/ 3I 39 Vol .1 Feb. I qq THE GREENING OF ECONOMIC POLICY REFORM VOLUME II: CASE STUDIES Edited by WilfridoCruz J Mohan Munasinghe Jeremy J. Warford The World Bank Environmental Department and Economic Development Institute ESD EDI The Greening of Economic Policy Reform Volume II: Case Studies Edited by Wilfrido Cruz Mohan Munasinghe Jeremy Warford The World Bank Washington, D.C. Copyright 0 1997 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing February 1997 Environment Papers are published to communicate the latest results of the Bank's environmental work to the development community with the least possible delay. The typescript of this paper therefore has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. Some sources cited in this paper may be informal documents that are not readily available. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. The chapter "Economic Policies for Sustainable Resource Use in Morocco" by Ian Goldin and David Roland-Holst first appeared in The Economics of Sustainable Development (1995, Ian Goldin and L. Alan Wmters, editors), copyrighted and published by Cambridge University Press, and is reprinted with the permission of Cambridge Un!versity Press. Related World Bank publications include Economywide Policies and the Environment by Mohan Munasinghe and Wilfrido Cruz (World Bank Environment Paper No. 10) and Environmental Impacts of Macroeconomic and Sectoral Policies (forthcoming) edited by Mohan Munasinghe. Library of Congress Cataloging-in-Publication Data Warford, Jeremy J. The greening of economic policy reform / Jeremy Warford, Mohan Munasinghe, Wilford Cruz. p. cm. Includes bibliographical references. Contents: v. 1 Principles-v. 2. Case studies ISBN 0-8213-3797-1 1. Sustainable development-Case studies. 2. Economic policy- Case studies. 3. World Bank I. Munasinghe, Mohan, 1945- II. Cruz, Wilfredo. III. World Bank. IV Title. HC79.E5W36 1997 333.7-dc2O 95-39068 CIP CONTENTS VOLUME II Foreword . ................................................... viii Acknowledgments ............................................. x Contributors and Editors . ........................................ xi 1. Natural Resource Management and Economywide Policies in Costa Rica: A Computable General Equilibrium (CGE) Modeling Approach ......................................... 1 Annika Persson and Mohan Munasinghe Annex 1-A: Summary of CGE Model Structure ...... ............ 17 Annex 1-B: Input-Output and Land Use Data, Costa Rica ..... ..... 23 2. Evaluating Economywide Policies in the Presence of Agricultural Environmental Externalities: The Case of Ghana ..... 27 Ramon Lopez Annex 2-A: Derivation of a Land Demand Equation ..... .......... 45 Annex 2-B: Derivation of a Labor Demand Specification ..... ...... 46 Annex 2-C: Data Sources and Definition of Variables ..... ........ 47 Annex 2-D: Effects of Policy Adjustment on National Income ....... 48 3. Economic Policies for Sustainable Resource Use in Morocco .55 Ian Goldin and David Roland-Host Annex 3-A: Model Equations: Formal Structure of the Morocco Model .67 Annex 3-B: Social Accounting Matrix (SAM) for Morocco, 1985 .... 73 Annex 3-C: Results of Sensitivity Analysis .74 iii iv Volume II 4. Poverty, Migration, and Deforestation in the Philippines ..... ...... 77 Wilfrido Cruz, Herminia Francisco, and Gregory Amacher Annex 4-A: Multinomial Discrete Choice Model of Upland Migration ........................................ 95 Annex 4-B: Descriptive Statistics of Explanatory Variables in the Migration Model ...................................... 98 5. Economywide Policies and the Environment: A Case Study of Poland .................................... 101 Robin Bates, Shreekant Gupta, and Boguslaw Fiodor 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka .................................. 131 Peter Meier, Mohan Munasinghe, and Tilak Siyambalapitiya Annex 6-A: Attributes ............... ...................... 156 Annex 6-B: The Concept of Dominance and Tradeoff Curves ...... 158 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment ........... .................... 165 Zeinab Partow and Stephen D. Mink 8. Economic Policy, Wildlife, and Land Use in Zimbabwe .189 Kay Muir-Leresche and Jan Bojo with Robert Cunliffe on Environmental Impacts Annex 8-A: Relative Environmental Impacts of Wildlife .212 Abbreviations and Acronyms ................... 221 Contents v Tables 1-1: Percent of Total Land Area in Forests and Agriculture ..... ...... 14 1-2: Effects of Different Future H-Values ....... .................. 14 1-3: Effects of Taxes on Production Factors (undefined property rights) .......................................... 14 1-4: Effects of Taxes on Final Products (undefined property rights) .... 15 lB-1: Base Case Data for Costa Rica, 1986 ....... ................. 23 1B-2: Land Use Data .23 2-1: Production Function, Ghana, 1988-89 .41 2-2: Land Equation, Western Ghana, 1988-89 .42 2-3: Net Effect of Increasing Cultivated Land by 10 Percent .43 2-3A: Extent of Overcultivation and Income Losses .43 2-4: Basic Parameters and Elasticities Used in the Simulations .... .... 43 2-5: Effects of Various Policies .............. ................... 44 3-1: Estimates of Structural Parameters ......... ................. 65 3-2: Aggregate Simulation Results ........... ................... 65 3-3: Sectoral Simulation Results .............. .................. 66 3C- 1: Results for Varying Factor Substitution Elasticity .75 3C-2: Results for Varying Water Prices .75 4-1: Land Use Classification in the Philippines, 1988 ..... .......... 88 4-2: Philippine Forest Land Population, 1950-85 ...... ............. 89 4-3: Incidence of Poverty in the Philippines, 1971-88 ..... .......... 89 4-4: Nominal Protection Rates for Main Agricultural Commodities, 1960-86 Averages . ...................................... 89 4-5: Frequency Distribution of Migration Streams, Mjj s, 1990 Census ............................................ 90 4-6: Description of Migration Determinants and Hypothesized Link to Migration ............................................ 90 4-7: Multinomial Discrete Choice Estimates of Determinants of Upland Migration .................. ................... 93 4B- 1: Descriptive Statistics of Explanatory Variables in the Migration Model .............. ..................... 98 5-1: Energy Intensity of National Income for Selected Countries ...... 120 5-2: Energy Intensity Relative to OECD for Selected CEE Countries, 1988 ................ .................... 121 5-3: Poland: Composition of Estimated Particulate Emissions .... .... 121 5-4: Poland: Composition of Estimated SO2 Emissions ..... ........ 121 5-5: IBRD Simulations: Scenarios for Energy Demand, Energy Intensities, and Emissions ........ .................. 122 5-6: MIT Scenarios for Energy Demand, Energy Intensities, and Emissions ......................................... 123 vi Volume II 5-7: EAP Scenarios for Energy Demand, Energy Intensities, and Emissions ......................................... 124 5-8: Poland: Increases in Nominal Emissions Fees ...... ........... 124 5-9: Poland: Increases in Real Emissions Fees-Actual and IBRD Simulations ........................................... 125 5-10: Poland: Costs of Emission Control Measures for Large Stationary and Transport Sources ...... ............ 125 6-1: Specific Generation Costs ........... ..................... 149 6-2: CEB Estimates of the LRMC .......... .................... 149 6-3: The Impact of Pricing Policy .......... .................... 150 6-4: Summary of Options Examined ......... ................... 151 7-1: Budget Costs of Animal Feed and Fertilizer Subsidies, 1975-91 .. 181 7-2: Geographic Distribution of Grazing Lands, 1986 ..... ......... 181 7-3: Changes in the Composition of Meat Consumption ..... ....... 182 8-1: Adjusted Contribution of Wildlife to the Economy, 1990/91 ..... 207 8-2: Comparative Analysis of Wildlife and Subsistence Agriculture: Total Revenue Estimates Using Nyaminyami District, Zimbabwe . 208 Figures 1-1: Main Linkages of the CGE Model ........................... 15 1-2: Resource Flows and Data Adjustments ....... ................ 16 1-3: Effects of Interest Rates When Property Rights are Well Defined .. 16 4-1: Expansion of Agriculture in Lowlands and Forestlands, 1960-87 . . 94 5-1: Poland: Energy Sector Balance ............... ............. 126 5-2: Emissions Under Base Case and CAC Scenarios (IBRD simulations) ................................... 127 5-3: Emissions Under Base Case and CAC Scenarios (IBRD simulations) ................................... 127 6-1: The Model ............... .................... 152 6-2: Expected Trends for CO2 Emissions in Sri Lanka ...... ........ 152 6-3: T&D Loss Predictions ................................... 153 6-4: The Real Price of Electricity ............................... 153 6-5: Balance Sheet Ratios ................................... 154 6-6: System Cost versus GHG Emissions .......... .............. 154 6-7: The Optimum Portfolio for GHG Reduction ....... ........... 155 6-8: The Cost Curve for GHG Emission Reductions ...... ......... 155 6B-1: Definition of Dominance and the Tradeoff Curve .160 6B-2: Local Health Impacts .160 7-1: Tunisia's Bioclimatic Zones ....... ............ 183 Contents vii 7-2: Rainfall and Herd Size, 1961-91. .an Herd.Size,.19 6 1-9 1.184 7-3: Distribution of Sheep by Region .184 7-4: Distribution of Goats by Region .185 7-5: Coverage of Feed Requirements by Zone Variation with Rainfall, 1984-86 .185 7-6: Beef and Mutton Prices .186 8-1: Comparison of Projected Household Income. .Incom 211 FOREWORD Recent years have seen a wide range of economywide policy reform prograns which have been undertaken in developing countries to address macroeconomic problems, such as those affecting international trade, government budgets, private investment, wages, and income distribution. Reform programs also address broad sectoral issues, such as those relating to agricultural productivity, industrial protection, and energy use. The economywide mechanisms for attaining these goals include altering the rates of exchange or interest, reducing government budgets, promoting market liberalization, fostering international openness, enhancing the role of the private sector, and strengthening government and market institutions-often coupled with pricing and other reforms in key sectors such as industry, agriculture, and energy. Although these policies are typically not directed explicitly toward influencing the quality of the natural environment, they may, nonetheless have major impacts upon it, either positive or negative. This book is the second of two volumes that address these economic policy and environment interactions. The case studies in this volume (and the synthesis of key principles in Volume 1) show that there are significant payoffs in attempting to understand such impacts better and to act upon them. Positive impacts of economywide reforms on the environment can be used to build constituencies for reform. Potential negative impacts need to be analyzed, monitored, and mitigated. The importance of addressing economywide reforms and environmental management in an integrated manner has received growing recognition. As noted in the Bank's Annual Report on the environment, most country assistance strategies (CAS) now address the environmental challenges facing each country (World Bank, 1995, Mainstreaming the Environment). The CAS is the principal statement of the World Bank's overall development strategy in a given country. In some country assistance strategies, the links between environmental problems and their underlying causes are discussed (e.g., natural resource or energy price distortions). In others, the focus is on government efforts to address environment issues at the country level. Recent examples include the CAS for Chile, China, Hungary, Latvia, Niger, Senegal, Swaziland, and Thailand. For exanple, in Senegal, although the main focus is on stimulating economic growth, the environment is listed as one of five problem areas that need specific attention. Environmental considerations are also featured in CAS discussions of sectoral policies and lending programs. These include water resource management (e.g., Brazil, India, FYR Macedonia, and Peru), urban sanitation (most countries), energy efficiency (China, and various countries in Eastern and Central Europe), industrial pollution control (Brazil, China, Ghana, and Thailand), mining (Estonia, Ghana, Peru, and Poland, among others), and agriculture (especially countries in Africa and Latin America) (World Bank, 1995, Mainstreaming the Environment). In some instances, the direction of environmental impacts stemming from economywide policy reform is fairly straightforward. The extent of the impact, however, invariably requires empirical analysis. In more complex cases, even the direction of the impact is ambiguous. Although much literature of a theoretical nature on this topic is emerging, there has still recently been little empirical work addressing the links between economywide policies and viii ix Foreword the environment. In view of their location-specific nature, involving a range of economic, physical, institutional, and cultural variables, there is a clear need for more case study material to enhance our understanding of these relationships. The case studies presented in this volume, which have been carried out within the World Bank over the last several years, are a contribution to this objective. They have already provided the basis for a comprehensive paper on economywide policies and the environment that was approved by the Bank's Board of Executive Directors, and presented at a special session of the 50th Anniversary Annual Meetings of the World Bank and International Monetary Fund in Madrid, October 1994 (World Bank, 1994, Economywide Policies and the Environment: Emerging Lessons from Experience). The case studies reflect a wide range of country situations and environmental problems. Pollution issues are addressed with reference to air quality and energy use in Poland and Sri Lanka, while a variety of natural resource- related issues are covered in the other studies: deforestation and land degradation in Costa Rica, migration and deforestation .n the Philippines, degradation of agricultural lands due to overgrazing in Tunisia, fertility losses due to extension of cultivated areas in Ghana, water resource depletion in Morocco, and wildlife management in Zimbabwe. The case studies also utilize a variety of analytical methods to illustrate the different approaches available for identifying the environmental implications of economywide reforms. These methods range from those tracing the links between economic incentives and resource use through direct observation, to others relying on more complex economic modeling of policies and their environmental effects. In all the studies, however, the analytical approach uniformly requires identifying key environmental concerns and relating these to the agenda of priority sectoral and macroeconomic reforms under consideration. The analysis underscores the formidable difficulties of developing a general methodology to trace all possible environmental impacts of a package of adjustment reforms. At the same time, it offers evidence that careful case-specific empirical work may help identify better ways to deal with potentially serious impacts of specific economywide policies on high-priority environmental problems. We obviously have a long way to go before environmental impact assessment of economic policies can be used systematically in public decisionmaking; nevertheless, this work helps to reinforce the case that imperfect knowledge should not be an excuse for inaction. In many important cases, the potential environmental consequences of economic policies may be discernible, and the direction in which compensatory or defensive measures should go, is quite apparent. Andrew Steer Vinod Thomas Director Director Environment Department Economic Development Institute The World Bank Washington, D.C. ix ACKNOWLEDGMENTS We acknowledge the comments and suggestions that were received during the course of many discussions, workshops, and reviews, from colleagues within and outside the World Bank. These include Richard Ackermann, Dennis Anderson, Maria Concepcion J. Cruz, Luis Constantino, Partha Dasgupta, Mohamed El-Ashry, Ved Gandhi, Christopher Gibbs, Gunnar Eskeland, Stein Hansen, T. J. Ho, William Hyde, Emmanuel Jimenez, Ronald MacMorran, Karl- Goran Maler, Paul Martin, David Pearce, Jorn Rattsoe, Robert Repetto, Iona Sebastian, Robert Schneider, Anand Seth, Hu Tao, Adriaan Ten-Kate, Laura Tuck, Hirofumi Uzawa, and David Wheeler. We have attempted to address all their comments and are solely responsible for any remaining inadequacies. We are especially grateful to our case study authors: Gregory Amacher, Robin Bates, Jan Bojo, Robert Cunliffe, Boguslaw Fiodor, Herminia Francisco, Ian Goldin, Shreekant Gupta, Ramon L_opez, Paul Martin, Peter Meier, Stephen D. Mink, Kay Muir-Leresche, Zeinab Partow, Annika Persson, David Roland- Host, and Tilak Siyambalapitiya. Publication of this book was sponsored by the Environment Department and the Economic Development Institute of the World Bank. We thank Mohamed T. El-Ashry, Andrew Steer, Vinod Thomas, and Hatsuya Azumi for their support. This book has also benefitted greatly from comments received when a condensed version was presented to the Board of Executive Directors of the World Bank, as well as at a special Environment Seminar convened at the Fiftieth Anniversary Annual Meetings