If it had lump-sum tax instruments to recover the gains from the gainers, it would do so-but these instruments do not exist. In fact, even distortional tax instruments to tax back the gains from the gainers do not necessarily exist. The answer may be that the government should borrow to compensate the losers now, and repay the debt as it taxes back revenue from the gainers over time. In this setting the interest payments on the adjustment loan are the price of not having a sufficiently developed set of instruments to tax the gains from the reform, in a situation in which preventing losses to politically key groups is a binding constraint. This simple story needs to be elaborated upon a little. First, in this "static" setting the case for borrowing has to be based on the government's ability to tax back progressively more and more of the gains while providing full compensa- tion every period to the losers. This may be plausible, but it either relies on a progressive improvement in the government's tax collection abilities or on another behavioral restriction that lays down how much of their gains the gainers are willing to give up (and which says that this increases over time). In a growing economy these analytical contortions are not necessary. The only behavioral restriction necessary is that agents will not countenance a decline in their standard of living. The compensation required is then temporary and can be financed through (say) a fixed proportionate tax levied on the continually increasing incomes of the gainers. A second point that might be raised is on the distorting effects of compensa- tion to the losers and the taxation necessary to recover the gains from the gainers. If lump-sum instruments do not exist, then might it not be the case that the compensation required to meet a binding political economy constraint may introduce a distortion that is much worse than the original distortion whose removal motivated the policy reform? Braverman and Kanbur (1987) have shown that with restricted types of compensation it might be best not to enter into certain policy reforms at all! However, other policy reforms are still worth- while, so the basic question of support for these reforms still remains. Lending for Projects versus Lending for Policy Reform Should the marginal dollar of lending (or borrowing) go to projects or to policy reform? Given the above characterization of policy reform as having a project-like profile of returns, the answer seems straightforward-calculate the internal rate of return on projects and on policy reform, and use resources to support whichever is higher. Before elaborating on this, however, we must do some ground-clearing. First, note that because of the fungibility of money, the notion of a particular loan being tied to a particular project or policy reform is questionable. World 410 Projects versus Policy Reform Bank (1989) mentions the legal reason, coming from the Articles of the Bank, that World Bank funds are to be used "only for the purposes for which the loan was granted." But this formalism should not hide the fact that the marginal lending may not necessarily be that which accompanies a project or a program. In this setting the lending that accompanies a program may simply reflect an overall increase in lending to an economy in which the marginal product of investment will increase as a result of the program. But it still leaves open the question of where marginal resources should go. Second, policy reform packages are currently appraised quite differently from traditional projects. Although for both there are projections of relevant eco- nomic magnitudes with and without the project or the policy reform, so far as we can tell, rate-of-return calculations are not presented for policy-based lend- ing as they are for project-based lending. World Bank (1989) notes that adjustment loans differ from traditional project loans in that their benefits are difficult to pinpoint and estimate. The benefits hardly ever accrue directly to the borrowing governments, but rather tend to be spread widely through the economy, accruing to labor and to capital, to farmers, retailers, importers and other entrepreneurs. There is no practical way for the government to track down such widely dispersed benefits of the loan, nor can it cause the beneficiaries to pay the cost of the loan. Instead, the government will typically turn to general revenues to service the debt. In a standard adjustment loan scenario, the cash to service the loan should be interpreted as being raised through taxes. In section II, I discussed projects, such as those relating to infrastructure, in which the benefits do not accrue directly to the government, and how evaluation of such projects would be complicated by this factor. Similarly, although it can be argued that the consequences of a policy reform package are subject to great uncertainties, so are various aspects of project outcomes. In any event, projec- tions of various macroeconomic quantities are routinely undertaken, including "with reform package" and "without reform package" scenarios. These are sub- ject to criticisms (most importantly, on the link between policy reform and increased productivity-usually reflected in the incremental capital-output ratio assumptions) but are nevertheless a basis on which the loan decision is judged. These projections (which can be, and often are, cast in the form of graphs similar to figure 1) can be used to calculate rates of return and would be subject to the same types of problems as rates-of-return calculations on projects. I would not wish to minimize the difficulties of achieving more of a congru- ence between techniques for appraising project loans and those for appraising policy-based lending, but I would argue that this congruence is something one should strive for-certainly the logic dictates it. Are there any general considerations that might lead to arguments in favor of policy-based lending? One way of approaching this might be to look at the ratio of long-run benefits to short-run costs. Policy reform has two main types of short-run costs-the resource underutilization costs typically associated with Kanbur 411 changing some relative price when others are inflexible, and the distributional costs. Resource underutilization can be thought of as analogous to the capital costs of a project, and one can perhaps plausibly argue that they are equally as large relative to long-run benefits. In fact, Squire's (1989) typical "industrial" project shows capital costs as 74 percent of the present value of total benefits, with all calculations made at market prices. Even at shadow prices, capital costs accounted for as much as 36 percent of the present value of total benefits. For Squire's.typical "infrastructural" project these numbers were 5 percent (at mar- ket prices) and 8 percent (at shadow prices), whereas for his "utility" project they were 5 percent and 6 percent, respectively. A calculation of the ratio of up- front costs of an adjustment package can be made for the policy reform scenario depicted in figure 1. If we take the critical dip in real GDP under the adjustment scenario (relative to nonadjustment) as the equivalent of the up-front capital cost of this "project," then with a horizon of seven years and a discount rate of 10 percent, the cost is about 75 percent of the present value of the benefits. On the distributional front, it can be argued that policy reform packages are likely to hurt many more groups of people than a typical project, especially if the latter is restricted to a specific sector. The likelihood of requiring short-run compensation on political economy grounds is thus far greater with a policy reform package than with an isolated project. For this reason if for no other, up- front financing is more likely to be needed for a policy reform package than for a project. The above discussion can perhaps be put in the context of the Special Program for Africa, a major effort on the part of the World Bank and bilateral donors that is making funds available in support of policy reform. One of the require- ments is that the growth of per capita consumption should not fall below 1 percent per annum, and the funds are seen as supporting this requirement. The judgment that seems to have been arrived at is, therefore, that it is better to use at least some of the funds to support consumption during a period of policy reform (to ensure its sustainability) than to use these same funds to do, for example, infrastructural projects but without the policy reform. Now, of course, this is too simplistic a characterization of a complex program with many components, but at the margin of use of funds this must be the decision that has actually been made-implicitly or explicitly. Although the validity of this choice cannot be assessed without detailed analysis, the general considerations outlined above would seem to support this prioritization in the use of funds. IV. CONCLUSIONS This paper has considered the interaction between projects and policy reform, using the analytical framework of conventional general equilibrium analysis. Although at the most general level of analysis sharp results are difficult to obtain, we can state some specific conclusions as follows: * Policy reform that reduces distortions in an economy also increases the social profitability of a range of typical projects. 412 Projects versus Policy Reform * Projects that increase supply responsiveness in a particular sector strengthen the argument for reducing price distortions in that sector. * Policy reform packages can and should be thought of as "projects" that have net return profiles. The rate of return on policy reform packages can and should be calculated from the standard "with policy" and "without policy" projections that are normally done. * Policy reform packages, because they have economywide consequences, are much more likely to cause losses to politically powerful groups than is the typical project. If the political economy constraints are binding, and some form of compensation is required, then the short-run financing needs are likely to be relatively greater for policy reform, at least on this score. In addition, price reform can lead to unemployment of resources in the short run, as the economy adjusts; these costs are akin to the up-front capital costs of a project. * The framework developed in this paper, which insists on the discipline of thinking about policy reform packages in terms of projects, may help clarify the choice between adjustment and investment lending. But, of course, it does not necessarily lead to any conclusions on the global balance between these two types of lending. All one can say is that if there are opportunities for substantial net benefits through policy reform, but this reform has large short- run costs, then adjustment lending is appropriate. To the extent that a coun- try has already advanced considerably in its policy reform (so that the oppor- tunities for further benefit are lower and the up-front costs are not as high), then the balance should shift away from lending for policy reform to project lending. REFERENCES Binswanger, Hans. 1989. "The Policy Response of Agriculture." In Stanley Fischer and Dennis de Tray, eds., Proceedings of the World Bank Annual Conference on Develop- ment Economics 1989. Washington, D. C.: World Bank. Boadway, Robin. 