of the World Bank and the International Monetary Fund in Madrid. The complex research program behind this book was made manageable by the excellent Economywide Policies and Environment project team. We especially thank Adelaida Schwab, Noreen Beg, Shreekant Gupta, Annika Persson, and Nalin Kishor. Karen Danczyk assisted with the editorial work; Luz Rivera and Daisy Martinez provided secretarial support. Jay Dougherty and Rebecca Kary of Alpha-Omega Services, Inc., provided editorial and layout services. The project was supported in part by generous contributions from the governments of Norway and Sweden. Wilfrido Cruz, Mohan Munasinghe, and Jeremy J. Warford x CONTRIBUTORS AND EDITORS Gregory Amacher is Associate Professor, Forestry Department, Virginia Polytechnic Institute. Robin Bates is Principal Economist, Industry and Energy Department, World Bank, Washington, D.C. Jan Bojo is Senior Environmental Economist in the Environmentally Sustain- able Development Division, Technical Department, Africa Region of the World Bank. Wilfrido Cruz is Environmental Economist at the Environment and Natural Resources Division of the Economic Development Institute of the World Bank. He was formerly Economist at the World Resources Institute, a leading environ- mental policy research organization. Robert Cunliffe is an Ecologist at the World Wildlife Fund, Zimbabwe. Boguslaw Fiodor is Professor, Instytut Ekonomil, Adademia Ekonomiczna, Wroclaw, Poland. Herminia Francisco is Associate Professor, Economics Department, College of Economics and Management, University of the Philippines at Los Banios. Ian Goldin is Chief Executive, Development BanK of Southern Africa, South Africa. Shreekant Gupta is Chief of the Environmental Economics Unit, National Institute of Public Finance and Policy, New Delhi, India. Ram6n L6pez is Professor, Department of Agricultural and Resource Econom- ics, University of Maryland, College Park, Maryland. Peter Meier is Chief Economist of IDEA, Inc., Washington, D.C. Stephen D. Mink is Senior Economist, Middle East and North Africa Region, World Bank, Washington, D.C. xi xii Contributors Kay Muir-Leresche is Senior Lecturer at the University of Zimbabwe, Depart- ment of Agricultural Economics and Extension, Harare, Zimbabwe. Mohan Munasinghe is Distinguished Visiting Professor of Environmental Management, University of Colombo, Sri Lanka. He is on leave from the World Bank, where he was Chief of the Pollution and Environmental Economics Division, Environment Department. Zeinab Partow is a Researcher, Banco de la Republica, Bogota, Colombia. Annika Persson is a Consultant in the World Bank Policy Research Depart- ment, Washington, D.C. David Roland-Host is Associate Professor of Economics, Mills College, Oakland, California. Tilak Siyambalapitiya is head of the Generation Planning Branch, Ceylon Electricity Board, Sri Lanka. Jeremy J. Warford is Visiting Professor of Environmental Economics, Centre for Social and Economic Research on the Global Environment, University College, London. He was formerly Senior Adviser in the Environment Depart- ment and Economic Development Institute, the World Bank. 1 Natural Resource Management and Economywide Policies in Costa Rica: A Computable General Equilibrium (CGE) Modeling Approach Annika Persson and Mohan Munasinghe CURRENTLY, THE ECONOMIC ANALYSIS of environmental issues relies mainly on project level studies, using cost-benefit analyses and environmental assessments. However, economywide policies (both macroeconomic and sectoral) frequently have much more powerful environmental effects than mere project level investments. Some progress has been made in identifying the environmental consequences of sectoral policies involving, for example, energy, water or agricultural pricing. Nevertheless, the impacts of broad macroeconomic reforms (such as exchange rate devaluation, trade liberalization, privatization, and other fiscal and monetary stabilization policies) on natural resource and pollution management are far more difficult to trace. This hampers efforts to design better sustainable development strategies that meet economic, social, and environmental criteria in a more balanced way (Munasinghe 1993). This chapter, with minor format and presentation changes, was initially published in the World Bank Economic Review, Vol. 9, No. 2 (pp. 259-85). With permission, it is presented here with the other case studies from the Economywide Policy and Environment Project. I 2 VOLUME II In the case of the World Bank, for example, well as nonmarket methods (such as the general lack of knowledge about links command-and-control techniques). between economic policies and the environ- The ideal approach is a "general equilib- ment has delayed attempts to gradually expand rium" analysis that traces both the economic the application of environmental analysis to and environmental effects of economywide cover economywide or policy-based lend- policy reforms. When such comprehensive ing-the second largest use of Bank resources methods are not possible in developing coun- (about $5.8 billion annually, or 27 percent of tries where data and skills are scarcer, "par- total lending in 1993). It has also hampered tial" approaches that help to identify the most efforts to develop more effective national important impacts of economywide policies, environmental action plans (NEAPs) (these are frequently used. Because the full conse- are prepared by borrowing countries, with quences of a policy are not traced, both quan- Bank assistance, to help determine priority titative and qualitative results of the partial activities to address national environmental equilibrium model may be wrong. For exam- issues). ple, taxes that are not "lump sum" may carry A recent study argues that many instances over from the sector for which they were of environmental damage are due to market intended into other sectors of the economy and failures and policy distortions, exacerbated by affect consumption and production decisions unemployment, landlessness, and poverty there as well. In this context, the main purpose (Munasinghe et al. 1993). Therefore, broad of this paper is to investigate the effects of policy reforms which usually promote effi- economywide policies in Costa Rica on forest ciency or reduce poverty, also should be areas and the environment (Persson 1994). We generally beneficial for the environment. also seek to determine whether new measures However, some of these reforms may have involving the allocation of property rights to negative environmental effects, depending on these forests will yield different results when pre-existing (and often localized) analyzed using a general equilibrium model constraints-such as inadequately defined than by applying a more conventional partial property or resource rights. equilibrium approach. It is important for decisionmakers to be In the remainder of this section, the main able to trace the complicated paths by which issues, analytical approach and results, are macro-level policy changes ultimately affect summarized. The next section describes previ- incentives for efficient resource use at the ous work on Costa Rica and the environmen- micro level of firms or households. The objec- tal priorities. The applicability of CGE models tive is not necessarily to modify the original to such problems is discussed in the section, broader policies (which have conventional The modeling approach. Further details of the economic or poverty related goals), but rather model and data used here are presented in the to design more specific or localized comple- next two sections. Finally, the last two sec- mentary measures that remove economic tions summarize the chief results and conclu- distortions or constraints (Munasinghe and sions of the study. Annex I-A provides more Cruz 1994). These additional measures would information on the model. help mitigate negative effects or enhance Deforestation and soil erosion are major positive impacts of the original policies on the environmental problems in Costa Rica. Some environment. Such complementary actions data on forest clearing over time are shown in would include both market-based approaches Table 1-1. To evaluate how sectoral and (like Pigouvian taxes on environmental exter- economywide policies can help control defor- nalities or allocation of limited pollution estation, the CGE model used here highlights rights coupled with marketable permits), as the economic activities and factors that specif- 1. Natural Resource Management and Economywide Policies in Costa Rica 3 ically affect deforestation in Costa Rica. The significantly higher than the value of the logs model goes beyond standard approaches in and the cleared land. two important respects. First, it can simulate In the model, tax policies may generate the effect of introducing property rights on unexpected side effects, and substitution forest resources, thus encouraging sustainable effects between inputs in the producing sec- management of forests by private individuals tors may be important. Therefore, when possi- who value future returns to forestry. Second, ble impacts of macroeconomic policies are it includes markets for logs and cleared land: investigated the general equilibrium approach loggers deforest to sell timber to the forest generates results that are different from those industry and for exports, and squatters clear derived from a partial equilibrium analysis. land for agricultural production and for sale to the agriculture sector as the latter expands and Status of Forests in Costa Rica requires more land. The model retains features that are fairly Deforestation in Costa Rica is proceeding standard in most CGE models. The tradable at a rapid pace, and there is a growing concern sectors-forestry, agriculture, and indus- about this both inside the country and from try-are price takers in the world market, environmental organizations in the rest of the while infrastructure and services produce world. Ministerio de Recursos Naturales, nontraded output. To focus on the natural Energia y Minas (1990) mentions the follow- resource sectors, the domestically mobile ing economic and ecological benefits that factors include, aside from capital and (skilled Costa Rica may lose if deforestation contin- and unskilled) labor, cleared land and logs. ues: access to construction materials and other The supplies of both labor and capital are wood products, unchecked species of plants exogenous. The demand for these factors arise and animals that have possible and future uses from the producing sectors (agriculture, indus- for consumption and industrial production, try, etc.) and from the deforestation activity of recreation and ecotourism, control of erosion loggers and squatters. The supply of cleared and sedimentation, and education and research land is initially based on Costa Rica's total possibilities. The greenhouse effect and con- land area that has been deforested. However, cerns about the rich biological diversity in additional cleared land is made available from Costa Rica may be important to other coun- increased deforestation. This rate of land tries and environmental organizations. clearing depends on the definition of property Deforestation and erosion are the main rights as well as on taxes (or subsidies) that environmental problems in the country affect the forest and agricultural sectors. In (Blomstrom and Lundahl 1989; Foy and Daly addition, the expansion of squatting activities 1989). Originally, most of Costa Rica was augments the cleared land factor. Agricultural forested, but in 1977 only 31 percent (16,000 production provides the demand for cleared square kilometers) remained covered with land. forests. Blomstrom and Lundahl (1989) esti- Poorly defined property rights in Costa mate that in 1983 14 percent of the area was Rican forests play an important role in defor- still covered with forests. Sol6rzano et al. estation. The model indicates that correction (1991) give the more conservative estimate of this market failure would reduce deforesta- that about 40 percent of the land is still cov- tion. If property rights are well defined and the ered with forests. This difference is probably interest rate is exogenous, the value that due to differences in what types of forests loggers assign to preserving the forests is were investigated. The lower estimates proba- crucial. In order to stop deforestation, the bly concern only primary forests, whereas the benefits from preserving the forests must be higher estimates include secondary forests and intervened forests (Sader and Joyce 1988). 4 VOLUME II Most of the deforestation has occurred since and only a few species are commercially 1950. If deforestation continues at the current utilized. About 54 percent of the logs are rate, the commercial forests of Costa Rica will processed, and of these about half finally be exhausted within the next five years. The reach the market (Ministerio de Recursos life zones with the highest rates of deforesta- Naturales, Energia y Minas 1990). The main tion are the tropical wet forests; these are also part of the logs used in the timber industry is the life zones in which biodiversity levels are bought from sources other than the industry highest (Sol6rzano et al. 1991). itself. Carriere (1991) describes the process of 2. Banana firms and other companies are deforestation as taking place in several stages. expanding their plantations rapidly. First, a logging company involved in high- The main products cultivated in Costa Rica grading clears a vehicle tract to extract lum- are rice, coffee, fruits, sugarcane, beans, ber. Thereafter, the road is improved by the maize, and sorghum (Hugo et al. 1983). Lutz government due to pressures from lobbying and Daly (1990) state that erosion is visible in groups, and this in turn enables local peasant some areas, but that farmers "do not produce families to clear and use the remaining forest in obviously unsuitable ways to destroy the for subsistence agriculture until the decreasing environment. For example, living fences are yields force them to sell or abandon the land, widely used, which reduce erosion, and pro- depending on whether it is titled or not. How- tective forest cover is left intact next to creeks, ever, the land is still suitable for pasture and is on contours or steep slopes, etc." therefore assembled by urban-based real estate The Costa Rican tax structure for income companies and sold to cattle ranchers. After a and property taxes is regressive. Sales taxes few years, the land is almost completely and other indirect taxes constituted 70 percent degraded and unsuitable for any kind of eco- of the total tax revenues in 1970, and there are nomic use. This view is shared by Keogh indications that this figure may still be high. (1984). Although property taxes are low (in some The Costa Rican government is taking steps cases about 1 percent of the actual market to preserve the forests. More than 13,000 value), property and income tax evasion is a square kilometers has been designated as problem that costs the country approximately national parks, although in the past deforesta- 100 billion colones a year. A remedy may be tion had been encouraged to diversify the to raise the price of land by increasing land country's production away from coffee and taxes, increase tax collection rates, and prose- banana crops (Biesanz et al. 1987). cute tax evaders more effectively. The following four groups as responsible 3. Cattle ranchers have expanded their for deforestation in Costa Rica (Lutz and Daly activities rapidly at the expense of for- 1990): ested areas in recent decades. However, 1. The timber industry may be responsible this type of land conversion may be for deforesting as much as 20,000 hect- limited now since most of the land that ares annually. Logging requires a special can be sustainably used for pasture has permit from the government, but about already been cleared. half the trees are cut illegally. In the 1950s and 1960s there was a large Domestically cut logs are processed locally increase in investment in cattle, encouraged by and are used typically in construction. Exports foreign aid and investment, as well as govern- of wood and wood products are small, and ment aid in the forms of credit and provision imports are negligible. The current import of infrastructure. This increase of cattle ranch- tariff on logs is 5 percent (Lutz and Daly ing caused deforestation to increase rapidly. 