1975. "Cost-Benefit Rules in General Equilibrium." Review of Eco- nomic Studies 42: 361-73. Braverman, Avishay, and Ravi Kanbur. 1987. "Urban Bias and the Political Economy of Agricultural Price Reform." World Development 15 (September): 1179-87. deJanvry, Alain, Andre Fargeix, and Elisabeth Sadoulet. 1990. "The Political Economy of Stabilization Programs: Growth, Welfare, and Sustainability." Department of Eco- nomics, University of California, Berkeley. Processed. Diamond, Peter, and James A. Mirrlees. 1971. "Optimal Taxation and Public Produc- tion." American Economic Review 61: 8-27, 261-78. Dreze, Jean, and Nicholas Stern. 1987. The Theory of Cost-Benefit Analysis." In A. J. Auerbach and M. Feldstein, eds., Handbook of Public Economics. Amsterdam: Elsevier. Gersovitz, Mark. 1990. "Transportation Policy and Pan-Territorial Pricing in Africa." University of Michigan. Processed. Kanbur 413 Harberger, Arnold. 1964. "Taxation, Resource Allocation and Welfare." In The Role of Direct and Indirect Taxes in the Federal Revenue System. Princeton, N. J.: Princeton University Press. Kanbur, Ravi. 1987. "Measurement and Alleviation of Poverty: With an Application to the Effects of Macroeconomic Adjustment." IMF Staff Papers 34, no. 1 (March): 660-85. . 1988. "Poverty and Adjustment: A Conceptual Framework." Fiscal Affairs Department, International Monetary Fund. Processed. . 1990. "Poverty and the Social Dimensions of Adjustment in C6te d'Ivoire," SDA Working Paper Series, no. 2. World Bank. Kanbur, Ravi, and Gareth Myles. 1990. "Policy Reform and Political Constraints: An Economic Analysis." Discussion Paper no. 100. Development Economics Research Centre, University of Warwick, U. K. Kouwenaar, A. 1988. A Basic Needs Policy Model: A General Equilibrium Analysis with Special Reference to Ecuador. Amsterdam: North Holland. Little, Ian M. D., and James A. Mirrlees. 1969. Manual of Industrial Analysis in Devel- oping Countries, II, Social Cost-Benefit Analysis. Paris: OECD. Neary, J. Peter. 1982. "Intersectoral Capital Mobility, Wage Stickiness, and the Case for Adjustment Assistance." In Jagdish Bhagwati, ed., Import Competition and Response. Chicago: University of Chicago Press. Squire, Lyn. 1989. "Project Evaluation in Theory and Practice." In Hollis Chenery and T. N. Srinivasan, eds., Handbook of Development Economics. Amsterdam: Elsevier. World Bank. 1989. "Adjustment Lending: An Evaluation of Ten Years of Experience." Country Economics Department, Policy and Research Series, no. 1. Processed. .1990. World Development Report 1990. New York: Oxford University Press. PR OCEE DIN GS OF THE WO RLD BANK ANN U A L C ONFE REN C E ON D E V E LOPM ENT E C ON OM ICS 1 99 0 COMMENT ON "PROJECTS VERSUS POLICY REFORM," BY KANBUR Enzo Grilli Ravi Kanbur addresses three questions: (1) In what way does policy reform affect project evaluation? (2) What are the effects of projects on policies? and (3) Is there a way for borrowers and lenders to choose at the margin between projects and policy reform? In addressing these questions, the author uses a stylized general equilibrium model in which projects are represented by changes in the production pattern of the public sector and policies by changes in taxes. In the author's analytical framework, the effects on welfare of any change in the pattern of public sector production (that is, project financing) and of any varia- tion in the pattern of taxes (that is, policy reform) can be neatly examined. One can look at the effects of a project in isolation or of a policy reform in isolation, or one can look at possible combinations of project and policy reform consistent with equilibrium in the economy. Within the framework he employs, Kanbur shows that in a number of stylized cases policy reforms that reduce distortional taxation also tend to increase the social profitability of projects. He also shows that projects that increase supply responsiveness in the productive sector strengthen the case for policy reforms. Finally, Kanbur makes a case for evaluating policy reforms in cost-benefit terms, in the same way as projects, and he attempts to show on this basis that a choice can be made between "investing" in projects or in policy reforms. I believe that the paper provides satisfactory, if rather restricted, answers to the first two questions. On the first, "how policy reforms affect project evalua- tion," the author's answer is in essence the following: in the presence of distor- tions in the economy, measured in terms of the marginal welfare cost of public revenue, if a project affects private incomes, as most do, then the value of the private gains must be adjusted by the shadow price of public income. The author goes through a series of stylized cases to make the point that policy reforms tend to increase the social profitability of projects. This result is familiar to us from the work of Little and Mirrlees (for example, Little and Mirrlees 1974). The net social profitability of a project can be decomposed in their approach into the standard "efficiency" net benefits and the net incremental cost of private con- sumption (using public income as the numeraire). The lower the marginal social Enzo Grilli is director of the Economic Advisory Staff in the Office of the Senior Vice President for Operations at the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 415 416 Comment cost of raising revenues, the lower would be the shadow price of public income. By lowering the shadow price of public income, a tax reform would reduce the penalty on projects that increase private incomes. Kanbur stresses the cost of raising revenues, whereas Little and Mirrlees, assuming a strict revenue con- straint, stressed the marginal value of public expenditures (a more general dis- cussion of the different ways of deriving the shadow price of public income is in Ray 1984). Another feature of Kanbur's examples is the neglect of distributional issues. This detracts from the interest of his story regarding his question one, because in that case the marginal social cost of taxation will depend on which income group pays the tax, and the cost of additional consumption will depend on who gains. To make his "story" even more interesting, Kanbur might have noted how a distorting policy environment is likely to induce the public sector to choose the wrong public sector projects. For example, if the prices of agri- cultural goods are too low relative to those of manufactured goods, say, because of differential tariff protection, the government would be likely to invest more in the manufacturing sector (just as the private sector would do). Recent history suggests that this has indeed happened. On Kanbur's question two, "how do projects affect policy reforms," in the sense of helping or hindering them, the author shows that within his model the familiar dependence of the welfare gains from tax reforms on supply and demand elasticities holds. Then he takes the case of infrastructural projects and suggests that greater supply elasticity brought about by such projects will help the effectiveness of policy reforms by making adjustments easier. This inference is of course correct, and its importance is borne out by our experience. Other types of linkages readily come to mind-for example, education and retraining projects that increase labor supply elasticity or even improve the quality of management of the economy. However, the neglect of income distribution effects makes the illustrations of the effects of projects on policies rather restricted. On the third question, regarding choices between projects and policies, Kan- bur's answer does not fully reflect the difficulties involved. The core of what he is saying, I believe, is that governments and lending agencies should calculate the rates of return to policy changes just as they do for projects; this is because policy changes have costs as well as benefits, and a cost-benefit framework of the Little-Mirrlees type should be readily applicable. This is quite straightfor- ward because, as the optimum tax literature of Diamond and Mirrlees (1971) and others has shown, the rules for evaluating tax changes are the same in principle as the rules for evaluating projects. There are a number of serious conceptual and practical difficulties here, how- ever, that do not appear in the author's treatment: * The identification of the costs of policy change is not at all easy in the general case; this is not simply a question of compensating some identifiable losers, such as those who lose jobs when inefficient public-sector enterprises are forced to close down. Policy reforms create obvious and not so obvious costs Grilli 417 throughout the entire economic system. Compensations, therefore, often are not paid in full. Even if they are, there are other costs in getting policy changes enacted. Much depends on the political and institutional setups, and it is diffi- cult to think of a separate budget category that governments carry for "policy reform costs." * There is also the problem of allocating costs to particular policy changes. The costs of getting tariff reductions on some commodities, once undertaken, may also permit several other policy changes. The whole set of policy changes then need to be grouped together for evaluation. * It is also not obvious that policy changes always incur costs, even in the simple sense of a budgetary loss. Reductions of the marginal income tax rates have been implemented in some countries in the expectation that revenues will increase along with work incentives. Reductions in near-prohibitive tariffs need not involve any net losses of customs revenue if smuggling is thereby reduced. In fact, the policy distortions have often been so great-in exchange rates, for example-that getting rid of them may well lead to a time profile of positive net benefits, even in the short term. * Just as the evaluation of projects for a group of interdependent activities is often difficult-for example, in electric power sectors with integrated grids or in railways-the evaluation of a policy reform program, containing within it both a large number of price changes as well as projects, will be difficult. Note that all the major policy reform programs of recent years, for example in China, Mex- ico, and Turkey, have involved large economywide changes in prices and taxes. Although I accept Kanbur's point that the basic approach to these issues is similar to project evaluation in the general sense that the principles of cost- benefit analysis should apply, I have a different perception of what is feasible and of what can and cannot be requested of governments and lending agencies. It may be useful in this context to distinguish between the choices open to governments and to lending agencies. Using Kanbur's notation, the govern- ment's concern is to design a policy program consisting of vectors of dt and dz, that obey economic constraints (such as public revenue and balance of pay- ments) and political constraints (support, compensation). The lending agency, in contrast, is concerned with how to best support a country's program through an array of lending instruments. Obviously, the lending agency does have to decide whether the government's program is worth supporting as a whole. For reasons I have discussed above, this is often a complex matter, and only rough judgments are typically possible. The World Bank, for example, prepares mac- roeconomic projections with and without policy reform programs to gain a broad idea of the benefits in terms of extra growth of national product and per capita consumption. But beyond this, the lending agency's perspective often differs because it is typically involved with a subset of dz and dt. Different agencies, such as the World Bank and the International Monetary Fund, typ- ically address different aspects of the overall program. Moreover, different loans from the same agency may deal with an interrelated set of policy measures. 