1990). Efficiency in the forestry sector is low, The pasture trend boomed in the 1970s, but 1. Natural Resource Management and Economywide Policies in Costa Rica 5 since then profits have decreased. More than ness to pay for the preservation of the Costa 70 percent of the farmland is in pasture, while Rican forests will not be included in the pri- only 2.5 percent is in coffee, and 1.1 percent is vate cost. That the difference in cost functions in bananas (Biesanz et al. 1987). is a major cause behind deforestation in the 4. Squatting is taking place on both pri- developing world is shown in Chichilnisky vately owned and government land. (1993), where a north-south trade model is Some of the squatters produce agricul- developed, and the difference in trade patterns tural outputs, but others sell the cleared between north and south is explained by a land to cattle ranchers or other land difference in property rights. Thus, deforesta- owners. Buyers who buy "in good faith" tion would be driven by the difference in from squatters are not prosecuted. About private and social objectives. For example, the twice as much is paid for cleared land as loggers' main interest may be the profitability is paid for forests. of the logging operation itself without much Squatting is an important cause of defores- consideration about future, alternative uses of tation in Costa Rica. By clearing the land, it is the land. possible to get formal ownership to the land Another critical economic factor may be the (Blomstrom and Lundahl 1989) or in some existence of higher private discount rates, i.e., cases at least to the "land improvements." deforestation may be caused by discounting Squatting by smallholders nowadays appears the future value of the forests. A high discount to constitute a less significant part of the factor implies that future gains from the for- deforestation in an overall context, although it ests is of much less value than the gains from may be locally important (Lutz and Daly deforestation today. The impacts of tropical 1990). forests are often more significant in the long If ownership may be obtained with no costs term than in the short term. However, the other than the clearing of the land, the forests regenerative capacity of tropical forests is low, can be seen as a type of common property and the discounting of future environmental while the cleared land is perceived as tradi- benefits may often make it more profitable to tional private property. However, in the case harvest forest resources as quickly as possible. of Costa Rica, it is not the traditional case of Forest investments, like replanting, take a long undefined property rights where there is an time to yield returns, and individuals therefore open access resource (cf., for example, find little attraction in conservation and refor- Dasgupta 1982). Here we are looking at inse- estation activities. In many developing coun- cure land tenure. This implies that there is no tries, private market rates are very high and crowding effect on the stock of the resource, often exceed the rate that would be socially which is what occurs when each agent maxi- justifiable (Barbier, Burgess, and Markandya mizes his own profit without taking the effect 1991). Poor people often face even higher on the stock of the resource into account. discount rates because of credit constraints. Instead of the traditional open access problem, we have a form of short-term property rights The Modeling Approach when deforestation occurs, but the property rights to the standing forests are not protected. As may be concluded from the above, the The logger or squatter will continue deforesta- main reasons for deforestation and thereby tion only until the marginal cost of deforesta- erosion are the following: tion equals the marginal revenue, because of this structure of property rights. The social * The price of land is too low, since the cost of deforestation will then be higher than total social opportunity value of the rain the private cost since "the world's" willing- forests is not included. 6 VOLUME II * Undefined property rights make the and taxation, and the land tenure system) private cost of deforestation lower than promoting deforestation. The approach im- the social cost of deforestation. plies that every unit of production in each producing sector produces, for example, a * Discount rates may be too high. This fixed amount of air pollution or deforestation. implies that the value of future gains The environmental impacts are not part of the from the forests are lower than the gains model per se-the environmental degradation from deforestation today. or improvement is not fed back into the model so as to affect future production and consump- * In addition, economywide policies such tion decisions. The results include the findings as the tax system may cause deforesta- that export taxes on rice and rubber increase tion. investment in soil conservation, increase the use of agrochemicals, and shift land from Computable general equilibrium (CGE) rubber to rice. models have been applied before to environ- Not much work has previously been done mental problems-mainly issues involving air on the modeling of undefined property rights pollution and pollution taxes. A short discus- in a general equilibrium context, where the sion of some CGE models relevant to the results may differ from those of a partial approach developed here is provided below. equilibrium model. Devarajan (1990) suggests Bergman's (1990a) model is designed to that a fruitful approach may be to incorporate simulate the effects of environmental regula- a partial equilibrium model in the general tion and energy policy on the Swedish econ- equilibrium framework by removing the first omy. The environmental market failure is in order condition that labor must be paid the this case corrected by the creation of a market value of its marginal product in some sectors for emission permits. The cost of emission and replace it with a condition that reflects the permits for carbon dioxide, sulfur, and nitro- suboptimal behavior of the sector. This will gen is incorporated in the cost functions. enable an analysis of the effects on deforesta- Jorgenson and Wilcoxen (1990) analyze the tion of policy interventions in the system. It is economic impact of environmental regulations emphasized that the model has to be dynamic on the U.S. economy. This is done by simulat- in order to take account of both the stock and ing long-term growth with and without envi- the flow effects of deforestation. ronmental regulations. The share of abatement Unemo (1993) models the suboptimal use costs in total costs is estimated for each indus- of land in Botswana, caused by overgrazing of try, as well as the share of investment in cattle due to undefined property rights of the pollution control equipment and the cost of land. Land is seen as an open access resource, pollution control devices in motor vehicles. and the effects of overgrazing are incorporated The model is run with and without these costs, in the cattle owner's production function in to estimate the economic impacts. the form of crowding effects. The quantity of There are not many examples of CGE output is determined not only by the number models dealing with the impact on the econ- of cattle the individual owns, but by the whole omy of overexploitation of natural resources. population of cattle grazing on the land. The Panayotou and Sussangkarn (1992) construct results include the finding that a fall in the a model against the background of environ- price of diamonds considerably increases mental problems in Thailand. The sources of pressure on land, as mining becomes less the environmental issues are economic profitable relative to cattle ranching. growth, exchange rate problems, and govern- In order to model property rights related ment policies (such as agricultural policies behavior in Costa Rican forests, it is assumed 1. Natural Resource Management and Economywide Policies in Costa Rica 7 that the private cost of deforestation is lower LOGGERS clear land for the purpose of than the social opportunity value of the forests obtaining LOGS for the FOREST industry and when property rights are undefined. When for exports, and SQUATTERS clear land and property rights are defined, the social value of sell it to the AGRICULTURE sector. the rain forests will be incorporated in the The domestic intersectorally mobile pro- utility functions of the squatters and therefore duction factors are unskilled labor in the private cost of deforestation. This ap- (ULABOR), skilled labor (SLABOR) and proach facilitates the airalysis of the role of capital (CAPITAL). Logs (LOGS) and cleared undefined property rights, and follows the land (DLAND) are specific to the sectors approach used by Chichilnisky (1993). FOREST and AGRICULTURE, respectively, although logs can be traded on the world General Features of the Model market. No reafforestation is possible in the model. This model is a static CGE model of an The key elements of the CGE model are open economy, although it has certain implicit introduced below. A more detailed mathemati- dynamic features, since the discount rate is cal description is given in Annex 1-B. included in the future valuation of forested land. It differs from the standard approach of Factor market equilibrium and the stock CGE modeling by the inclusion of undefined of forested land property rights and by the modification of the The supplies of both labor and capital are functioning of the markets for logs and cleared assumed to be exogenously given and, for land. This is discussed in detail below. Land factor markets to clear, these supplies must cleared by squatters is assumed to be sold to equal the demands for labor and capital, re- the agricultural sector. eultedmnsfrlbradcptl e The model has two types of sectors. The spectively. Demands arise from the producing tradable-producing sectors (T-sectors) FOR- sectors plus the amounts used for deforesta- EST, AGRICULTURE, and ITNDUSTRY are tion by squatters and loggers. The demand for EST,~~ ~ AGIUTR,ad' DSR r each production factor (like capital or labor) assumed to be price takers on the world mar- eachp n factor (ectal or lbr ket in the standard Heckscher-Ohlin fashion both the T and N sectors, as well as the The nontradable-producing sectors (N-sectors) deresation secos, isngiven by the partial are INFRASTRUCTURE and SERVICE.' In derivatve of the cost function for the relevant addition, there are two sectors that clear land. sector with respect to the price of the same addin tproduction factor. There is demand for un- skilled labor for deforestation by both loggers 'The subsectors of the input-output table are and squatters, but only loggers generate de- aggregated into the five production sectors as follows: mand for capital. FOREST (forestry and fishing)- AGRICULTURE Costa Rica's total area has been divided (bananas, unprocessed coffee, sugarcane, cacao, basic grains, cotton, tobacco, livestock, other agricultural, into two types of land: cleared land and for- coffee processing, grains milling, sugar refining); ested land. Cleared land is produced through INDUSTRIES (meat and milk, fish tinning, edible oils, deforestation. The production of cleared land bakeries, other manufactured goods, drink, tobacco will depend on the definition of property products, textiles and clothing, leather and shoes, rights as well as taxes and subsidies on the timber and furniture, paper and printing, chemical products, oil refining, tire products, plastic and rubber, factors of production and the profits in the glass and ceramic, construction materials, metal FOREST and AGRICULTURE sectors. products, electric products, transport equipment, other Logs are assumed to be tradable. There- manufacturing); INFRASTRUCTURE (construction, fore, the demand for forest land by the logging transport, electricity, gas and water); SERVICES sector and the world market price determines (banking and finance, commerce, ownership of the rate of deforestation. This demand is equal dwellings, general government). 8 VOLUME II to the partial derivative of the logging cost In the nontradables-producing sectors, the function with respect to the user cost of logs sector-specific capital is endogenously ad- plus the net export of logs. justed so that price equals marginal cost. The supply of cleared land is composed of the stock of cleared land, plus deforestation by Prices, domestic demand, foreign trade squatters (the production of cleared land by and market clearing squatters is further discussed in the next sub- For a good produced in the tradables-pro- section, Technology, costs, and producer ducing sectors, the domestic producer price is behavior). The demand for cleared land is the equal to the world market price of the identi- demand from the AGRICULTURE sector, cal to and marketrlesoducing which is set equal to the partial derivative of cal good, and in the nontradables-producing the agriculture sector cost function with re- ser the timesthe tax rate. spect to the user cost of cleared land. producer price times the tax rate. specTtothe userbinatio of pereducto lacs cThe intermediate demand of a good is given The combination of production factors can by the technology assumptions. Domestic final be influenced by taxes and subsidies. Thus, a demand is given by a linear expenditure sys- given user price will be greater (smaller) than tem, derived from the consumers' utility the corresponding supply price by a percent- maximization. age tax (subsidy). To equilibrate the market for a good, the Technology, costs, and producer behavior net export for that good is defined as the difference between domestic supply and The production factors have been aggre- demand. gated into a composite input, Y. DLAND is combined with CAPITAL to yield an aggre- Deforestation sectors gate R, which in turn is combined with LOGS to generate M. The latter is combined with In this model, there are two sectors that are SLABOR to produce V, which is combined responsible for deforestation, as discussed withSULABOR to yioduceld thecs compsited fbelow. They interact with the rest of the econ- with ULABOR to yield the composite factorI omy through their demands for capital and input, Y. This aggregationI IS accomplished lao,bsupynfretrdcsadcer through the use of constant elasticity of substi- labor, by supplytg forest products and clear tution (CES) production functions. The tech- land to the rest of the economy, and through nology is specified to exhibit constant returns the changes in relative prices of factor inputs to scale. The relation between inputs and and the sectoral outputs. output is given by typical Leontief production The logging sector functions for each sector. Since the technology exhibits constant The logging sector is assumed to have a returns to scale, the marginal cost and the capital intensive technology (Repetto 1988). average cost of production in a given sector Further, the technology is assumed to exhibit can be expressed as a linear function of prices, decreasing returns to scale in order to reflect relevant input/output coefficients, and indirect the diminishing amount of available forests, as tax rates. well as the fact that much of the logging is Producers are assumed to maximize profits. done illegally. The production of LOGS is The producer output prices, Pi, in the assumed to depend only on two factors of tradables-producing sectors are given by the production: LABOR and CAPITAL. A world market prices. Assuming perfect com- loglinear production function is used. Since petition, this implies that pure profits are the technology used to model the diminishing nonpositive, output is non-negative and posi- yields in deforestation exhibits decreasing tive only if pure profits are equal to zero. returns to scale, this implies that the returns to 1. Natural Resource Management and Economywide Policies in Costa Rica 9 the production factors fall with increased environmental damage. Therefore, the total deforestation. Deforestation for land and social cost of deforestation is this private cost deforestation for logs are assumed to be inde- plus the future benefits from cleared forests pendent of each other, and therefore the in- that are foregone by clearing the land today. creased deforestation for logs does not affect The future value of the forests is assumed to the returns to deforestation for land, and vice be greater than the value of the forests today. versa. However, increased deforestation for The analysis of the definition of property logs implies decreasing yields in the logging rights can then be accomplished through the sector, and increased deforestation for land simulation of two regimes. In the case of implies diminishing returns in the squatting undefined property rights, the present-day sector. squatters do not take the future value of the In the case of undefined property rights, forests into account. When property rights are loggers take only the private cost of deforesta- well defined, the squatters own their land and tion into account. When property rights are do take the future value of the forests into well defined, the opportunity value of saving account. The owners of forested land (i.e., the the forests is included in the loggers' cost "squatters aware of the future"), decide function. whether to preserve the forests or clear land. When property rights are undefined, there Squatters is no market for the forests available to squat- The forested land cleared by squatters is ters. A simple partial model of land clearing seen as a common property, although there is by squatters (in which each squatter receives n crowding.effect since thestockoffoan equal share of the private profits) is used to no crowding effect slnce the stocktofsfrestd show that land will be cleared until marginal land is not included In the squatters' produc- coteusmaialrvn.Thseut tion function. The base case assumes unde- cost equals margmal revenue. This result fined property rights. This section is inspired corresponds to maximization of private profit, mainly by Johansson and Ldfgren (1985). given the insecure land tenure. The squatters have a production function When property rights are well defined, for cleared land that increases monotonically there is a market for the forests. The squatters with labor inputs. Their total revenue from take the future value of the forests into ac- clearing the land is the price paid for the count, and they can choose to clear forested r.learmg ~pic land orn to preerv the forsts This ise coss cleared land. Part of the land cleared by squat- land or to preserve the forests. This IS consis- ters is sold to the AGRICULTURE sector; the tent with