418 Comment Consequently, rates of return to the changes supported by a particular loan would not normally make sense. To conclude, let me say that I admire Kanbur's efforts to deal formally with a set of difficult and relevant issues. I think that he has helped us all to look more systematically at some of the key choices that need to be made and that we make almost daily, sometimes without thinking about them in the way we should. REFERENCES Diamond, Peter A., and James A. Mirrlees. 1971. "Optimal Taxation and Public Pro- duction." American Economic Review 61: 8-27, 261-78. Little, Ian M. D., and James A. Mirrlees. 1974. Project Appraisal and Planning. Lon- don: Heinemann Educational Books. Ray, Anandarup. 1984. Cost-Benefit Analysis: Issues and Methodologies. Baltimore, Md.: Johns Hopkins University Press. P ROCE E D ING S OF TH E WO R L D BANK ANN U AL CONFE R E NCE ON D E VE LOPM E N T E C ON OM I C S 1 9 9 0 FLOOR DISCUSSION OF THE KANBUR PAPER A World Bank participant said that Ravi Kanbur's paper inappropriately charac- terized the Bank's project lending and policy lending as separate activities. He thought the Bank had always used its project loans to try to facilitate policy reform. The Bank had learned in sector lending that policy reforms took a long time-seven to ten years was not unusual-and was a difficult process with many bureaucratic and political obstacles. Policy reform through sector lending was sometimes made even more difficult because decisions were made not at the sectoral level but at a higher level of government. Loans for structural adjust- ment could be disbursed faster, usually over two or three years, and their real merit was that policy reform decisions were now explicitly made at the highest levels of government. The two types of lending are part of a continuous process. The real issue is not one kind of lending versus the other but how the Bank uses its total lending resources, given the lending instruments available, to make policy reforms most effective. The same participant also pointed out that readers should be clear about the Lyn Squire's "infrastructural" project cited in Kanbur's paper. The example of an infrastructural project that Squire uses-an irrigation scheme in which public revenues provide water free to the private sector, and for which the main issue is cost recovery-is not the typical infrastructural project at the Bank. What the Bank calls infrastructural projects-roads, ports, railways, and airports- typically have explicit user charges that generate fiscal revenues, and with such projects, referred to as utility projects in Squire's terminology, Squire's work indicated that the Bank's methodology generally underestimates the rate of return. A participant assumed that when Kanbur's paper stated that the Bank's rate of discount should be the International Bank for Reconstruction and Development (IBRD) lending rate, this was because the Bank would be the lender of last resort. Would it then imply that if the Bank is the lender of last resort to an Interna- tional Development Association (IDA) country, the rate of discount be the IDA lending rate? On this point, another participant observed that in thinking about the IBRD rate of interest, it is very important to distinguish between the rate of interest and the cost of funds; the latter can become as much as double the rate of interest because of commitment fees and cross-currency risk. This session was chaired by J. Price Gittinger, a consultant with the Population, Health, and Nutrition Division of the Population and Human Resources Department of the World Bank. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 419 420 Floor Discussion of Kanbur Paper Kanbur responded to comments by Enzo Grilli (discussant) about the practi- cal difficulty of evaluating policy lending. Kanbur felt that when the World Bank prepared projections and drew up scenarios to justify policy reform packages, Bank staff did somehow decide that it was a good loan. Surely these projections could also be used to calculate rates of return. Although conceding that there would be practical difficulties, Kanbur suggested that at least conceptually, both project evaluation and policy evaluation were desirable. Kanbur responded to another participant and agreed that he also held project and program lending to be on a continuum, and therefore both should be evaluated similarly. However, he maintained that it is not just a matter of the sequencing of program and project loans; if resources are constrained, there is also a question of where the marginal unit of lending goes. On the lending rate, Kanbur said that if it is an IDA loan, then one should use the IDA rate. Gittinger (chair) thanked the participants and brought the session to a close. PROCE E D INGS OF T H E W O RLD BAN K AN N UAL C ONFE RENCE ON D E VE L O PME N T E C ON O M I C S 1 9 90 ROUNDTABLE DISCUSSION Development Strategies: The Roles of the State and the Private Sector The roundtable discussion at the end of the conference sought to highlight the panelists' views on the roles of the state and the private sector in development. The panelists were Amartya Sen, professor of economics and philosophy at Harvard University; Nicholas Stern, professor of economics at the London School of Economics; Joseph Stiglitz, professor of economics at Stanford Uni- versity; and Stanley Fischer (moderator), vice president, development economics and chief economist at the World Bank. AMARTYA SEN This is a good moment to look both at the connection between the public and private sectors and at the role that the public sector can actually play. To begin with, I would like to make some observations on the theme as stated for the roundtable "The Roles of the State and the Private Sector", but I would, after that, also like to raise a question about this way of seeing the problem. We know that the private sector has various shortcomings in dealing with problems of the type discussed a lot in the literature: public goods; situations of strong externalities; possibly large inequalities in the distribution of incomes, achievements, and freedoms; and so forth. What is interesting to note in terms of this general theoretical background is that when one looks at different coun- tries with different types and extents of public sector activism, one finds that some countries have benefited considerably from such activism. This is impor- tant to mention because of the failures of public-sector-oriented economies, especially in Eastern Europe, which have been so widely discussed recently. One way of presenting the problem is this: if you took all the developing countries and looked at performance measures of what they had achieved over a certain period-not just in terms of growth in gross national product (GNP) but increasing living standards, say, increasing life expectancy-the results would show a very peculiar mixture of performance. In our recent book, Jean Dreze and I used the criterion of under-five mortality (taking note of both infant and The panelists' comments, taped at the conference, have been reviewed by them for publication. © 1991 The International Bank for Reconstruction and Development / THE WORLD BANK. 421 422 Roundtable Discussion child mortality) and looked at the percentage reduction brought about between 1960 and 1985 (Dreze and Sen 1989). We found that the group of top ten countries in this ranking includes some in which the private sector has been immensely powerful and a major engine of expansion, raising GNP per head very fast, but where much resources and efforts have been put into public health and public education. However, the group also includes some that have remained poor in terms of GNP, but that have benefited significantly, especially in health and education, from the public sector. The reduction in child mortality we are talking about, ranging from 71 to 83 percent, is dramatic in this period. Of the top ten countries, there are five in each of these subgroups. On the growth-utilization side are Hong Kong, the Republic of Korea, Kuwait, Singapore, and the United Arab Emirates. On the pure public support side are China, Costa Rica, Cuba, Jamaica, and-rather interestingly-Chile. Chile is in that category because although its growth rate has been low, its policy interven- tions have continued to be very strong, particularly in health and infant care, a fact not widely known. Incidentally, the latter group with strong public sector interventions would have also included Sri Lanka if we had started not at 1960 but earlier, because the health transition in Sri Lanka took place in the 1940s and 1950s, when public health arrangements were dramatically expanded and when the system of free and subsidized rice distribution was initiated. Between 1940 and 1960 Sri Lanka's death rate went down from 20.6 percent to 8.6 percent. A region that would have also been included in the second subgroup if it had been a country is Kerala, but being a state in India, it was excluded. So, in terms of at least some of the criteria to which we might attach impor- tance, health and education particularly, the public sector's performance has been significant. This is not only the case in China, Costa Rica, Cuba, and so on; it is also the case in some of the countries with active private sectors and with high growth of disposable incomes. Nearly all of these countries have devoted a tremendous amount of growth-generated resources to the public sector for health, education, and