the condition for socially optimal rest is used for subsistence agriculture by the forestry-that a tree should be harvested when squatters themselves. However, since both the market value is equal to the shadow value activities occur, the returns at the margin must (Hellsten 1988). This result corresponds to the o timization of net social benefits. be the same in each case. The squatters are assumed not to sell the timber from their It can be deduced from the foregoing that deforestation, as discussed in the section, more land will be cleared when property rights Status of Forests in Costa Rica. Other uses of are undefined rather than when property rights the timber, such as for firewood, are assumed are well defined. This is because the squatters' thbe t gimbe . marginal cost of deforestation is lower when to be negligible. The squatters' total private cost for the property rights are undefined, rather than clearing of the land will depend only on the when property rights are well defined, and the amount of labor needed in order to clear the cost includes the future value of the forests. land when property rights are undefined. This MR is the marginal revenue curve. is the private cost, which does not include the A more detailed analysis of the supply future value of the forests and the cost of function indicates that when property rights 10 VOLUMEII are well defined, deforestation will be in- Base Case Data, Assumptions, and creased by (a) a change of technology towards Limitations of the Model more efficient use of labor in the production of cleared land, (b) an increase of the time The data used in this version of the model preference rate, and (c) an increase in the originate from Bricefio (1986),2 the University supply price of cleared land. Conversely, of Costa Rica, and the Costa Rica National deforestation is reduced by increases in: (a) Accounts (Central Bank 1990). However, the the future value of the forests; and (b) the sectors of production are not consistent be- price of labor. tween the two studies, and data were therefore When property rights are undefined, the adjusted in Ravent6s (1990). The input-output clearing of land is not affected by the future matrix in Annex 1-B, Table IB-I has been value of forests and the rate of time prefer- calculated from the disaggregated data used in ence. The effects of other variables are the Ravent6s (1990 and 1991). The remaining same as in the earlier case. differences have been added to the net export The profit maximization condition for the column. Land use data are shown in Annex 1- squatters in the general equilibrium model B, Table IB-2. The economic rent to timber includes a term reflecting the opportunity has been calculated from Sol6rzano et al. value of saving the forests for alternative uses (1991). Deforestation in 1986 was assumed to or deforestation in a later period. When prop- equal average deforestation between 1973 and erty rights are undefined, the weight given to 1989. The value in 1986 prices was calculated this term is zero, since future tenure of the using the increase in the consumer price index forest is uncertain. When property rights are between January 1985 and December 1986 well defined, this term is included in the profit (calculated from Sol6rzano et al. 1991). The maximization. rent to the production factor DLAND was subtracted from the rents to CAPITAL in the Macroeconomic closure and measures of agricultural sector, and the LABOR used for welfare land clearing by squatters was subtracted from * ~~~~~~the LABOR used in the same sector. The current account is assumed to be con- stant, and the current account surplus is de- The LABOR and CAPITAL used for log- fined as the sum of net exports. There are ging were subtracted from the payments to three welfare measures in the model: (a) the those factors in the forest sector. Logging is disposable income (which is implicitly deter- assumed to be responsible for half of total mined from the current account), (b) the green deforestation, while land clearing by squatters is assumed to be responsible for the other half. theosum natiofal faoctr i icms ps aetermind that No estimates of elasticities of substitution diminishes with incres dfrstation toat between production factors are available. It is reflect the negative welfare effects of defores- reasonable to assume that they are imperfect tation), and (c) utility (which is determined substitutes, and all substitution elasticities from the consumer's utility function). Utility were therefore assumed to be less than one. As maximization results in a linear expenditure a base case, the substitution elasticity between system for goods, based on a transformed LAND and CAPITAL in the sector AGRI- Cobb-Douglas utility function. CULTURE was set at 0.5. The substitution elasticity between the capital aggregate R and LOGS was assumed to be 0.8 in the FOREST sector, and the substitution elasticity between 2Calculated from Sol6rzano et al. (1991). 1. Natural Resource Management and Economywide Policies in Costa Rica 11 the aggregate M and LABOR was set at 0.8 in framework discussed earlier. This is due to all producing sectors. Compared to other substitution effects in the producing sectors. studies, such as Bergman (1990a and 1990b), The current situation with undefined prop- these values appear reasonable. The remaining erty rights was taken as a base case. As a first elasticities concern aggregates involving land step, property rights were defined, and the and logs in sectors which cannot use those opportunity value of the forests (the H-value) factors as inputs in production, and therefore was set 28 percent higher than the value de- the shares of those inputs in production always rived from deforestation (Sol6rzano et al. have to be zero. Those elasticities where set to 1991). The discount rate was set at 10 percent. zero, which is consistent with a fixed coeffi- The results are displayed in Table 1-2. A cient (Leontief) technology. comparison of the first and second columns The parameters in the production functions shows that the definition of property rights for squatters and loggers are judgment-based results in a dramatic decrease in deforestation, estimates, assuming a labor-intensive technol- and an increase in the net import of logs (not ogy for the squatters and a capital-intensive shown in the table). Activity in the forest technology for loggers. sector declines significantly, since this sector In concluding this section, we note several is relatively capital intensive and the price of limitations in the data and model formulation. capital is elevated. Deforestation by squatters First, because of the various data adjustments, ceases, and activity in the agricultural sector the results of the simulations are mainly indic- declines, but to a smaller extent than the forest ative and not necessarily precise quantitative sector, since the existing stock of land re- measures. mains. The welfare measures remain constant, Second, since the model developed in this since consumption of different goods is un- paper is essentially static, the results are com- changed. parative snapshots of different policy experi- Sensitivity analysis (remaining columns in ments. A more dynamic version of the model Table 1-2) shows that even a relatively small is being developed, which takes the stock and opportunity value of forests (e.g., H=0.0792) flow effects into account, and derives valid will decrease deforestation dramatically. results for a longer-term planning horizon. However, for deforestation to cease com- Third, the approach developed here does pletely, a high value (H=0.4792) is required. not include some other possible linkages with Both the opportunity value of forests and the deforestation. Migration and population interest rate are exogenous to the model. growth are two causal factors that may be Varying the interest rate while keeping oppor- important (Harrison 1991), but they are not tunity value fixed shows that high interest investigated. Furthermore, the model neither rates promote deforestation, and vice versa. allows for reafforestation, nor includes erosion The results of varying the interest rate may and other external effects of deforestation. The be deduced from Table 1-2, since a decrease economic valuation of such environmental in the interest rate is equivalent to an increase effects (to incorporate them in the conven- in the opportunity value, and vice versa. The tional economic analysis) would be a formida- equivalent of Table 1-2 in terms of interest ble task (see, for example, Munasinghe 1993). rates is shown in Figure 1-3. We can conclude that though deforestation is increasing with Results interest rate, the relationship is not linear. Next, the effects of taxes on logs, land, Simulations of different policy experiments unskilled labor, and capital were investigated, and the definition of property rights generated as summarized in Table 1-3. A 10 percent tax some results that are different from what increase on logs generated predictable results, could be expected from the partial equilibrium 12 VOLUMEII with no deforestation from loggers and no the forest sector itself suffers. The effects are production in the forest sector. Resources reversed for a doubling of the tax on forest were shifted to the agricultural sector, with an products. Deforestation remains largely unaf- increase in deforestation for land and an fected in both cases. increase in total deforestation. The increase in When the tax on agricultural products is total deforestation can be explained by low- reduced to half, the agricultural sector actually ered prices of capital and unskilled labor, decreases. The industrial sector benefits be- resulting from the discontinued production in cause of its extensive use of agricultural the forest sector. The tax increase actually products as intermediate inputs, and the forest results in a higher level of utility, as well as an sector is reduced. Deforestation for logs re- increase in "green" gross domestic product mains constant, while deforestation for land is (GDP). Smaller tax increases than 10 percent somewhat reduced. Utility, income and green generated the similar results, with the forest GNP measures are reduced. A double tax on sector declining and the agricultural sector agricultural products generate the opposite expanding as taxes on forest products were effects. A tax on products from the industrial increased. sector generate the same effects as a tax on Taxes and subsidies on land generate ex- agricultural products, although the magnitude pected results, with a corresponding change in of the changes is larger. deforestation by squatters and roughly con- stant deforestation by loggers. Both the tax Conclusions and the subsidy are distortionary, and reduce utility as well as income and GDP. The results of the CGE study support the A 10 percent tax increase on unskilled labor more conventional partial equilibrium ap- adversely affects the forest sector, but logging proach that establishing property rights tend to continues and logs are exported. The price of decrease deforestation. The reason is that such unskilled labor in the deforestation sectors are rights allow forest users to capture the future actually reduced, since those sectors are con- benefits of reduced logging damage today. sidered informal in the sense that their activi- Initially, this potentially avoidable loss is ties are to a large extent illegal, and the sectors presumed to be 28 percent of the value of the remains unaffected by government tax poli- residual stand, based on a recent environmen- cies. Resources are shifted to the agricultural tal accounting study (Sol6rzano et al. 1991). sector, with a large increase in land clearing Using an interest rate of 10 percent, the simu- by squatters as a result. The agricultural sector lation indicates that deforestation is dramati- gains an advantage relative to the industry cally reduced to 5 percent of the base level as sector, and industrial production is reduced. both the logging and squatters sectors internal- These results also hold for the experiments ize the losses associated with deforestation, with capital tax policies. and reduce corresponding activities. Signifi- Substitution effects prove to be important cant reductions in deforestation occur even for policy experiments involving tax changes when the estimate of logging damage is sub- on goods produced in tradables sectors, as stantially reduced. The CGE results concern- summarized in Table 1-4. The effects of tax ing the effects of discount rate changes paral- changes on goods from the forest sector gener- lel the predictions of partial equilibrium ate small economywide effects, since this models-higher interest rates promote defor- sector is small compared with the others. The estation, while lower interest rates contribute industrial sector (which uses forest products to conservation. relatively intensely as an intermediate input) Beyond confirming the direct results of gains from the tax reduction and grows, while partial equilibrium analyses, the CGE ap- proach also makes an important contribution 1. Natural Resource Management and Economywide Policies in Costa Rica 13 by clearly identifying the indirect effects rium results. If the wage of unskilled labor arising from intersectoral linkages. This im- were increased (e.g., due to minimum wage pact must be combined with the direct effects legislation), the model predicts that deforesta- attributable to policies that are specific to the tion could worsen instead of declining. Al- forest sector-to determine the total impact. though logging declines due to increased For example, partial equilibrium analysis direct costs of higher wages, this is more than predicts that stumpage price increases will act offset by the indirect effect of intersectoral directly to reduce logging. On the other hand, flows since the industrial sector (where mini- the model shows that while deforestation from mum wage legislation is more binding) is logging will indeed decline, total deforestation much more adversely affected by the higher nevertheless increases. This phenomenon labor costs. Labor and capital thus tend to arises from indirect linkages captured by the flow to agriculture, leading to the conversion general equilibrium analysis. The contraction of even more forest land for farming. of the logging and forest industry sectors Finally, both these last two examples un- causes a shift of resources toward agriculture derline the importance of pursuing sectoral and, as agriculture expands, deforestation reforms in the context of growth. Without increases. alternative employment opportunities, reduc- The importance of such indirect effects is ing logging activities will tend to direct labor also demonstrated in the analysis of and capital resources toward agriculture, economywide policy changes, such as an industry, and other sectors. Expansion of some increase in the wage rate. Because of of these sectors may lead to a second round of intersectoral resource flows, the general equi- effects on foresry, which could ultimately librium model captures effects of changes in result in more severe deforestation. wages that are different from partial equilib- 14 VOLUME!! Table 1-1: Percent of Total Land Area in Forests and Agriculture Year 1963 1973 1986 Agriculture 30 40 57 Forest 67 57 40 Source: Sol6rzano et al. 1991. Table 1-2: Effects of Different Future H-Values (10 billion colones) Undefined Defined property rights H-Value 0.0000 0.4792 0.2792 0.0792 Utility 0.0232 0.0232 0.0232 0.0232 GNP 3.1962 3.1972 3.1971 3.1971 Income 3.7681 3.7679 3.7679 3.7679 Deforestation Squatters 0.0020 0.0000 0.0000 0.0000 Loggers 0.0020 0.0000 0.0002 0.0010 TOTAL 0.0040 0.0000 0.0002 0.0011 Production Forest 0.0552 0.0713 0.0711 0.0691 AGRI 1.3984 1.3876 1.3876 1.3879 IND 1.8477 1.8416 1.8417 1.8424 Table 1-3: Effects of Taxes on Production Factors (undefined property rights; IO billion colones) Base tax subs tax subs tax subs tax subs case logs logs land land ulab ulab cap cap Utility 0.0232 0.0243 0.0214 0.0231 0.0226 0.0229 0.0232 0.0232 0.0233 GNP 3.1962 3.2017 3.1848 3.1960 3.1757 3.1797 3.1965 3.1943 3.1958 Income 3.7681 3.7748 3.7561 3.7678 3.7592 3.7620 3.7679 3.7680 3.7682 Deforestation Squatters 0.0020 0.0044 0.0004 0.0000 0.0255 0.0210 0.0000 0.0020 0.0020 Loggers 0.0020 0.0000 0.0018 0.0020 0.0019 0.0029 0.0013 0.0036 0.0011 TOTAL 0.0040 0.0044 0.0023 0.0020 0.0273 0.0239 0.0013 0.0056 0.0031 Production Forest 0.0552 0.0000 0.1562 0.0672 0.0000 0.0000 0.0689 0.0515 0.0574 Agricul- 1.3984 1.4204 1.3748 1.3877 1.5122 1.4922 1.3877 1.3987 1.3983 ture Industry 1.8477 1.9248 1.7017 1.8430 1.7656 1.7946 1.8424 1.8487 1.8471 1. Natural Resource Management and Economywide Policies in Costa Rica 15 Table 1-4: Effects of Taxes on Final Products (undefined property rights) Halve Double Halve Double Halve Double tax on tax on for- tax on tax on tax on tax on Base Case forest est agriculture agriculture industry industry Utility 0.0232 0.0229 0.0240 0.0135 0.0459 0.0064 0.0572 GNP 3.1962 3.1941 3.1980 3.1589 3.2815 3.1324 3.3253 Income 3.7681 3.7669 3.7709 3.7317 3.8544 3.7051 3.8984 Deforestation Squatters 0.0020 0.0020 0.0020 0.0019 0.0022 0.0018 0.0025 Loggers 0.0020 0.0020 0.0020 0.0020 0.0020 0.0020 0.0020 TOTAL 0.0040 0.0040 0.0040 0.0039 0.0042 0.0038 0.0045 Production Forest 0.0552 0.0528 0.0609 0.0405 0.0902 0.0160 0.1363 Agricul- 1.3984 1.3984 1.3985 1.3258 1.5706 1.3964 1.4030 ture Industry 1.8477 1.8503 1.8418 1.9243 1.6663 1.8268 1.8909 Figure 1-1: Main Linkages of the CGE Model Factor