social security. The positive role of using growth- generated resources for skillful exposition of the public sector linkage is brought up very sharply if you contrast the experience of the countries with this linkage with the experience of some other countries that were already rich and have become richer in this period but that have neglected this linkage. The contrast between, say, Korea and Brazil, and between Kuwait and Oman, is quite strik- ing. The point is that even for the cases of high growth performance coupled with substantial private sector success, one sees the fruitfulness of using growth- generated returns as a means for expanding social security and the quality of life through the public sector. China presents a very interesting set of contrasting experiences. During the pre-reform period between the Revolution in the late 1940s and the economic reforms of 1979, the Chinese experienced very little expansion of food output Roundtable Discussion 423 per head, and also very moderate increase in GNP per head (not according to World Bank estimates, but we know that there are tremendous problems with those estimates). However, in contrast to the low growth of GNP per capita, the reduction in mortality was very high. Life expectancy in China went from being close to forty years to the high sixties in that period. In contrast, in the post-reform period, a time during which economic reforms and incentives had a dramatic effect on the expansion of agricultural and food output (agricultural output nearly doubled between 1979 and 1986), the fast expansion of life expectancy was halted. The Chinese official statistics indicate that mortality reached a floor in 1978-79 and has been higher every year since then. Now, I think, the Chinese mortality statistics are questionable. There is the problem of the aging of the population, and there is also the problem of increased coverage. But even after these adjustments are made, and even if one does not accept some of the early calculations that indicated that life expectancy had actually fallen substantially since the reforms, there is no doubt at all that the sharp decline in mortality has been halted and there has been a considerable decline in public health services. The reasons for this are not altogether unknown to standard economic the- ory. One is dealing with the problem of financing of public goods. Communal agriculture did little for agriculture, but it did a lot for public health, for reasons that are not surprising. Conversely, the remedy that was so successful in raising agricultural output-to wit, privatization-did not work so well when money for public health became scarce and when patients had to pay individually for medical care, when social medical insurance coverage (largely related to commu- nal agriculture), which at one stage covered 85 percent of the rural work force, fell to 15 percent. Perhaps better coordination between the public and private sectors could have made a difference. On the coordination issue, one of the things that Dreze and I show in our book is the very effective famine relief the public sector can ptovide in conjunc- tion with the private sector. The public sector, through programs such as cash for work, can do this by creating employment, recreating entitlements lost as a result of the disaster, and thereby generating income. This can be effectively combined with use of private food trade in the affected region. This combina- tion of public-sector-based income generation and private-sector-based food trade (provided there is also a credible threat of a public sector intervention to prevent supply manipulation of the market) can be very effective. Its effective- ness contrasts sharply either with the public sector taking over famine relief completely (as has often been tried, without much efficiency, especially in Sub- Saharan Africa) or with leaving famine relief entirely to private initiative (which has been also tried, with unmitigated failure in famines across the world). This brings me to the concern I expressed at the beginning of my remarks about seeing the problem as one of the relationship between the private sector and the state. My concern is that the view suggested by the title of this discus- 424 Roundtable Discussion sion, "The Roles of the State and the Private Sector," leaves out a major partici- pant, namely the public. This worry is not a fuddy-duddy, romantic one. The public can play an important political role in putting pressure on state policy, and on the private sector-directly or through state policy. This is, of course, partly related to the issue of democracy. The connection is most immediate in the case of famine and famine prevention. Major famines have taken place in market economies and in nonmarket socialist economies, but not in any country with a democratic system, with opposition parties, and with a relatively free press. This is a remarkable fact. Famines are, obviously, terrible for the people who suffer or die, but they are often not very expensive for nondemocratic governments, which can survive famines. This applies both to nondemocratic left-wing governments supported by the Soviet Union and to nondemocratic right-wing governments supported by the West; many examples of each can be given from Sub-Saharan Africa. It seems to me that the absence of democracy and public pressure is at the root of many policy failures in development strategy. The connections are not always obvious, nor always strong or immediate. But some links can be seen. Countries such as Botswana and Zimbabwe, which have done well in preventing famines, have a relatively better record in terms of having opposition parties, stronger media, and public opinion pressure. The picture for India is fairly clear, and the contrast between the pre- and post-Independence India is very sharp indeed. There were famines in India right up to Independence; the last one was in 1944-the so-called great Bengal famine. There has been no famine in India since Independence. This is obviously partly due to agricultural achievements, but to a great extent, it is because democracy makes famines much more expen- sive for governments. A government that can be criticized by opposition parties and by the media and that has to seek reelection by the people cannot afford to have a famine. It has to take quick action to prevent it. The contrast between India and China is also quite sharp with respect to famines. China's economic record is in many ways very much better than India's. Nevertheless, the famine that took place in China between 1958 and 1961, with various estimates of the death toll ranging from 23 to 30 million, makes it arguably the largest recorded famine in history. It would be very hard for such a famine to take place in any country where the government has to face elections, and the newspapers are free. Chinese newspapers did not criticize the government-indeed did not even cover the disaster. The government main- tained, without a basic revision, the same disastrous policies through the three years of the famine. Public pressure and a free press have a creative role to play, not just in preventing major disasters such as famines but also in making social security programs less fragile. If China had had opposition parties and free newspapers, the abrupt and widespread withdrawal of social medical insurance in the rural areas after the economic reforms of 1979 would not have been that easy to put through. Roundtable Discussion 425 If I had the time, I would go on to argue that some of the criticisms of the "softness" of the state, and the corollary view (characteristic of early writings on development economics) that the "harder" the state the more effective it must be, are dead wrong. Quite often what appears as softness is the responsiveness of the state to the public asserting itself and demanding that the state should take heed of the public's welfare. That need be no bad thing. NICHOLAS STERN I will focus on three things. The first is the meaning of "public" and "private" in today's discussion. The second concerns the reasons for state action. And finally, I will look at the consequences of state action. I will argue that an examination of the reasons and consequences gives us strong pointers as to where and when the state should be involved. First, an insistence on discussing the meaning of public and private is not just the act of the evasive academic stalling for time who says, "It all depends on what you mean." The definitions really do matter. Different aspects of "public- ness" and "privateness" have different economic implications. To focus on just some of these aspects can divert us from the key issues for economic perfor- mance and social welfare. Ownership is typically seen as the defining characteristic of public and pri- vate, but ownership is not, by itself, uniquely defined. Let me briefly list four of the many aspects of ownership: the right to manage for certain purposes; the right to income arising from the use of property; the power to transfer property; and the right to exclude others. One could go on with this list, but I want to make the point that measuring ownership in order to say what we mean by public and private is not as simple as it sounds. Let me illustrate why these things matter. I will focus on the first and second definitions I mentioned-the right to manage for certain purposes and the right to income arising from the use of the property. In passing, however, let me note that the right to exclude others is not always a feature of private property. British farmers, for example, do not have the right to stop people entering their farms. Similarly, for the public sector, the government does not have the right to exclude anyone from public roads, except under very special circumstances. So what about the right to manage? I will give three examples that underline the importance of the right to manage in the performance of firms and how variations in these management rights can influence firm behavior. First, take firms in socialist economies where output planning is changed to a form of planning that gives the firm some discretion. That is exactly what happened in the mid-1980s in China. It was not a change in ownership, in the conventional sense of selling the firm or changing the rights to income in any major way. It was a change in the scope of management rights, and the firms that now had greater discretion in managing their own activities made a significant leap in productivity. 