demand capital Factor supply Pactor markets Faator WageSs, costs rents| , Demand for J . _ < ~~intermediate goods Prdcrs | ousehold Sales Cnue revenues flexpniue . , s ~~~~~markets Supply of Consumption of produced goods Imports Exports final goods 16 VOLUMEII Figure 1-2: Resource Flows and Data Adjustments Domestic Total de- Forest Agriculture Industry Service Infrastructure Squatters Loggers Net export demand mand Forest Agriculture Final ~~Total Industry Intermediate demand Deforestation sectors demand production Service Infrastructure Squatters +Isq Defor demand Loggers +logf Labor capital -Igl -Isq +Isq +Igl Factor -Igk -Iv +Igk demand Land Indtax +Iv Indirect taxes Govemment revenues Total Output-producing sectors Deforestation Total final demand lsq = labor input in squatting; logf = logging; 1gl = labor input in logging; lgk = capital input in logging; Iv = land value Figure 1-3: Effects of Interest Rates When Property Rights Are Well Defined 0.0012 0.001 ' 0.0008 --- Squatters '0 0.0006 -A/ Loggers 0.0004 0.0002 0 0% 20% 40% 60% Interest rate 1. Natural Resource Management and Economywide Policies in Costa Rica 17 Annex 1-A: Summary of CGE Model Structure Factor market equilibrium and the stock of forested land The supplies of labor and capital are assumed to be exogenously given. According to Shephard's lemma the demand for a production factor within a sector is the partial derivative of the cost function with respect to the price of the same production factor. The market equilibrium conditions for capital, unskilled labor, and skilled labor, respectively, can be written a +U (1) where PK' is the user price of CAPITAL and k`' is the CAPITAL used in deforestation by loggers, and L = _jeT.N a + 1Ju + 1'°g. PLU ~~~~~~~~~~~~~~~~~~~~~(2) where Puu is the user price of UNSKILLED LABOR, 1sq and i"'g is the LABOR used in deforestation by squatters and loggers, respectively, and PLu is the user price of SKILLED LABOR. Equations (1) and (2) state that the supplies of labor and capital must equal the demand from the producing sectors plus the amount used for deforestation by squatters and loggers. Costa Rica's total area has been divided into two types of land: cleared land and forested land. Cleared land is produced through deforestation. The production of cleared land will depend on the definition of property rights, and taxes and subsidies will depend on the factors of production and the profits in the FOREST and AGRICULTURE sectors. It is assumed that there is a world market for LOGS. The market equilibrium condition for LOGS can then be written r dFOREST C FOR - dnexp a PUF ~~~~~~~~~~~~~~~~~~~(3) where dF RES is deforestation from the LOGGING sector, PFU is the user price of logs, f nep is the net export of logs, and a is a fixed coefficient reflecting the amount of timber extracted per unit of deforested land. The production of LOGS is further discussed in the section, The logging sector. The value of the deforestation is calculated here from the loss of standing volume in Sol6rzano et al. (1991). The coefficient a was assumed equal to one. There is no general agreement among biologists regarding the amount of biodiversity loss resulting from high-grading. If the value of deforestation is defined to represent biodiversity loss as well as loss of standing volume, and high- grading causes large losses of biodiversity, this modified coefficient would still be close to one. An alternative interpretation, consistent with the view that the biodiversity loss is proportional to the number of trees extracted, is that squatters and loggers each remove half the timber following the process described in Carriere (199 la and 1991b). In this case also, the modified coefficient would be nearly equal to one. Thus the model is valid for either of the last two interpretations, but the actual value of the economic loss would have to be adjusted upwards. Due to the lack of data on the value of biodiversity, the first approach of valuing only the timber was selected, since it is a conservative estimate. However, the true value of the loss of forests is higher. 18 VOLUME II The supply of cleared land is composed of the stock of cleared land, DLC, plus deforestation by squatters, d". The production of cleared land by squatters is further discussed in the section, Squatters. The demand of cleared land is the demand from the AGRICULTURE sector. The market equilibrium condition can then be written: a CAGRICULTURE -DL* do = O. a PUDL -4' The combination of production factors can be influenced by taxes and subsidies. The user price, Pju, will exceed the supply price, Pjs, by a percentage tax, Tj: PV = Pj (1+ T1); j = DLAND, LABOR, CAPITAL PLOCS = PLO/GS(J+TLOGS)- (5) In the case of LOGS, the supply price is determined by the world market price. Technology, costs, and producer behavior The production factors have been aggregated into a composite input, Y, using CES functions. The technology is specified to exhibit constant returns to scale. The relation between inputs and output is given by sectoral Leontief production functions of the following type: Xi = min[ i,Jij] i,jET,N Ai ai (6) where Xj is the gross output in sectorj, Yj is a composite input of production factors in sectorj, Xij is input of output from sector i in sectorj, and Aj and avj are Leontief input-output coefficients. Since the technology exhibits constant returns to scale, the marginal cost and the average cost of production in sectorj can be written as Cj = PyjA,+LiPjD aiT+tj; i,jET,N(7) where Cj is the marginal and average cost in sectorj, Pyj is the producer price of the composite input of production factors, pD iS the domestic price of sector i output, Aj is the use of production factors per sectorj output, a i is the use of sector i input per sectorj output, and tj is indirect tax per unit of sector j output. Producers are assumed to maximize profits. The producer output prices, Pi, in the tradables producing sectors are given by the world market prices. Assuming perfect competition, this implies that pure profits are nonpositive, and that output is non-negative and positive only if pure profits are equal to zero: Pi-ci < O, icT (Pi - Ci) Xi = 0; i E T Xi Ž 0; is T (8) In the nontradables-producing sectors, the sector-specific capital is endogenously adjusted so that price equals marginal cost: Pi = Ci; i E N (9) 1. Natural Resource Management and Economywide Policies in Costa Rica 19 Prices, domestic demand, foreign trade, and market clearing For goods produced in the tradables-producing sectors, the domestic producer price is equal to the world market price of identical goods, and in the nontradables-producing sectors, the domestic user price is equal to the producer price times the tax rate: PP = (1+a)P'e= (e + a) P; i E T, (10) PO = (I+ a1)P,; i E=N (11) where PiD is the domestic user price of goods produced in sector i, p,w is the world market price of good i, Pi is the producer price of good i, e is the exchange rate, and o, is the ad valorem tax rate on good i. The intermediate demand of good i is given by the technology assumptions. Domestic final demand (D) is given by utility maximization (equation 35). The net export (Z) is determined as the difference between domestic supply and demand. The market equilibrium conditions then become Xi = 7j.TNaijXj+Di+Zi; i ET (12) Xi = 1jE.TNaijXj+Di; iEN (13) Deforestation sectors In this model, there are two sectors that are responsible for deforestation. They are discussed below. The logging sector The logging sector is assumed to have a capital-intensive technology (Repetto 1988). Further, the technology is assumed to exhibit decreasing returns to scale in order to reflect the diminishing amount of available forests, as well as the fact that much of the logging is done illegally. The production of LOGS is assumed to depend only on two factors of production: UNSKILLED LABOR and CAPITAL. The loglinear production function is dFOREsT = (k)og)a(Iog0)3 c a+pf 0, 'dil (1q) < o, (19) and lim d(l.q) = B, n -> -0 (20) where n is the number of squatters, P' is the labor used in the clearing of the land, and B is the total amount of land available for deforestation. The squatters total revenue from clearing the land is the price paid for the cleared land: I (do) = PLL d" (21) where PDLS is the supply price of cleared land. Part of the land cleared by squatters is sold to the AGRICULTURE sector, and part of the cleared land is used for subsistence agriculture by the squatters themselves. However, since both occur, the revenue must be the same in both cases. The squatters are assumed not to sell the timber from their deforestation. Other uses of the timber, such as for firewood, are assumed to be negligible, and are therefore set to zero. 1. Natural Resource Management and Economywide Policies in Costa Rica 21 The squatters' total private cost for the clearing of the land will depend only on the amount of labor needed in order to clear the land when property rights are undefined, CY(d°) = P d-'J(do) (22) where PLsq is the price of labor in squatting. This is the private cost, and it does not include the future value of the forests and the cost of environmental damage. The total social cost of deforestation includes the future value of the cleared forests foregone by clearing the land today. The future value of the forests is assumed to be greater than the value of the forests today: H(F) > 1+i (23) Undefined property rights When property rights are undefined, the land is seen as a common property. There is no market for the forests available to squatters. The production of cleared land by squatter i is d,!'= I d(j,q) N (24) where N is the total number of squatters, and the squatters' total production of cleared land is do = EN, d17 (25) The total private profit from clearing the land is g' (do") = IX (di") - Cs (d`) (26) of which each squatter receives an equal share: 1 g' (do°) = - g' (d ") N (27) However, for the sector as a whole, land will be cleared until marginal cost equals marginal revenue. From the profit function, we have that the squatters' demand for ULABOR is I (PDL, PU) Pu 'y PDL (28) and the supply of cleared land is Y PDL (29) Well-defined property rights When property rights are well defined, there is a market for the forests. The squatters take the future value of the forests into account, and they can choose to clear forested land or to preserve the forests. This is consistent with the condition for socially optimal forestry that a tree should be harvested when the market value is equal to the shadow value (Hellsten 1988). Assuming that all squatters are identical, and that every squatter owns lAN of the land that previously was available for squatting, the total private profits from clearing the forested land is now (Johansson and Lofgren 1985): gf (d o) = P'. do - C- (d") -H(d°) I + i (30) 22 VOLUME I Deforestation will occur until marginal cost equals marginal revenue, and from the profit function, we have that P U+H(pqr1(1 + i)) s rPDL (31) ( (H(pqY) /(J -i))'~y Pu + ap d=l alV 1 YPDL (32) This result is equal to the result of the social optimization problem. Macroeconomic closure The current account is assumed to be constant. This implies that LeT.NPiZi = S (33) where S is the current account surplus. Equation (33) indirectly determines the disposable income (1). The gross national product is determined as the sum of factor incomes, plus a term reflecting the value of diminished deforestation: GNP = PU K + PU L + PU U + PUL DL* + E ,,ti - A(d- + d'og)H(l).+n(dS) + H(dg). (34) The model is solved by maximizing consumer utility subject to a budget constraint: Max U= rl,;,1 b,= 1, ieN,T S.t. I- XiPpDD= 0 (35) 1. Natural Resource Management and Economywide Policies in Costa Rica 23 Annex 1-B: Input-Output and Land Use Data, Costa Rica Table IB-i. Base Case Data for Costa Rica, 1986 (10 billion colones) FOREST AGRI IND INFRA SERVICE CONS NEXP TOTAL forest 0.0003 0.0022 0.0391 0.0002 0.0000 0.0124 0.0011 0.0552 agri 0.0004 0.4033 0.2488 0.0000 0.0000 0.3535 0.3924 1.3984 ind 0.0137 0.1405 0.7390 0.3418 0.1343 1.4426 -0.9643 1.8477 infra 0.0004 0.0293 0.0826 0.0647 0.0684 0.8366 1.0821 service 0.0038 0.0602 0.1546 0.1487 0.2160 1.1230 1.7062 land 0.2070 0.2070 logs 0.0022 0.0022 capital 0.0176 0.2052 0.2168 0.1633 0.5256 1.1285 ulab 0.0141 0.2714 0.1525 0.1754 0.1439 0.7573 slab 0.0002 0.0045 0.0999 0.1299 0.4904 0.7250 indtax 0.0025 0.0747 0.1145 0.0580 0.1276 0.3773 total 0.0552 1.3984 1.8477 1.0821 1.7062 3.7681 -0.5708 9.2870 GNP 2.6148 Note: indtax = indirect tax; nexp = net export; slab = skilled labor; ulab = unskilled labor. Source: Calculated from a disaggregated input-output table constructed by Edgar Bricefio, University of Costa Rica, San Pedro, and Costa Rica National Accounts. The adjustments have been calculated from Sol6rzano et al. (1991). Table 1B-2: Land Use Data 1963 1973 1986 Hectares Percent Hectares Percent Hectares Percent Agriculture 1,544,796 30.09 2,048,512 39.90 2,944,616 57.36 Primary for- est 3,154,280 61.44 2,666,005 51.93 1,760,622 34.30 Secondary forest 299,011 5.82 283,571 5.52 292,850 5.70 Other 135,593 2.64 135,593 2.64 135,593 2.64 Total 5,133,680 100.00 5,133,681 100.00 5,133,681 100.00 Source: Calculated from Sol6rzano et al. (1991). 24 VOLUME I Bibliography Alsabah, M. S. S. 1985. General Equilibrium Papers. Boulder, Colo., and London: Analysis of Government Expenditures in Westview Press. an Oil Exporting Country: The Case of Blomstrom, M., and M. Lundahl. 1989. Kuwait. Ph.D. dissertation, Harvard Uni- COSTA RICA en landstudie. Blickfang versity. Centralamerika, Latinamerika-institutet, Banco Central de Costa Rica. 1990. Cuentas Stockholm. Nacionales de Costa Rica 1980-1989. Bojo, J. 1991. "Econornics and Land Degrada- December. tion," Ambio 20(2). 1991 a. Cuentas Agropecuarias Bricenio, Edgar. 1986. "Input-Output Tables 1980-1989. January. for Costa Rica." Unpublished data. San . 199 lb. Estadisticas del Sector Indus- Pedro, Costa Rica: University of Costa trial Manufacturero 1980-1989. March. Rica. Barbier, E. 1989. Economics, Natural Re- Brook, A., D. Kendrick, and A. Meeraus. sources and Development. Conventional 1988. GAMS-A User's Guide. Redwood and Alternative Views. London: Earthscan City, Calif.: Scientific Press. Publications Ltd. Burns, D. 1986. Runway and Treadmill De- Barbier, E., J. Burgess, and A. Markandya. forestation: Reflections on the Economics 1991. "The Economics of Tropical Defor- of Forest Development in the Tropics. estation." Ambio 20(2). London: Earthscan. Benjamin, N. C. 1990. "Investment, the Real Carriere, J. 199 la. "The Crisis in Costa Rica: Exchange Rate, and Dutch Disease: A an Ecological Perspective." In D. Good- Two-Period General Equilibrium Model man and M. Redclift (eds.), Environment of Cameroon." Journal of Policy Model- and Development in Latin America. The ing 12(l):77-92. Politics of Sustainability. New York: Benjamin, N. C., S. Devarajan, and R. J. Manchester University Press Weiner. 1989. "The 'Dutch' Disease in a Carriere, J. 1991b. "The Political Economy of Developing Country: Oil Reserves in Cam- Land Degradation in Costa Rica." Interna- eroon." Journal of Development Econom- tional Journal of Political Economy 21(1): ics 30:71-92. 10-31. Bergman, L. 1988. "Energy Policy Modeling: Chichilnisky, G. 1993. "North-South Trade A Survey of General Equilibrium Ap- and the Dynamics of Renewable Re- proaches." Journal of Policy Modeling sources." Structural Change and Eco- 10(3). nomic Dynamics 4(2). --. 1990a. "Energy and Environmental Cruz, W., and R. Repetto. 1992. The Environ- Constraints on Growth: A CGE Modeling mental Effects of Stabilization and Struc- Approach." Journal of Policy Modeling tural Adjustment Programs: The Philip- 12(4):671-91. pines Case. Washington D.C.: World . 1990b. "General Equilibrium Effects Resources Institute. of Environmental Policy: A CGE-Model- Dasgupta, P. 1982. The Control of Resources. ing Approach." Research Paper 6415. Oxford: U.K.: Basil Blackwell. Economic Research Institute, Stockholm. Dasgupta, P., and K.-G. Maler. 1990. "The Biesanz, R., K. Biesanz, and M. Biesanz. Environment and Emerging Development 1987. The Costa Ricans. Englewood Issues." Proceedings of the World Bank Cliffs, N.J.: Prentice-Hall. Ainual Conference on Development Eco- Bishop, R., and S. Andersen (eds.). 1985. nomics 1990. Reprint. Natural Resource Economics. Selected 1. Natural Resource Management and Economywide Policies in Costa Rica 25 Denevan, W. M. 1988. "Causes of Deforesta- 1982. COSTA RICA Country Environmen- tion and Woodland Degradation in Tropi- tal Profile. A Field Study. San Jos& Tropi- cal Latin America." In L. Fortmann and cal Science Center. J. W. Bruce (eds.), Whose Trees? Propri- Hellsten, M. 1988. "Socially Optimal For- etary Dimensions of Forestry. Rural Stud- estry." Journal of Environmental Econom- ies Series. Boulder: Westview Press. ics and Management 15:387-94. Dervis, K., J. de Melo, and S. Robinson. Hugo, V., V. Gonzalez, R. Jimenez, and T. 1982. General Equilibrium Models for Vargas. 1983. Problemas Economicos en Development Policy. Press Syndicate of la Decada de los 80. San Jose: Agencia the University of Cambridge, Cambridge. para el Desarrollo International. Devarajan, S. 1990. Can Computable General Johansson, P. O., and K. G. L6fgren. 1985. Equilibrium Models Shed Light on the The Economics of Forestry and Natural Environmental Problems of Developing Resources. Oxford, U.K., and New York: Countries? Paper prepared for WIDER Basil Blackwell. conference on "The Environment and Jorgenson, D. W., and P. J. Wilcoxen. 