426 Roundtable Discussion One cannot, of course, completely separate the right to manage from the right to residual income, because changes in the right to manage need some accom- panying incentives. Managers will not become more efficient unless there is also some incentive for them to do so. The incentives involved, however, can be quite a small proportion of total profits. The second example of the right to manage-or the lack thereof-is the borrowing of public-sector corporations, which is treated differently in different countries. In the United Kingdom, when we still had a substantial public sector, there were considerable restrictions on the borrowing of public firms. The restrictions arose because the borrowing of these corporations was counted in the public sector's overall borrowing requirement, and the government focused on the public-sector borrowing requirement as an instrument of control. One could argue that for the British railway and telecommunications systems prior to privatization, these limits on their right to manage really had to do with a peculiar accounting convention, and the focus on a particular overall number. And yet they greatly damaged the infrastructure of the United Kingdom. In contrast, borrowing by public corporations in France is not counted in the same way. Borrowing by Electricite de France is not part of public-sector borrowing, but that by the electricity industry in the United Kingdom is, at least until its planned privatization goes through. So what we have then is an administrative control on the right to manage quite dramatically affecting economic performance. In many ways, one can argue that there was no need to privatize these corporations to remove the controls. The last example concerning the right to manage I would like to give is from India, where Wade's analysis of the control of irrigation at the village level in Andhra Pradesh shows how communal management could produce quite posi- tive and efficient results in allocating water, even though the irrigation system was publicly owned at a much higher level (Wade 1988). These examples-the relaxation of output planning, public-firm borrowing requirements, and communal control of irrigation-show that the right to man- age can vary a great deal within the public sector even without a change of ownership, and these different rights can make a big difference to performance. Let me now turn to the meaning of public ownership and the right to income as defining what is public. This perspective of the right to income can show us that public ownership in some ways is very much more extensive than it might appear at first sight. A 50 percent profits tax with a full loss offset provision is equivalent to the government taking a 50 percent share in the firm. Hence, taxation of this kind is akin to public ownership. In fact, it may be that when the government underwrites firms it may have to bear 100 percent of the losses even though it gets only 50 percent of the profits. The upshot of all this is that we have to look at public ownership very closely. It has many different dimensions, and looseness in definitions can divert us from Roundtable Discussion 427 some of the more important aspects. Later on, I am going to emphasize the importance of competition, because the history of British privatization suggests that it is not so much public or private ownership that matters-it is the compet- itive environment in which the firm operates that really counts. I now come to the second issue I want to focus on: the reasons for state action. The first reason is from standard welfare economics, where we point to market failure. Among the reasons for government intervention are externalities and public goods, and missing markets, in addition to conditions leading to the violation of perfectly competitive behavior such as imperfect information, increasing returns, and entry barriers. These reasons, we know, have to be set against the possibilities of government failure. The second reason for state action concerns poverty and deprivation, to which Amartya Sen referred. Perceived responsibility to alleviate poverty can provide strong grounds for state action involving social security. The third reason for government intervention concerns rights. Many, includ- ing myself, would argue that individuals, as part of their citizenship or involve- ment in society, also have certain basic rights, including, among others, certain aspects of education and health. Some others-though I am not sure about this one-extend the notion to the right to housing as well. Many discussions of provision for the disabled turn on the notion that it is wrong to so organize affairs that the disabled are excluded from the "right to participate" in society and economy. Questions of basic rights and equality of opportunity again point to government action. Fourth, we have what you might call paternalism, where individual prefer- ences are overridden by the government. Drugs are clearly an important exam- ple, when an individual's preferences for drugs are overridden. But I also think compulsory pension schemes are another example in which the government acts to reinforce the "higher" self against the "lower" self. Last, there is reason for government action arising out of accepting responsi- bility for future generations, which current generations, left by themselves, might not take. Concerns such as forestalling global warming, conserving the rain forests, and protecting endangered species fall in this category. We now come to the consequences of state action. A popular view is that when governments intervene (through quotas, prohibitions, restrictions, and the like), all you get is rent-seeking and unproductive activities. The costs may be very large, in contrast to those associated with traditional calculations of dead- weight losses (usually in the context of taxation). Rent-seeking is no doubt important, but from an empirical point of view, it is very hard to know what the costs of rent-seeking are. Most estimates of rent-seeking measure the size of rents, and then assert that the cost of rent-seeking, in terms of the resources used up in competitive markets, is equal to the size of rents. This rests on the rather dubious assumption that rents are competed for in perfect markets. But in fact the markets for rent are very clearly not competitive. The rents earned by the 428 Roundtable Discussion Marcoses in the Philippines clearly were not rents for which there was open competition. Government actions may or may not be beneficial. I want to focus on a few examples that show how economic theory and policy experience can combine to help us in understanding when governments should intervene and what limits government action. The first example-one I first looked at more than twenty years ago-is the Kenya Tea Development Authority (a successor to crop development authorities promoted under the colonial government). Smallholders grew tea, the govern- ment organized the activity, private factories processed the tea, and it was sold in London. Quality control was exercised through the management of market- ing outlets. The government in this particular project took responsibility for roads, for agricultural extension, for information and teaching of the peasant tea growers, and for coordination. These were all areas where economic theory tells us that there are good reasons for governments to act-in areas where markets might fail-and my cost-benefit analysis of the programs in the late 1960s was in line with the common judgment that the initiative was very suc- cessful. The second example of state action and its consequences I want to describe concerns social security in Maharashtra State in India. There, the Employment Guarantee Scheme has been quite successful in protecting the unemployed against destitution. One of the central reasons the scheme has worked is that it was "incentive compatible," and participants were self-selected. In order to receive the payment, one has to present oneself for employment at not very attractive wages. This aspect helps overcome the problem of adverse selection. In contrast, other poverty alleviation schemes in India, such as the Integrated Rural Development Programme, that have tried to make subsidized loans to people on the basis of their incomes-whose measurement is easily manipulated-have performed much less well. My third example has to do with privatizations in the United Kingdom. The most successful privatization in the United Kingdom, in terms of efficiency and performance, involved commercial areas in which publicly owned firms had to compete in the marketplace to get customers. I am speaking of entities such as Cable and Wireless, Jaguar, and British Aerospace. Less successful in terms of performance and more troublesome in terms of regulation has been the privati- zation of natural monopolies-British Gas, British Telecom, and, more recently, water and the soon-to-be privatized electricity industries. The primary lesson here then is that competition, rather than who actually receives the profits, is the more important factor in performance. Fourth, we may draw attention to the very large experience of development projects in the World Bank over the last decades and their history and documen- tation, particularly by the Operations Evaluation Department. Roads and irri- gation, classic examples of infrastructure, seem to have been rather more suc- Roundtable Discussion 429 cessful than other areas. I know that analysis of these and other development projects is well under way, and it is something those of us outside should greatly encourage and watch with enthusiasm. Fifth, I would note that a number of authors over the last two decades have attempted to provide a long-run and comparative perspective of growth performance-in a sense, to set out in broad terms the lessons of history regard- ing the appropriate degree of intervention (see, for example, Chenery, Robin- son, and Syrquin 1986; Morris and Adelman 1989; Reynolds 1983). In general, the analysts avoid asserting that there is a unique best strategy. Some countries, such as China and Korea, have performed well with a great deal of government intervention; others, such as Hong Kong, seem to have flourished with a mini- mum of government interference. Sen has argued strongly here and elsewhere that democracy may be of great importance in preventing famine, but it would be hard to be confident from comparative history that democracy is good for growth. Has it been democracy that has propelled Hong Kong and Singapore? The evidence appears to be ambiguous. Where does all this lead us? We should not delude ourselves that transition is easy, or indeed that the same formula is appropriate to all countries. A careful and creative economic analysis of potential reforms should leave us a number of options and not a single economic blueprint. Nonetheless, I would argue that the examples I have cited illustrate a conjunc- tion between theory and experience and give us fairly direct answers about when governments ought to intervene. From the point of view of income distribution and protection, governments ought to be active in social security. From the point of view of rights, they ought to be active in education and health. From the point of view of market failure, they ought to be active in infrastructure, in roads, power, and so on. Where governments should not be active-because none of these arguments apply-is in the private production of hair pins, motor- cycles, motor cars, and the like. Let me conclude with some speculation on how the relative roles of public and private sectors might move as development progresses. In poor and backward economies, market failures may be more severe, but the same is also true of government failure. How do these considerations balance? I would suggest, notwithstanding the weakness of management and the scarcity of resources in poor countries, that there are certain crucial activities for which the government should take responsibility. In poor economies, infrastructure, health services, and education are usually weak. Here, surely, are areas where government can act and be assisted with resources and know-how from wealthier countries and international institutions. Governments obviously need a tax system to finance their activities. That is a subject for another day, but I would just like to note that developing countries do seem to be learning quite a lot of lessons about taxation, and it is impressive how much in fact they do collect. 