1990. Emerging Development Issues," Helsinki, "Intertemporal General Equilibrium Mod- September 3-7, 1990. eling of U.S. Environmental Regulation." Edelman, M. 1985. "Extensive Land Use and Journal of Policy Modeling 12(4): the Logic of the Latifundio: A Case Study 715-44. in Guanacaste Province, Costa Rica." Keogh, R. M. 1984. "Changes in the Forest Human Ecology 13(2). Cover in Costa Rica through History." Fisher, A. 1981. Resource and Environmental Tuirrialba 34(3):325-3 1. Economics. Cambridge, U.K.: Cambridge Larson, B. 1991. "The Causes of Land Degra- University Press. dation along 'Spontaneously' Expanding Forsund, F. 1985. "Input-Output Models, Agricultural Frontiers in the Third World: National Economic Models, and the Envi- Comment." Land Economics 62(2): ronment." In A. V. Kneese and J. L. Swee- 260-66. ney (eds.), Handbook of Natural Resource Lutz, E., and H. Daly. 1990. "Incentives, Economics, Vol. I. New York: North- Regulations, and Sustainable Land Use in Holland, Elsevier Science Publishers. Costa Rica." Environment Working Paper Forsund, F., and S. Strom. 1980. Miljf og No. 34. Washington, D.C.: World Bank. ressurs fkonomi. Oslo: Universitets- Maler, K.-G. 1985. "Welfare Economics and forlaget. the Environment." In A. V. Kneese and Foy, G., and H. Daly. 1989. "Allocation, J. L. Sweeney (eds.), Handbook of Natural Distribution and Scale as Determinants of Resource Economics, Vol. I. New York: Environmental Degradation: Case Studies North-Holland, Elsevier Science Publish- of Haiti, El Salvador and Costa Rica." ers. Environment Working Paper No. 19. Ministerio de Recursos Naturales, Energia y Washington, D.C.: World Bank. Minas. 1990. Estrategia de Conservaci6n Harrison, S. 1991. "Population Growth, Land para el Desarrollo Sostenible de Costa Use and Deforestation in Costa Rica, Rica. San Jose: ECODES. 1950-1984." Interciencia 16(2). Moltke, K. von. 1990. "International Eco- Hartshorn, G., L. Hartshorn, A. Atmella, L. nomic Issues in Tropical Deforestation." Diego G6mez, A. Mata, L. Mata, R. Mo- World Outlook, summer. rales, R. Ocampo, D. Pool, C. Quesada, C. Munasinghe, M. 1993. Environmnental Eco- Solera, R. Sol6rzano, G. Stiles, J. Tosi, A. nomics and Sustainable Development. Umania, C. Villalobos, and R. Wells. Washington D.C.: World Bank. 26 VOLUMEII Munasinghe, M., and W. Cruz. 1994. Schelhas, J. 1991. "A Methodology for As- Economywide Policies and the Environ- sessment of External Issues Facing Na- ment. Washington D.C.: World Bank. tional Parks with an Application in Costa Munasinghe, M., W. Cruz, and J. Warford. Rica." Environmental Conservation 18(4). 1993. "Are Economywide Policies Good Shephard, R. W. 1981. Cost and Production for the Environment?" Finance and De- Functions. New York: Springer-Verlag. velopment 30(3):40-44. Sol6rzano, R., R. de Camino, R. Woodward, Panayotou, T., and C. Sussangkarn. 1992. J. Tosi, V. Watson, A. Vasquez, C. "Case Study for Thailand." In D. Reed Villalobos, J. Jim6nez, R. Repetto, and W. (ed.), Structural Adjustment and the Envi- Cruz. 1991. Accounts Overdue. Natural ronment. Boulder, Colo., and Oxford, Resource Depreciation in Costa Rica. San U.K.: Westview Press. Jos6 and Washington, D.C.: Tropical Persson, A. 1994. "Deforestation in Costa Science Center and World Resources Rica: Investigating the Impact of Market Institute. Failures and Unwanted Side Effects of Southgate, D. 1990. "The Causes of Macro Policies Using Three Different 'Spontaneously' Expanding Agricultural Modeling Approaches." Beijer Discussion Frontiers in the Third World." Land Eco- Paper Series No. 48, Stockholm. nomics 66(1):93-101. Ravent6s, P. 1990. Commercial and Tax Thrupp, A. 1981. "The Peasant View of Con- Reform in Costa Rica in the Mid 1980's servation." Ceres, July-August. (preliminary version). Alajuela, Costa Unemo, L. 1993. "Environmental Impact of Rica: INCAE. Government Policies and External Shocks Ravent6s, P. 1991. Personal communication. in Botswana-A CGE-Model Approach," INCAE, Costa Rica, July. Beijer Discussion Paper Series No. 26. Repetto, R. 1988. "Economic Policy Reform Varian, H. 1984. Microeconomic Analysis. for Natural Resource Conservation." Envi- New York: W. W. Norton & Company. ronment Department Working Paper No. Wilen, J. E. 1985. "Bioeconomics of Renew- 4. Washington, D.C.: World Bank. able Resource Use." In A. V. Kneese and Sader, S., and A. Joyce. 1988. "Deforestation J. L. Sweeney (eds.), Handbook of Natural Rates and Trends in Costa Rica, 1940 to Resource Economics, Vol. I. New York: 1983." Biotropica 20(1):11-19. North-Holland, Elsevier Science Publish- ers. _ 17 IZ,SJX 2 Evaluating Economywide Policies in the Presence of Agricultural Environmental Externalities: The Case of Ghana Ramon Lopez THE CLOSE DEPENDENCE OF RURAL INCOME on environmental factors in poor areas of the world is by now widely accepted. Excessive deforestation, significant declines in natural biomass, and reduction in soil quality, in many cases induced by rapid population growth, have been identified as key factors explain- ing an apparent fall or stagnation of rural income in Africa and in certain areas of Latin America and Asia (ADB, ECA, and OAU 1984; FAO 1986). The fact that a significant part of the land used in farming and most forested areas in developing countries are controlled through various forms of common property and, sometimes, public property, has been emphasized as a source of resource overexploitation (Perrings 1989; Sinn 1988; Glantz 1977; Allen 1985; L6pez and Niklitschek 1991).' The author is grateful to Erik Lichtenberg for useful comments. 1. Feder et al. (1988) have empirically documented the negative effects of insecure land tenure property rights on agricultural productivity. 27 28 VOLUME II Whether indigenous common property economywide policies on agricultural income resource controls do cause overexploitation by explicitly accounting for such environmen- (i.e., whether the "tragedy of the commons" tal effects. applies) has been the subject of much contro- In doing this, we estimate a production versy. Various authors have argued that tradi- function that, in addition to the conventional tional communities develop controls on the inputs, accounts for the effects of environmen- use of the resources inducing socially efficient tal factors. The main environmental factor exploitation (Dasgupta and Mailer 1990; included in the analysis is the role of land Larson and Bromley 1990). That is, traditional degradation. Based on these estimates, we are systems would internalize the potential exter- able to obtain an idea of the extent of nalities arising from lack of individual re- overexploitation of the environmental re- source ownership. Whether or not this is true source (i.e., the extent by which fallow peri- can only be elucidated through empirical ods are too short). In doing this analysis, we work. To our knowledge, the only empirical use a unique data set for an area in western test of the commons' efficient resource exploi- Ghana. These data match household survey tation hypothesis has been developed by information from twelve villages covering the L6pez (1993) for western Cote d'Ivoire. period 1988 through 1989 with remote sensing L6pez rejects this hypothesis, documenting data on (a) area under closed forest, (b) area socially excessive deforestation and biomass under natural bush (fallow area), and (c) depletion which causes large income losses density of biomass in the fallow areas. The for the rural communities. detailed survey includes data on production, The large number of studies focusing on the land use, employment, use of conventional implications of rural environmental degrada- inputs, and demographic characteristics for a tion have permitted a qualitative understand- large number of farm-households in the ing of how the income of rural communities is twelve villages. Field visits to the area pro- affected by policies that directly or indirectly vided important qualitative information to impinge on resource utilization patterns. complement the survey and remote sensing However, these qualitative analyses have not data. yet been followed by empirical work that A secondary goal of the study is to relate would allow us to obtain a clearer idea of the the agricultural implications of over- quantitative importance of the environment as exploitation to the wider economy. A simple a factor of production.2 The absence of such general equilibrium model is used to measure estimates has, of course, made it impossible to how agricultural price policies, trade policies, measure the effects of government policies on and wage and public sector employment agricultural income although suitable theoreti- policies are likely to affect the contribution of cal models on this subject do exist (e.g., agriculture to national income. Because of L6pez and Niklitscheck 1991; Deacon 1992). data limitations, this second part of the exer- The principal purpose of this study is to cise is necessarily provisional and less de- estimate the value of environmental resources tailed than the first. Thus, the results have to as factors of agricultural production and to be interpreted in this light. Our motivation is measure the potential effects of various mainly to illustrate how unintended environ- mental impacts of economywide policies in agriculture could also have possible conse- 'An exception is the above-mentioned study by quences on national income. For example, it is Lopez (1993) who obtained estimates for the contribu- not clear that complete elimination of tariffs tion of biomass to agricultural production and for the and export taxes will necessarily cause an effects of output prices or biomass and agricultural increase in national income (or better welfare) productivity. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 29 in the presence of other distortions. Based on liberalize the economy. The real exchange rate empirical estimates of the magnitude of the has been adjusted to levels that are consistent various distortions, environmental and con- with external equilibrium; the public sector ventional, we try to measure the impact of has been downsized; protection to the indus- reducing import protection and export taxation trial sector has been substantially decreased; on national income. Additionally, we discuss and prices of agricultural goods, particularly the consequences of using primarily wage cuts exportables (cocoa), have been brought to or public employment reductions to limit the levels much closer to international prices. The fiscal deficit in the presence of multiple dis- above reforms have had a positive impact on tortions. economic performance. Evaluating economywide policy reform in Despite the impressive depth of the re- the context of environmental externalities forms, there are still several remaining policy constitutes a major challenge. First-best in- issues: struments to remove existing externalities are (a) Although quantitative import restrictions not always available, and the transaction costs have been completely eliminated, the associated with them can be extremely high. industrial sector is still protected via Setting first-best taxes is in many cases politi- tariffs. The average import tariff rate cally difficult and, moreover, many govern- including special taxes is about 20 per- ments, particularly in poor developing coun- cent ad valorem with considerable varia- tries, do not have the administrative and legal tions across goods. capabilities to enforce first-best regulations. (b) Although the industrial sector is the This may leave second best approaches as the most important import substitution sec- only options to mitigating the negative effects tor, a few agricultural goods, mostly of externalities. The environmental impact of cereals, are also protected via import economywide policy reforms may, in this duties that reach about 18 percent. context, considerably affect the quantitative (c) The implicit export tax on cocoa is still and even qualitative effect of these reforms on substantial, about 24 percent ad valorem production and eventually on national income (this implicit tax has been reduced from and welfare. almost 50 percent in 1985). The chapter is organized as follows: in the (d) The public sector deficit is still large, section, Policy Issues in Ghana, we provide a and efforts are now under way to reduce brief review of the evolution of the Ghanaian it. This requires both large further reduc- economy as well as a discussion of the major tions in employment in the public sector current policy issues. The empirical estimation and, possibly, additional cuts in real of the production function and of land cultiva- wages paid by the public sector. tion decisions is presented in the section, Each one of the above policy issues implies Estimating Production Functions and Land a potential need for further policy reforms. Allocation Decisions. The general equilibrium These additional reforms include further model is presented in the section, General decreases in protection to the import substitu- Equilibrium Simulations, where we also tion sector, further cut on the implicit export discuss the results of simulating various policy tax to agriculture and wage and public em- reform scenarios. The final section presents ployment reductions to control the public our conclusions. deficit. Policy Issues in Ghana Since 1983 the government of Ghana has introduced structural reforms to stabilize and 30 VOLUME II Estimating Production Functions not necessarily imply overcultivation of the and Land Allocation Decisions resource. It is possible that the community exerts certain controls on the allocation of the In the context of traditional agricultural land to allow for its optimal exploitation. The systems prevailing in western Ghana, the controls may be inadequate or insufficient; in predominant cultivation practice is shifting which case the community as a whole may cultivation. This involves long cycles over experience income losses, associated with which land is cultivated for one or two years excessive loss of biomass, erosion, and flood- and left idle for four to ten years to restore its ing. Based on estimates of the production productive capacity. The fallow or rest period function and land allocation decisions, we allows the natural vegetation to re-emerge and provide here some clear indication that indeed generate biomass that is used as natural fertil- such controls are not sufficient and that too izer in the next cultivation period. Also, the much land is cultivated. Fallows are too short root system of the natural vegetation provides and the village communities could signifi- physical support to the inherently unstable cantly increase their total income by reducing tropical soils. At the same time, the patches of the cultivated area. bush spread out over the land help protect the Since individual farmers can move around cultivated areas from erosion, flooding, and within the village land areas in search of land other detrimental factors. This provides the to cultivate, the relevant biomass variable rationale for postulating that biomass consti- needs to be defined at the village level as well. tutes an important factor of agricultural pro- Thus biomass is defined as the total village duction, and that its depletion, via more exten- area under fallow times the average vegetation sive cultivation and the shortening of fallow density per acre of fallow. These data are periods, is likely to have a negative effect on obtained from remote sensing based on satel- productivity. In this section, we estimate an lite observations for each village area in 1988 extended production function that allows us to and 1989 (see Annex 2-C). We have data on empirically measure the importance of natural ten villages in western Ghana. These data biomass as a factor of production vis-a-vis the were matched with Living Standards Survey more conventional factors of production. (LSS) data (World Bank 1993), which provide Another important feature of traditional information on sixteen farm-households systems is that land property rights are not in sampled in each village and each year. The general well defined. A large proportion of the household data that we use are output value, land available in a village is reserved exclu- family and hired labor used by the farm- sively for use by the villagers. The land is household, land cultivated by each farmer, generally owned by the community (based on farm tools and labor used, use of fertilizers or lineage) rather by individuals, but its use by other purchased inputs, and some demo- outsiders is normally restricted. That is, this is graphic variables, such as education, age, a classic case of common property with closed family size, and ethnic background. In esti- or highly restricted access for external poten- mating the land allocation decisions we also tial users. The community control of the use data on wages deflated by output prices resources is consistent with and facilitates the per village. shifting cultivation practices. The individual has exclusive rights on the land that he or she Production function estimation actually cultivates, but, once the land is left Table 2-1 provides various estimates of a idle in fallow, the land can be reallocated by double-log production function for the ten consensus and consent of the village chief villages in Ghana. Not all household data Common property in these conditions does turned out to be consistent, and some house- 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 31 hold data were simply nonexistent. For this cocoa production could be that cocoa produc- reason, we could use data for 139 households ers are usually more integrated into the agri- from the ten villages and over a period of two cultural extension systems and other govern- years. For the environmental variable, we used ment support services than farmers that do not two variables, one that accounts only for the produce cocoa.4 proportion of fallow land in the village area The high and significant value of the ordi- and another one that considers biomass vol- nary least squares (OLS) coefficients of the ume properly defined as area under fallow environmental variable (Table 2-1, Column 1) times the average biomass density per acre of indicates that the environmental factor is a fallow. The use of fallow land as a factor of very important determinant of agricultural production is appropriate, since fertility of the productivity. Another possible interpretation, land actually cultivated is determined by the however, is that a village may have more proportion of land under fallow.3 Besides the biomass than another because it has better soil fertility effect, the level of biomass in the quality and climate, which allow the natural areas surrounding the cultivated land patches vegetation to grow faster. This better climatic help protecting against flooding and other and soil characteristics would also lead to physical damages. more productive agriculture in this village, The estimates of the production function and thus the positive correlation found be- appear highly plausible. Moreover, the tween output and biomass would be spurious. goodness-of-fit statistic is large, and the de- Two approaches were followed to check the gree of stability of the coefficients is quite relevance of this interpretation: (a) we used remarkable. The estimates of the effect of the area under fallow per village as an instrument biomass variable are statistically significant for biomass volume, and (b) we control for and are comparable in magnitude with those village specific characteristics such as climate associated with conventional factors. We also and soil quality using village dummy vari- used other variables, such as fertilizers used ables.5 and family education, but these variables The area under fallow variable need not turned out extremely insignificant. The fact suffer from the shortcomings discussed above. that only a small proportion of all farmers use Fallow area is not likely to be positively fertilizer and usually in small amounts may correlated with soil quality and good climatic explain the lack of significance of this vari- conditions. Nonetheless, it is positively corre- able. Also, education is quite homogeneously lated with biomass density. In a sense, fallow low among most farmers, and thus the lack of area is an almost ideal instrument for explanatory power of this variable. The esti- mates suggest that farmers who produce cocoa tend to be more productive than those who do not, and that productivity declined in 1989 vis-a-vis 1988. A possible reason for the 4Additionally, we used two dummy variables positive correlation between productivity and (Dummy size A and B) to control for the possibility of technological differences associated with either ex- tremely small or extremely large farm sizes (i.e., for farm sizes outside the mean + 3 standard deviations 3This is clearly appropriate under the assumption of range). Further, we also use a dummy variable (Dummy steady state, i.e., when the stock of biomass is constant. 89) to control for possible changes in productivity due Outside the steady state the relationship between to weather or other reasons between 1988 and 1989. cultivated land fertility and the proportion of fallow land is less direct and is also affected by the rate of 'Procedure (a) is consistent with a long-run biomass accumulation or depletion. However, even interpretation because biomass density depends on the outside the steady state, cultivated land fertility is still proportion of the land under fallow in steady state (see affected by the proportion of fallow land. L6pez 1993). 32 VOLUME II biomass.6 The fact that using the village fal- for labor and 26 percent for capital. Since the low area variable as an instrument yields share of agricultural output in national gross highly significant and positive estimates for domestic product (GDP) is about 50 percent, the effect of biomass (see Table 2-1, Columns this means that the contribution of biomass to 2 and 3) suggests that the causality interpre- national income is about 7.5 percent. Thus, the tation-that more biomass causes greater stock of biomass is an important determinant agricultural productivity-is correct. of not only agricultural production but also of Using village specific dummies should total GDP. A 10 percent reduction of the control for unobserved variables, such as biomass stock, for example, could lead, ceteris climate and soil quality, and thus elirninate the paribus, to a fall in agricultural output of the influence of such variables from the coeffi- order of 1.7 percent and to a reduction of cients associated with the other explanatory aggregate GDP of 0.75 percent. variables.7 The use of village-specific dum- mies (rather than household dummies) is an Land cultivation decisions adequate procedure because the biomass Table 2-2 provides estimates of the land variable used is also defined at the village cultivation decisions by the individual farm- level. Column 3 provides the estimates ob- ers.8 Column 1 provides estimates conditional tamed using village-specific dummy variables on a given level of family size, while the and the instrumental variables estimators. The estimators in Column 2 are obtained by as- estimators of the biomass coefficients re- . . . . mained large and sinfcn. Tu,bt suming that family size is endogenous. In this sense, the estimates in Column 2 can be inter- procedures suggest that there is a meaningful preted as more long-rn than those in Column causal relationship going from the level of the 1. The key variable here is the agricultural environmental factors to agricultural output. wage deylatedaby he t p ric.Th ri- wage deflated by the output price. This vani- Apart from the statistical significance of the able is defined per village for each year. Its coefficients associated with biomass, it is . . important to emphasize their quantitative ncoefficent suggests that the wage rate exerts level. Th bioass oeffiientestiated a negative effect and that the agricultural output price has a positive effect on land using instrumental variables and controlling cultivated. The elasticity for the wage rate is for village dummies was 0.17 (Table 2-1, about -I for the short-run estimates (Column Column 3). This means that the contribution 1) and about -1.26 for the long-run estimates of biomass to agricultural production is about (Column 2). 17 percent, or equivalently, that the share of (oumnc2). ' . ~~~~~Thus, increasing agricultural prices or wage biomass in total agricultural production once reductions cause an expansion in the culti- this factor is measured at its shadow value is vated area and, therefore, a reduction in the 17 percent. This compares with a contribution fallow area and a fall in biomass, which in of 26 percent for land cultivated, 25 percent turn reduces agricultural productivity. Accord- ing to the estimates in Table 2-2, the elasticity of land cultivated with respect to agricultural 61f anything, fallow area should be negatively output price is in the 1 to 1.26 range. The correlated with soil quality and good climatic condi- other explanatory variables have all the ex- tions. One could reasonably argue that villages that pected signs and are significant. Family size have good soils and climate would tend to cultivate has a large positive effect on the amount of more land (and hence have less fallow) because land is more valuable. land cultivated. This finding is very consistent 7The village dummies also account for other village specific variables such as distance to markets, transpor- 'The conceptual basis for the derivation of the tation facilities, etc. land use specification is provided in Annex2-A. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 33 with the idea that population pressure is an = 0. In this case, important factor behind environmental degra- dation. Capital equipment represented by the _ .3x number of tools and implements owned by a x o (3) household also exerts a positive effect on the area that families cultivate. Similarly, it ap- That is, the level of biomass density in the pears that, other conditions being equal, estab- long run is inversely related to the proportion lished local residents (Ashanti and Bron of the total land area that is cultivated. Using tribes) tend to cultivate less land than rela- (3) in (i), we obtain tively new immigrants (the rest of the popula- tion). Tradeoffs between expansion of cultivated land, biomass, and agricultural That is, the total biomass volume is decreasing productivity with the proportion of cultivated land (xli) and The estimates of the production function increasing with the rate of growth of natural expose the tradeoffs associated with expand- vegetation, a. From (1) it is clear that the ing the cultivated land. The direct positive elasticity of total biomass with respect to area effect of expanding cultivation on production under cultivation can be defined by has to be weighted against the negative effect d In 0 of such an expansion on the level of biomass d In x _ I (4) and, therefore, on output. To evaluate this d In x 1 x/x tradeoff, it is necessary to quantify the effect If, for example, 25 percent of the village land of cultivated land on the level of biomass. is exate, 25 incent cultated Total biomass can be expressed as total fallow is cultivated (xp x = 0.25), increasing cultivated area imesthe aerag bioass dnsit per land by 1 percent will decrease total biomass area times the average biomass density per by 1.33 percent, or equivalently, the elasticity acre of fallow land, is -1.33. Using (4) and the estimated coefficients for 0 = ( - x) (1) the land cultivated and biomass variables, one where Y7 is average biomass density per acre of can calculate the full effect of increasing wherew land, x is total land area in a village cultivated land on agricultural output.9 If, for fallow land, x is total land area in a village, example, we use the estimates in Column 3 of and x is the total land cultivated. Thus, x - x is Table 2-1, we have the following result: in- the area under fallow. The average density q creasing land cultivated by 10 percent would is not independent of the area cultivated ei- induce a 2.7 percent increase in output as a ther. Following L6pez (1993), we can postu- direct effect. However, the 10 percent increase late a simple specification for the dynamics of i Q m~~~~~~~~~~~i land cultivated also induces an approxi- mately 14.5 percent reduction in total biomass (in fact, the average value for x/x in the 10 nt = a- (2) villages analyzed is about 0.31). A 14.5 per- cent reduction in biomass in turn would cause where a > 0 represents the natural growth of vegetation, and x/x is really the rate of depre- 91n doing this, we are assuming that the system is in ciation of biomass in the context of a rota- long-run equilibrium. To consider situations outside long-mn tional system. In the long run when the level equilibrium, the specification of the empirical and theoretical of biomass becomes stable, we have that models needs to be much more complicated and requires a much larger data set than what we have (see L6pez 1993). 34 VOLUME II according to the estimates in Column 3, a 2.5 percent decrease in agricultural output. Thus, - x) F. the net effect of expanding area cultivated on r + x/x agricultural output is still positive, but of the order of 0.2 percent, many times smaller than in which case the effect of expanding culti- the direct effect. The first row of Table 2-3 vated land on national income is shows the net effect of expanding the area cultivated by 10 percent using the estimates from Table 2-1. In general, the net effects on d In YA a Iln F (1 + r)x/x output of expanding land cultivation are d In x a In x r + x/x positive, but very small. The second row of Table 2-3 shows the net a In F d In 0is (6) effect of land cultivated on net farm income a ln 0 d ln x J rather than on output under each scenario. Net farm income is defined as the gross value of output less land clearing costs and less (plus) the value of the biomass decrease (increase). where s £! is the share of land clearing That is, since expanding land cultivated re- c y quires investing labor and other resources in cost in net revenue. The first right-hand term first clearing and preparing the land, a more in Equation (6) captures the direct effect of appropriate analysis of the impact of expand- expanding land cultivation, which is positive. ing cultivated land on farm income should The second right-hand term accounts for the account for the costs embodied in expanding present value of the future effects on income land into cultivation (since the system is of reducing biomass associated with the land rotational, these clearing costs are recurrent cultivated expansion. (Note that this term is rather than once-and-for-all). In making their negative because dln /dlnx is negative.) The land allocation decisions, however, farmers last right-hand term accounts for the direct affect their level of biomass assets as well. costs involved in expanding land cultivation. Hence, total agricultural income is The expression in square brackets in the right- YA =pF(L,x,z,0) -cx+ull (5) hand side of (6) is evaluated by using the coefficients in Table 2-1 and an evaluation of Equation (4). According to information ob- where p is output price, L iS total labor used, tained in the field, it was estimated that the x is land cultivated, z are other inputs used, c share of land clearing cost in net revenues is is the per acre land clearing cost, q is defined between 12 percent and 17 percent. We use in (2), and u is the shadow price of the stock the lower bound estimate and thus assume that of biomass density, sC = 0.12. Using this, one can calculate the total effect of increasing cultivated land by 10 percent on net farm revenues. These estimates u(t) = jP(x - x)F6 (L, X, Z, 0) are provided in the second row of Table 2-3. As can be seen, expanding cultivated land has e r(u- t) d, a negative effect on farm income despite that we use the most conservative estimates of where rsis the discount rateandy Ft a F/8. It land clearing costs. If we assume that can be shown that in steady state XC = 0.16 rather than 0.12, the effect of ex- panding land cuitivate on income becomes substantially more negative, reaching a value 2. Evaluating Economywide Policies with Agricultural Environmental Externalities. Ghana 35 of -0.75 percent when the OLS estimates are two general equilibrium models, one that used and -1.0 percent when the Instrumental postulates a tight and competitive labor mar- Variables estimates are used. Thus, the empir- ket, and another that assumes urban unem- ical evidence suggests that too much land is ployment or subemployment in the Harris- cultivated and that any further expansion of Todaro tradition. The main feature of the the cultivation level is detrimental for village Harris-Todaro model is that rural and urban income once a full account of the indirect wages for identically skilled workers are not effects is considered. Table 2-3A provides equalized."1 The first model assumes a com- estimates of the degree by which land is over- petitive labor market which leads to equal cultivated and the income losses implied by wages for given skill levels. Both models such cultivation. As can be seen, despite the assume a small open economy with exoge- use of conservative estimates, the income nously given border prices. losses are quite large, reaching almost 3 per- cent when our best econometric estimates are Environmental distortions in a Harris- used. Todaro model To specify a general equilibrium model we first need to specify the land clearing cost (c) In the policy analysis, we also need esti- in terms of real factor use. We assume that mates for the agricultural labor demand elas- land clearing requires labor in fixed propor- ticities. Given the estimates of the production tion. We specify that y days of labor are function and the land equation the labor elas- necessary to clear one acre of land. In this ticities are indeed implicitly defined. Using case, c = yw and the agricultural production the production function estimates of Table function need to be modified slightly to ac- 2-1, Column 3, and the land equation esti- count for this, F(L - yx, x, 8). This specifica- mates reported in Table 2-2, Column 2, it can tion accounts directly for the cost of clearing be shown that the implicit labor demand land. An increase in x has a direct positive elasticity is about -1.26.' "By directly estimat- effect on output and two indirect negative ing a labor demand equation, we obtain esti- effects, decreasing 8 and decreasing the mates somewhat higher. Thus, it appears that amount of labor that is effectively devoted to demand for labor in western Ghana agriculture production. is highly elastic. National income for a small economy evaluated at the opportunity cost of resources General Equilibrium Simulations needs to be specified in terms of world prices rather than at domestic prices. Thus, national In this section, we evaluate the impact on income can be defined as national income of various government poli- cies. The policies analyzed are the following: (a) a net reduction of agricultural export taxes, (b) a decrease in tariff protection to the indus- trial sector, (c) a decrease in government employment as a measure to control the fiscal deficit, and (d) a decrease in public sector 1"Editors' note. This is in contrast to the standard wages as an alternative to reduction of public assumption that wages equalize with competition and employment. We perform the analysis using labor mobility. The Harris-Todaro framework recog- nizes the role of imperfect information, limited labor mobility and institutional constraints, leading to signifi- cant differences in wages in distinct sectors of the '"See Annex 2-B for a derivation of this result. economy. See, for example, Todaro (1989). 