430 Roundtable Discussion JOSEPH STIGLITZ The burden of my remarks is to deal with what economic science has to say about the appropriate roles of the state and the private sector. It is an issue on which most economists will have a strong opinion. But if you ask a taxicab driver, he will also give you an opinion. The question is, should our opinions as economists be listened to more carefully than those of the taxicab driver? One can attempt to address the question on grounds of theory and on grounds of evidence. I am basically a theorist, so I began by asking the question: do we have any theorems that provide insights into the question? Well, there are two old theorems that speak to the issue, and two new theorems. One of the old theorems-the Lange-Lerner-Taylor theorem-says that mar- ket socialist economies and private capitalist economies are equivalent. Yet anybody who has visited a socialist economy-even people who claim to be market socialists-would express some skepticism about the theorem. Of course, a theorem only proves that certain conclusions are true based on some given assumptions. The question is, do the assumptions underlying this theorem describe either the market socialist economies or the capitalist economies? And I think there is a growing consensus that neither part of the theorem is good. The other old set of theorems are the fundamental theorems of welfare eco- nomics. These underlie the market-failure approach that motivated Stern's remarks about how to define the role of government. The theorem says that the markets are efficient, except for certain well-defined market failures, such as externalities, or the public provision of infrastructure. The view has been exten- sively developed over the past two decades, on the basis of this theorem, that the government should intervene in these well-defined areas and keep out of every- thing else. The two new theorems I want to talk about address more current issues. The first tries to explore two of the hidden assumptions in the earlier fundamental theorems of welfare economics. In particular, the new theorem asks: if we have incomplete risk or futures markets and we have imperfect information-both of which we all agree we do-what can we then say about the efficiency of market economies? The answer-in general-is that market economies are essentially never constrained Pareto-efficient. The theorem is interesting because it has taken away the intellectual founda- tions for the belief in the efficiency of a market economy. The old welfare theorem said that government, no matter how well organized, no matter how competent, could not do any better than the market. And if that is the case, then we do not have to inquire very much into the nature of government. In contrast, this theorem says that in the presence of imperfect information, incomplete risk markets, and incomplete futures markets, there is the potential for government intervention. However, the theorem does not say what governments must do to realize this potential. We have to then inquire into whether that potential can be realized. Roundtable Discussion 431 The second new theorem has to do with privatization. Much of the earlier literature did not address the central issue which has become the focus of discus- sion in the last five years. The kinds of interventions that were often talked about in the earlier literature, say, in the context of externalities, do not address the question of whether there should be government production. They only require some form of government intervention, often in the form of taxes or subsidies-that is, Pigouvian interventions of a very limited kind. The issue that has been more recently under debate is whether the government should be involved in production. Or should government leave production- that is, privatize? The new theorem, which I refer to as the fundamental privati- zation theorem, does address these issues. Like the welfare theorems, it says that private production can emulate public production, that is ideal public produc- tion, only under highly restrictive conditions. So that, in general, the two are not equivalent. This provides a weak intellectual foundation for understanding the precise roles of the government and the private sector. What it does say is that we do not have any confidence necessarily that markets are efficient, that there is a potential scope for government intervention, but it is a fairly diffuse and ill- defined role until we fill out the details. I now come to the empirical evidence on the roles of the state and the private sector. Here economists, for the most part, specialize in anecdotal evidence and tend to be fairly selective, depending on which side they are on. There are many examples of incompetent government enterprises, and those who criticize government have a rich set to draw upon to verify their prejudices. Yet there are also many successful government enterprises. I think there is a general consensus that many of the French enterprises have been very successful. Many people cite some examples in Singapore. Studies show that there is no significant difference between the efficiency of the Canadian National Railroad and the Canadian Pacific Railroad. So, if one is not quite so selective, then one can find examples on both sides of the issue, and we need to inquire in more detail what accounts for the success or failure of particular government programs. At the same time, I should also emphasize that we have a number of instances of incompetent private corpora- tions. In fact, some of the incentive issues extensively discussed in the context of public sector activities are almost as important for the private sector. In large U.S. corporations there is a separation of ownership and control. However, the takeovers in recent years have graphically demonstrated the ability of managers to walk off with millions of dollars of their companies' resources. For example, the management of RJR Nabisco in the recent takeover walked off with over a hundred million dollars of bonuses for their efforts in getting rid of the com- pany. Also, an audit showed that whenever the president of the company went on a trip, he was accompanied by a person called G. Shepherd on a separate plane. The auditors looked through the company's organization chart, but could not find any G. Shepherd. Finally, they found out it was the president's German Shepherd, who bit people and "needed" to travel in a separate airplane! I do not 432 Roundtable Discussion mean to prove anything with this story, but it does make the point that anec- dotes can also be produced on either side. The only other piece of anecdotal evidence is that governments, besides their role in the provision of infrastructure, have played an important role in almost all successful economic development efforts. So the question is not whether the government ought to play a role, but to define more precisely the appropriate role. Moving from these fairly general issues to more specific ones, I want to turn to financial markets and entrepreneurship. Well-functioning financial markets are now recognized to be central in any development effort, for which the allocation of capital is a vital function. This function is not done well at a centralized level or with a planning mechanism, because the issue is not so much the choice of sectors that ought to get funds, but the particular project and the particular managers-involving complex institutional questions-that should be selected. In the United States we know what happens if you do not have well-functioning financial markets-savings and loans institutions have misallocated an amount of capital equal to somewhere between a quarter and a half of the U.S. econ- omy's total annual saving. And that is a lot of money to be squandered. The fact that the U.S. economy has not done such a good job with its financial markets suggests that other countries-particularly developing ones-are going to have a hard time as well in the development of their financial markets. The recent literature on this has focused on the pervasiveness of information prob- lems in financial markets. In addition, these markets are characterized by credit rationing and by equity rationing, and this has important effects both on resource allocation and macroeconomic behavior. I should point out that the fact that there are such problems in the private financial sector does not neces- sarily mean that governments can do any better. We are now increasingly aware that we have to accompany an analysis of market failure with an analysis of government failure. Stern referred to the recent arguments about government activities generating rent-seeking behavior, which would completely absorb all the rents. I agree with him that this is not a general result and only holds under certain special assump- tions. Under more general assumptions, rent-seeking does take place, but will not necessarily dissipate all the "rectangles"-the areas between the demand and supply curves that arise, say, with government-imposed trade restrictions. Governments face many of the same information problems that the private sector faces, sometimes more acutely, and sometimes less acutely. We are increasingly trying to define more precisely the advantages and disadvantages of government relative to the private sector. But it is only detailed comparative analysis that will lead us to a sense of the appropriate assignment of responsibil- ity between the public and the private sector. Let me also emphasize a point that Stern raised, which is that the issue should not just be viewed as a contrast between the private sector and the government. It really is a question of the whole complex of relationships among market, Roundtable Discussion 433 nonmarket, government and state, and voluntary institutions, and the particular forms that the government intervention takes. My concluding comment on the issue being discussed is that the appropriate analysis should not be based just on ideology, should take into account the institutions, the nature of government, and the nature of the private sector in particular countries. This needs to be done against the background of the emerg- ing general principles about circumstances that are more conducive to efficiency in either the public or the private sector. Among these conditions are the impor- tance of having appropriate incentive structures-including subjecting markets and firms to competition-either in the private or the public sector. I also agree with Sen's earlier comment that these conditions also should be embedded into democratic societies that would then provide checks and balances that are important to a successful development strategy. FLOOR DISCUSSION A participant asked the panelists whether the general principles they had enunciated about the roles of the private and public sectors should vary across different stages of development. What does the historical evidence tell us about it? Another participant expressed disappointment at the impression Stiglitz had given that all economists have for empirical evidence are anecdotes, and one can choose one's own anecdotes to make one's case. The participant had expected the speakers to go beyond the well-known welfare theorems and draw on his- tory, which over the last forty years or so offers a wide selection of development strategies and performance. A World Bank participant said that the stark juxtaposition of the private sector and the public sector overlooked the role of private voluntary organiza- tions and the nongovernmental organizations; within the public sector itself, it also ignored various levels, such as national, state, and local governments, and community organizations. Another Bank participant noted that apart from the question of ownership, one element missing from the discussion was the ques- tion of who sets the rules of the game, and the public sector had to be involved in that. Stiglitz responded that the problem with history is that different people read it differently. Some general principles, however, can be gleaned from the historical record. First, there is an important role for the government in setting rules for the private sector, including incentives, private property laws, and contract law. Second, competition is more important than the private-public division; giving a private firm a monopoly is not likely to improve efficiency as much as pitting a public enterprise against an open economy and subjecting it to competition. The general implication from reading history is that ownership is not as important as the environment in which a firm operates. On a related issue, Stiglitz said there was at one time a concern that in the 434 Roundtable Discussion early stages of development the state would have to intervene to promote entre- preneurship, since it would not otherwise be forthcoming. It is now recognized that innovation within state enterprises requires entrepreneurship just as much as it does in the private sector. One of the problems with state enterprises is that they often are unable to provide adequate incentives for entrepreneurship and innovation. Sen said that he agreed with Stiglitz that what appears, at one stage, to be a hard reading of the facts often looks different when one looks back. For exam- ple, in the mid-1950s, the Soviet Union's high growth rate impressed econo- mists. Sen cited Wiles as saying that "In the Soviet economy there are, as it were, always too few hairbrushes and too many nailbrushes in view of the resources available, while in a 'capitalist' economy this proportion is always more nearly right. But the production of both these articles is growing at about 10 percent per annum in the USSR and about 2 percent per annum in the 'capitalist' coun- tries. In the end the Soviet citizen will be supplied better even with hairbrushes" (Wiles 1962, p. 217; orginally stated by Wiles in Oxford Economic Papers, October 1953, pp. 315-16). Sen noted that such a comment seems extraordi- nary now, but it did not then. So although he agreed with the participant who had asked the question about the lessons of history, he felt that one needed to be cautious. One needs theory and reasoning to give depth to the brute observation of facts. Commenting on Stiglitz's presentation, Sen said that Stiglitz had drawn a distinction between information constraints and incentive constraints, and focused on the former. Sen felt that while both the issues are important, the incentives problem was the more serious of the two in looking at the perfor- mance of the public sector. He said that the Lange-Lerner-Taylor theorem also focused on the information issue-whether market socialism could emulate the information-generating aspects of a market economy. However, the problem with market socialism is not so much with signals but with more incentives and entrepreneurship. Finally, Sen contrasted the need to emphasize political as well as economic rights. Stern had focused mainly on economic rights, but this can be supple- mented. Sen reiterated that the political rights were important both in them- selves and in terms of their consequences, in giving those in authority appropri- ate incentives to be concerned with the well-being and the misery of the people. Responding to the question about how the role of the state should change in the course of development, Stern said that when countries are poor and disor- ganized, it is possible to argue that the problems of market failure, as well as the problems of government failure, are more severe. This made the question diffi- cult to answer. However, his own judgment was that in very poor countries the argument for a more active state is stronger than in richer countries. He pointed to four reasons for this: first, the state needs to be more active in providing infrastructure, such as roads; second, the state needs to help in the learning process and in accumulating knowledge and technology; third, the state has a Roundtable Discussion 435 possible role in fostering capital markets; and fourth, the problems of economic and physical vulnerability are especially severe. On the question about the lessons of history, Stern said that there is no single route to rapid growth. Most people who have looked at the question in histori- cal perspective, and without prejudice, have suggested that there are several ways to grow. The comparative growth experience does not point unam- biguously to the superiority of certain development strategies. Stern noted that the one single most important explanatory variable indicated by Reynolds in explaining long-term growth (extending over a hundred years) was the adminis- trative competence of the government (Reynolds 1983). Of course, that still leaves the question of what is to be done when such competence is weak. Fischer (chair) then thanked all the panelists and other conference partici- pants. He noted that the purpose of these conferences was to open up communi- cation between the World Bank and the academic and policy communities, and that the conference had highlighted some stimulating perspectives that varied from the standard thinking in the Bank. Fischer suggested that among the many things that the conference had raised, Sen's emphasis on the role of democracy deserves a great deal of thought, particularly on the difficult question of how it might affect the Bank's operations. REFERENCES Chenery, Hollis, Sherman Robinson, and Moshe Syrquin. 1986. Industrialization and Growth: A Comparative Study. New York: Oxford University Press. Dreze, Jean, and Amartya Sen. 1989. Hunger and Public Action. Oxford, U.K.: Claren- don Press. Morris, C., and I. Adelman. 1989. Comparative Patterns of Economic Development 1850-1914. Baltimore, Md.: Johns Hopkins University Press. Reynolds, J. 1983. "The Spread of Economic Growth to the Third World." Journal of Economic Literature 21: 941-80. Wade, Robert. 1988. Village Republics: Economic Conditions for Collective Action in South India. Cambridge, U.K.: Cambridge University Press. Wiles, Peter. 1962. The Political Economy of Communism. Oxford, U.K.: Blackwell. Subscription Coupon If you are not already a subscriber to The World Bank Economic Review or The World Bank Research Observer, you may begin a one-year subscrip- tion by completing and returning this coupon to World Bank Publications. The Economic Review is directed at an international readership among professional economists and social scientists. It publishes original research that emphasizes empirical applications and testing. The Research Observer is for everyone who has an interest in economic development. Its overviews of key issues in development economics are intended for policymakers, practitioners in the field, students, and economists seeking surveys of work outside their own specialties. Indicate below the type of subscription desired. Please complete both sides of this subscription coupon. -- @Er [N SONME$ ,~~~~~~~~~~~~~~. . . . . . . . . . ..... :,READERS IN NON-OECDCOUNTRIES AND TURKE Cilomplimentar susciptos bo"th joral reaa lab::t edesw mailing a ses .i developing countries. Complete th:' o an.d.' '.to: World Ban P C c : one.. or k bot 118 H Str t'NW Ro he Wor a '''-':., 'i x ......~~~~~~~~~~~~~~~~~~~~.... f''''".U.S.' ...... : Wsi DC 233 USA 0 T W ' : 'arc .. ' cz'wn,'in ud'i g-Xi-"' ~~~~~~~~~. . . . .. .. . . . . . . . . . . . . . , ' "." ' '.' ' ,''".. .:'" READERS 0N ECJ)COUNTRIES. --w CM s -" " ': '-'I"'d -' I " h'' ' :' Readers wihmiigadesses in, O CD contis shud opet coupon incling th meho of p4yment, j.'andmlto Th-e Economic Review 0........... : - .. .U$2, 0 US$45 . .,-i..,,Ae as: ...h Check method of payment: O Enclosed is my check, payable in U.S. dollars to World Bank Publications. O Charge my 0 VISA 0 MasterCard 0 American Express Credit card number Signature Expiration date O Bill me (institutional customers only). Include purchase order or list p. o. number Please complete both sides of this subscription coupon. Please print clearly. Name Title Organization name Organization address Country Postal code Please check only one O001 International noncommercial organizations (including UN agencies and development banks) 0 02 Governmental and nongovernmental agencies 0 05 Academic and public libraries and research institutes 0 06 University Faculty 0 08 Regular bookstores 0 °09 Commercial, financial institutions and corporations 0 12 Individuals (using home address) 0 13 Newsmedia . .... ... ... ........... ... . ... . . . .... ............ Check up to six categories 0 R0 General economics: theory, history, systems 03 R 1 Economic growth: development, planning, fluctuations 0 R2 Economic statistics O R3 Monetary and fiscal theory and institutions 0 R4 International economics O R5 Administration, business finance, marketing, accounting 0 R6 Industrial organization, technological change, industry studies 0 R7 Agriculture, natural resources, rural development 0 R8 Labor, population and human resources 0 R9 Welfare programs, consumer economics, urban and regional economics PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1990 A copy of the Proceedings is sent to current subscribers to The World Bank Economic Review or The World Bank Research Observer as part of their subscription, and is included in new subscriptions at no extra charge for as long as stocks are available. Additional copies of the Proceedings can be ordered with the coupon that follows. ISBN 0-8213-1607-9 / US$17.95 each / Order Stock #11607 Development and the Environment Living with Wildlife Wildlife Resource Management with Local Participation in Africa Africa's wildlife are a valuable natural resource currently in danger of being destroyed by widespread misuse and degradation of the land. This report examines plans that encourage the local people to use the wildlife resources effectively for mutually beneficial results. ISBN 0-8213-1697-4 / US$13.95 / Order Stock #11697 The Prospects for Agroforestry in the Tropics A variety of agroforestry practices are reviewed in this technical paper in light of their agricultural productivity as well as their environmental benefits. ISBN 0-8213-17024 1 US$6.95 1 Order Stock #11702 l'o order aniy of these putblicationis please use tlhe couIponI thlatfollows. The Economy of the USSR Summary and Recommendations This is the most comprehensive assessment of the Soviet economy to date. It recommends immediate introduction of economic reforms to shift the Soviet Union toward a market economy. This report was prepared jointly by the International Monetary Fund, the World Bank, the Organisation for Economic Co-operation and Develop- ment, and the European Bank for Reconstruction and Development. ISBN 0-8213-1768-7 / US$15.95 / Order Stock #11768 Improving the Performance of Soviet Enterprises John Nellis The growing private sector in the USSR has had a number of problems. While private enterprises are more economically efficient than state enterprises, there are still considerable constraints on private ownership and operation. John Nellis, a member of the World Bank team that participated in the 1990 Joint Study of the Soviet Economy, discusses these constraints and their effects on the privatization of the Soviet economy. ISBN 0-8213-1777-6 / US$5.95 / Order Stock #11777 Poland Economic Managementfor a New Era This World Bank Country Study presents a macroeconomic frame- woik and outlines a series of systematic reforms that would allow Poland to develop a competitive market economy. ISBN 0-8213-1650-8 / US$9.95 / Order Stock #11650 The mlost con1ven1ient anid tser-f ienidly softvare ever released b1y the Wor ld Banik World Bank Data on Diskette An increasing number of data collections published by the Bank are now available on diskette for use in personal computers. Using the Bank's newly developed Socioeconomic Time-series Access and Retrieval System (STARS), the data can b e easily exported to spreadsheet and word processing programs. Each diskette package includes double-density 5 1/4" diskettes and 3 V½" dis- kettes for use in PCs with a hard disk and MS-DOS versions 2.0 or higher. Also included is a User's Guide to the software and an English paperback edition of the corresponding book. World Tables 1989-90 Data on Diskette ISBN 0-8213-1531-5 / US$70.00 / Stock #11531 The most detailed collection of economnic data published by the Bank. World Development Indicators 1990 Data on Diskette ISBN 0-8213-1532-3 / US$95.00 / Stock #11532 Drawn from the World Development Report 1990, the Bank's most widely circulated statistical tables on economic and social development. African Economic and Financial Data on Diskette ISBN 0-8213-1529-3 / US$95.00 / Stock #11529 Designed to track the performance of development programs and aid flows in the region. New This Spring Watch for details in upcoming issues of The Economic Review and The Research Observer. Social Indicators of Development 1990 Data on Diskette World Debt Tables 1990-91 Data on Diskette Both titles are currently available in book form. To order any of these data collections on diskette please use the coupon that follows. Title Stock Number Price Subtotal US$ Shipping and handling US$ 3.50 Airmail surcharge outside USA (US$6.00 per copy) US$ Total US$ Check your method of payment. Orders from individuals must be accompanied by payment or credit card information. Ask local distributors about forms of payment acceptable to them. O Enclosed is my check payable to World Bank Publications in U.S. dollars. ECharge my O VISA [ MasterCard [ZAmerican Express These credit cards are accepted for orders addressed to Philadelphia. (Check with your local distributor about acceptance of credit cards in your country.) Credit card ac ount number Expiration date Signature Institutional customers only: Ol Send invoice and reference purchase order no. Please print clearly Name Title Firm Address City State Postal Code_ _ Country Telephone Customers in the United States: Detach this form and mail to World Bank Publications, P.O. Box 7247-8619, Philadelphia, PA 1917048619, U.S.A. Customers outside the United States: Contact World Bank Publications, 1818 H Street, NW, Washington, DC 20433, U.S.A. for a list of World Bank publications distibutors. Write to the local distributor in your area for information on prices and terms of payment. Coupons received in Washington from countries with authorized distributors will be returned to the customer. Customers in countries with no authorized distributor, return your coupon to World Bank Publications, P.O. Box 724748619, Philadelphia, PA 19170-8619, U.S.A. WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1991 Ii : April 25 and 26, 1991 Washington, D.C. CORRECTIONS TO THE PROCEEDINGS OF THE WORLD BANK ANNUAL CONFERENCE ON DEVELOPMENT ECONOMICS 1989 Page 29, fourth paragraph, sixth line: The percentages should be -50.1 and -44. Page 30, table 1: The price change for United States, coarse grains, should be -3; for developing countries, rice, should be -12; and for all, rice, should be -8. The trade volume change for EC, sugar, should be -5; for United States, rice, should be -2; and for OECD, wheat, should be -1. Page 30, table 2: First column, first row: 11. 8 should be -11. 8; third row: 11.1 should be -11.1. Second column, second row: 10.2 should be -10.2; third row: 13.1 should be -13.1. Third column, third row: 23.1 should be -23.1. Distributors of World Bank Publications ARGEN3INA Car Hinch, bRL FINLAND M33CCO O513AFRIUCA, BOrSWANA Calkori G Akalterinior KRir$ ppr INN)TFC Prrsi.Jelidk F to I 1 4th 14 or-Ok. 453/465 P.O. Bo- 126 Ap.rhdo Pol 22-F60 Dlord onwtl Pin 1333 BSto Aim SP40101 1406 ialpa, Mtesoo DR Southen Afro. HieldnkilO PD.orbll4 AUSTRALIA, PAPUA NEW GUINEA, MOROCCO C pe Ton 1141 FiJL SOLOMON ISIANDS, RANCE M 6dEOClreCpMaiisg Maroraire VANUATU, AND WESTERN SAMOA Woad b-rkPbbcati- 12 ort Mor, bd. d&Afn rrs-&-ipti-s rota DOA. Berth &Jo-als 66, ar vd C.th_ btr rnl Ssert S-re 6dgWhitbe-Road 75116Prio PO. Be 41095 Mitcham 3132 NETHERLANDs Crm1ght Virtoria GERMANY I.OrPtbha be. 20nbergS)24 UNO V.dig P.O. bc 14 AUSTRIA PeppeldormnAlA S55 7240 BA Loche SPAIN C-oI d and Co. D-53Co B-n, I M-ndi-Pnss libro, SA. CoAkbe 31 NEW ZEALAND Cootello 37 A-1011 Wie GlREECE Htills Liteay d nfolososatis os-ina 2M0i1MaUdnd KEMEi Privae Sg BAHBALN 24, Iypda eretP Plat lt p ttiros N.w Mnrkt lfbLterIta.lo dae1 AEDOS bataio tiesor-h adCo-acy Ath-o11635 AnMood CooselldoCot,31 Airdts ltd. 060)9 BonIer. P.OS B 22103 GUATEMALA NEGERIA MaoaomuaTosoo 317 Lirterias FPYm. dntart. Ihiresiry PF Li-id SRI LANKA AND THE MALDIV ES 5. Clr 7-5r IflrrHC-oa Being Jinito La}rHoeseBoohop BANGLADESH Zoo. P-vate MalBag o9 PDO. BS244 MiErobld i -volapoet Geatrsl City Ida 10, rS.Coispol-o A Asetai-ceSiedy OtIDAS) GadiroMat-attha Houe S, Ra-d 16 HONG KONG, MACAO NORWAY Cobhor 2 Dhmrotdi R/A Asi 2U0E Ltd Na hbmuti,o C-N Dhala 1239 6FL,146 Pnnao Ed-ard Boo'krDpe SWEDEN Rani,W. P.O.116125s5rE-td Prr-oglktilr: S-rls offir Kowlorn N-C602MOob 6 FPtzer Factokbsoragia MaioSoRe Hooglg Fgosg 12atal,So 10356 Mai,eCosot OMAN [-103275 oktheln Narkhnir - 3800 INDIA MINn IfoerastoBe Stuv Anid Pulh P-vat Ltd P.O. Sn 1613. Sb Awort Frriherie ortiror 76,KD.D. Avtne 751 Mot Road Mwt W-ntrrgrr-Wdhto AB Kobl. Mldro - 600 002 Box 10004 PAKISTAN 5-104 25 ktohs BELGIUM Roe9 hffir: Mie SBnkAgoty Jeer Or larroy 15 JfN. Htidi M-g5 65. u, h-hadAzem SWITZERLAND A. duRoiE202 Blld Esat- P.O.SBN. 729 F-rsi,gititk5: IO6) BRsnb Brely -4C0038 LT, 3 LibroinePsyot CANADA ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~6, ruCG-v CANADA 13/14 AsfAi Rad PERU C-rpoeL36] Le Diffuecr New Dek- 110 012 EdftoialeD-ol, SA CH 1211 Crovo 11 C.P. 85.1501B oY A.p&.e Apsdo 3624 B-haclovtlleAl Q-lb 17 Chritlraopr A-n Lin Foitii.-aiplioe soid-s j4B5 SE6 Ca -utta- 700 072 LibtAwr Payot PHILIPPINES S-desrAb-noeoont CHIN-A Jny.d- Hostel Btoldwsg L BnalookC-m-- Case,peela3312 Chime Fimdol & flonooic 5th X4ic Read G-di-ags Fiftlh F.- Filipiras LifeBeIdmg CH 100Z Lausooe Pebtbshimg Hue Sargao- 560039 Ayal AvnO MoIsim D. DFo SitDeoJi Met,aroM-nl TANZANIA Beiis 3 5-1129 Koclega Cr. Road ODfa d Uli-irty Prer Hydtetrd - 50 027 POLAND P.O. BS 529 COLOMBIA ORPAN DoaBSale Ifocnbce Ltd. P-rathslo Flab, 2nd FTor PIac Koilury i N-ela ApartdoAorA 34270 N-e Thalr Bwug NovoSopre 0-901 W wa THALLAND Bo.ota D.E Ahtobad - 30 0339 C-otsl Depareort SNer PORTUGAL 336 Sda Read CO dE"D-VOIRi Poia Hoesf Livisia Psatral B-g"k C-ntr dE tisr dtferoro 16-A Ahk terg Rue Do Con 70-74 Alir (CEDA) Le.kow- 226001 1200 Libo TRINIDAD IkTOBAGO, ANTICUA 04 B.P 541 BARBUDA, BARBADOS, Abidjk 04 Hlanu INDONESIA SAUDI ARABIA, QAT'AR DOMINICA, CRENADA, GUYANA, Pt. brdit Ltmited J-ot bSn Sto JAMAICA, MONTSE RRAT, ST. CYFPRUS JL SBarr RBarl-gi 37 PRO BS 31% KrTS & NEVIS, ST. LUCIA, MEMRB Inf- foatoros-rs PRO. BS 181 Riyadh 11471 ST.VINCENT & GRENADINES P.0. BS 2036 Jok P-t Bystiics tuds ieoLit Ni oue titMEMRBS rnfit-urtSon 5 n 69 Wats Sret rTALY Ertis sfiin C-mPe DENMARK l.sCe-oo rosaw S.-sont5PA Al Alssts TriidadW,Wetlodies tao,ieodsLitoortur Vta Bendletto FoOtri0 120/10 Al Dihr- Cet-r R -oea A1e 11 CareS. Piisl 562 First Fl-aso UNITED ARAB FMIRATES DK-19S Fdknitg C 5017Z1 Flonee PRO. BS 716B ME1I0RB GCfaCo Riyadh ' P.O. Bo6097 DOMINICAN REPUBLrC JAPAN 0usrlh EdrrorTsb, C par A. b.n Book Seers Ht Abdudll Ahtva DdiMg RSatsirl e Isabel b Ca64iha 309 37-3, Hoo-g 3-Chosr, BSeokyo-ku 113 Kng Khlhd str UNrTED KINGDOM AprtadoPoa I2190 Tokyo o .Sn356 Mi9N4iftlid. Srto Dosoorgo Dcitn P.O. B-o 3 KFNYA An, InHa eIoreiUi4 21PG EL SALVADOR Ahir BSn Ser-(E A) Ltd 33. Mdor-d Har Awad Ser Erglrd F-der PRO. Sn 45245 P.O. BS597S A-id. Maed Eo-iqe A-to 63533 Naiw'oi Jeddh VENEZUELA Edihko SISA, Ie Pise Ldt d