36 VOLUME II Y* =p *F(L-yx,x,() +q G(LN;KN)1 + t) q *G1( ) =w (9) + aLg + Utl (7) where t is the ad valorem tariff rate that pro- tects the nonagricultural sector (note that the where p* and q* are the international prices of domestic price of the nonagricultural good is agriculture and the urban goods, respectively, q -=(1 + t)q*). Thus, from (9), employment in G( ) is the production function of the urban the private urban sector is L = L((1 + t)qw, wv). goods, LN is labor, and KN is other factors of The relationship between iwv and w is given by production used in production of the urban the well-known equilibrium condition that the good. Lg is government employment, a is wage in the rural sector is equal to the ex- labor productivity in government, and thus pected wage in the urban sector, atLg is production of government services (we have normalized the price of government where L is the total labor force. The expres- services to one). L + L The other components of the model are the w = N W (10) profit maximization conditions for farmers L - L that determine L and x, sion in square brackets corresponds, therefore, to the rate of employment in the urban sector. (1 -u)p *F, O = w (8)(i) Since w is given exogenously by government regulations regarding public sector wages, wage indexation, minimum wages and fringe F2 - yF~ - X'nF3 1 + r a8i benefits, in general there is (urban) unemploy- r + x/x ax ment or underemployment. This implies that the term in square brackets in (10) is less than one and hence, w < W. The level of unemploy- ment or underemployment (U) is determined where r is the ad valorem tax rate affecting by agriculture production ((I - v)p* --p is, there- fore, the domestic price of the agricultural - 1 q goods), and 0 ,? I is an efficiency parame- U=L-L -LN((l + t) q , w) ter determining the extent by which the effect - *, w) (1 of changing cultivated area on biomass is taken into consideration in the land allocation decisions. Subscripts in the function indicate We assume that Lg is determined by the gov- first derivative with respect to the correspond- ernment exogenously. This implies that U is ing term. Note that yF, is the real cost (de- determined recursively after w, L, and LN are flated by (1 - z)p*) of clearing land (what was solved from the rest of the model. called c before). From Equations (8(i)) and With the above specification, we can illus- (8(ii)), one can solve for x((l - r)p*,w) and trate the possible effects of the following L((1 - -r)p*,w). policies on national income, Y*, and the envi- The level of employment in the urban ronmental resource, d: private sector, LN, is determined by the level of the urban wage rate, W, which in a Harris- (a) Decreasing the degree of taxation of Todaro model is different from w, agriculture, i.e., decreasing vor equiva- lently increasing p for a given p*. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 37 (b) Reducing protection to the industrial rural wage dispersion, although significant, is sector or trade liberalization, i.e., reduc- not too large (smaller than in many other ing t (or, equivalently, increasing q for a developing countries) and open urban unem- of given q*). ployment is small compared to other develop- (c) Government wage policies that reduce ing countries even if one assumes consider- labor regulations and public sector ably downward biases in the available statis- wages, i.e., a reduction in W. tics. This decreases our confidence that the (d) Reducing public sector employment, i.e., Harris-Todaro hypothesis is clearly better than a cut in Lg. the competitive labor market hypothesis. For (The detailed derivations are presented this reason, we report here the policy evalua- in Annex 2-D.) tions obtained using both models. The idea is to determine how the results may vary under The main distinctive feature of using this alternative assumptions. model for analysis is that we explicitly ac- To evaluate the policies considered, we count for the interactions between the rural need information on basic parameters pertain- environmental distortion documented in the ing to the agricultural sector, as well as the section, Policy Issues in Ghana, and the con- private urban and government sectors. The ventional distortions, including trade protec- agricultural sector parameters are all obtained tion, labor market regulations, and oversized from our own econometric estimates reported government. The policy analysis focuses on in the section, Estimating Production Func- the consequences of reducing the conventional tions and Land Allocation Decisions. The distortions for national income, real wages nonagricultural parameters are mostly derived and the rural environmental stock. from recent World Bank publications. The We compare the effects of the policies exceptions are the elasticities of labor demand under each one of the maintained assumptions in the urban private sector, which are not regarding the labor market, i.e., the Harris- available for Ghana. We use instead elastici- Todaro and the competitive labor market ties reported for other African countries in a assumptions. Although the level of open study by Riveros (1992). unemployment recognized in government Table 2-4 presents the parameter values statistics is very low (2-3 percent), there are used to evaluate the expressions derived in indications that these statistics considerably Annex 2-D. Some limitations of these data are underestimate the actual level of unemploy- worthwhile to emphasize. The parameters for ment. Moreover, it appears that the level of the agricultural sector are reliable and consis- disguised unemployment is quite high in tent given that have been obtained from a Ghana. Also, the available sectoral wage data single source. However, the sample used to suggests that rural wages tend to be signifi- estimate these parameters is not necessarily cantly lower than urban wages in sectors that representative of the country as a whole be- use levels of skills that are comparable to cause, as discussed before, the entire database those in agriculture (i.e., wholesale, restau- corresponds to the Bron-Ahafo, Ashanti, and rants). Finally, there is some evidence that the Western regions of Ghana. These are impor- public sector still tends to lead the urban tant agricultural areas, but the degree of repre- private sector in wage adjustments. sentativeness of these three regions for the All these would be consistent with a Harris- country's agriculture is not clear. The data Todaro model where the urban wage rate required for the nonagricultural sector is not determination is significantly affected by too demanding, consisting mostly of shares of nonmarket forces and is for this reason higher the different sectoral outputs and sectoral than in the rural areas. However, the urban- employment in the economy. Data on labor 38 VOLUME II demand elasticities for the nonagricultural ing imports is negative. In the presence of sector are also needed. Unfortunately, there is unemployment (Harris-Todaro case), any no estimate of these elasticities for Ghana, and reform (such as decreasing of agricultural thus we have assumed "reasonable" values for implications) that increases output prices them. We performed sensitive analysis to see will contribute to decrease unemployment how the results are affected by changing these and thus to induce larger gains. On the values. contrary, policies that reduce output prices, The results of the policy simulations are such as import liberalization, cause higher presented in Table 2-5. The policy changes unemployment, which, in turn, reduces the simulated are not very large because the esti- income gains. The level of unemployment mated elasticities are valid only as local mea- is increased by price increases because sures. That is, the elasticities are valid only urban wages are set in nominal terms, and within the range of the exogenous variables thus any increase in output price implies a for which they were estimated. As the exoge- de facto reduction of the real wage distor- nous variables (i.e., policy variable) change, tion (i.e., by reducing real wages) and any the elasticities and other parameters used also output price reduction causes and increase change, thus increasing the error of the simu- in the real wage and thus a greater distor- lation. Thus, we postulate relatively gradual tion. This employment effect of trade re- policy adjustments of the order of 5 percent. form is strong enough to change the signs More important than the actual quantitative of the competitive labor market effects estimates reported in Table 2-5 is the qualita- (Table 2-5, Row 1). This explains the posi- tive pattern of response to policy changes. The tive income effect of reducing agricultural major patterns observed in Table 2-5 are the taxation and the negative effect of reducing following: tariff protection. By contrast, in the absence 1. The effects of the policy adjustment on of wage distortions (i.e., in the competitive national income are small in both the labor market case) decreasing agricultural Harris-Todaro and in the competitive mod- taxation reduces income because the wors- els (see Row 1). This is, of course, not ening of the environmental distortion that surprising. Most analyses typically find that such a policy induces more than offsets the the income or welfare effects of removing gains from eliminating the price distortion. even large distortions is very small, less Similar reasoning explains the negative than 1 percent of GDP. In the case of effects of import tariff reductions on na- Ghana, the remaining distortions are not tional income both with and without wage very large, and we simulate its partial, not distortions. total, removal. Moreover, accounting for 3. Both trade reforms have positive income the environmental distortion tends in gen- effects in the absence of wage and environ- eral to decrease the magnitude of the policy mental distortions (see numbers in brackets changes on national income. in Table 2-5). In the absence of environ- 2. Under the assumption of competitive labor mental distortions, but with wage distor- markets, the effects of trade liberalization tions, the agricultural tax reduction also has (i.e., of reducing taxation to agriculture and a positive effect on income, but decreasing decreasing tariff protection to the industrial tariff protection induces a negative effect. sector) on national income are negative The reasons for this are similar to those once the environmental distortion is ac- discussed in paragraph 2 above. counted for. In the Harris-Todaro specifica- 4. Somehow surprisingly, public sector em- tion, the effect of reducing agricultural ployment policies affect income in the taxation is positive while that of liberaliz- same direction whether wage distortions 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 39 exist or not. The effect, however, is more of devastating the remaining biomass re- strongly positive in the competitive case. sources of the rural areas. The fact that labor productivity is higher in the private than in the government sector is Conclusion responsible for the positive effect on na- tional income of reducing public employ- The major findings emerging from this ment. This effect is, however, much more paper are the following: remarkable in the case of competitive labor markets where the rest of the economy 1. Biomass is an important factor of produc- competes for labor directly with the gov- tion in Western Ghana. Its contribution to ernment. In the wage distortion case, the agricultural output ranges from 15 percent existence of unemployment considerably to 20 percent of the value of output. This dampens the effects of changes in govern- means that the contribution of the biomass ment employment on private sector em- factor of production to national GDP ranges ployment. In this case, the government between 7.5 percent and 10 percent. competes for labor with the private sector 2. The empirical evidence suggests an only in an indirect way. overexploitation of biomass through a more 5. As indicated before, the environmental than optimal level of land cultivated. Or, distortion is sufficiently strong to render the equivalently, fallow periods appear to be effect of trade liberalization perverse in the too short and thus the stock of the environ- competitive labor market case. mental resource is below the socially opti- 6. Perhaps the most important result of this mum levels. In this chapter, we have de- section is that, although the effects of the fined the social value of biomass quite reforms on national income vary in direc- narrowly, considering only its contribution tion and magnitude depending on the labor to agricultural output. We have excluded market assumption used, their effects on further roles of biomass that transcend the the rural environmental stock are consis- Ghanaian economy, such as its contribution tently negative regardless of the model to biodiversity and carbon sequestration. used. All four reforms cause a decrease in 3. The quantitative importance of agricultural biomass ranging from 3 percent for agricul- prices, rural wages, and population pressure tural price liberalization to 1 percent for as a source of biomass degradation has public sector wage reductions. The implica- been clearly demonstrated. tions of this are very serious, suggesting 4. The evaluation of a deepening of the struc- that even mild additional reforms are likely tural reforms in Ghana suggest that, in to cause a significant loss of forest and general, their impact on national income is biomass. This is important given that bio- rather small once the existence of the rural mass and other forest resources have value environmental distortion is considered. The beyond their contribution to national in- effects of furthering trade liberalization come. For example, biomass contributes to (decreasing implicit tax to agriculture or welfare beyond the Ghanaian economy reducing tariff protection) is in general such as biodiversity preservation and car- ambiguous. The possibility of adverse bon sequestration. The combined net effect effects is present for agricultural tax reduc- of a series of reforrns to further liberalize tion, depending on the characteristics of the goods and labor markets and to reduce the labor market. Import liberalization has a size of the government without tackling the negative effect on national income regard- environmental distortion may be mildly less of whether or not a wage distortion positive for national income, but at the cost exists. The effects of reducing the fiscal deficit through either lowering public em- 40 VOLUME II ployment or lowering public wages, how- is that the main source of supply response in ever, are unambiguously positive for na- agriculture is the expansion of the cultivated tional income regardless of the perfor- area rather than agricultural intensification. If mance of the labor market. agricultural price responsiveness relied less on 5. Without parallel efforts to address the land expansion and more on intensification, environmental distortion, the impact of a the policy evaluation of trade liberalization deepening of trade liberalization on bio- would probably be more favorable. Currently, mass depletion may be quite adverse. Fur- Ghana is gradually improving its agricultural ther losses of biomass of the order of 5-10 research and extension services which could percent by completing the trade liberaliza- in the future allow for greater reliance on yield tion process are not unlikely. Decreasing increases. On the other hand, it is plausible to public sector employment and public assume that as long as (forest) land is avail- wages, on the other hand, also causes a able for cultivation, a significant component negative impact on the biomass stock, but of the supply response is likely to continue to their quantitative impacts are more moder- be based on agricultural area expansion. To ate than those linked to trade liberalization. the extent that land is a normal factor of pro- duction (and there are no reasons to question The above implications are obtained using this), any increase in real output prices is parameters estimated by exploiting cross- likely to induce a greater demand for land. sectional and time variability during the This, in turn, will be manifested in an expan- 1988-89 years in western Ghana. A key em- sion of the agricultural frontier as long as pirical result underlying the policy simulations more land is available. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 41 Table 2-1: Production Function, Ghana, 1988-89 (3) (1l ) (2) Instrumental variable and OLS Instrumental variable village-specific dummies 9.44 9.08 9.31 Constant (13.30) (13.6) (13.6) 0.27 0.22 0.27 Land cultivated (3.16) (2.65) (3.12) 0.25 0.30 0.25 Total labor (2.47) (2.89) (2.51) 0.15 0.20 0.17 Village biomass (1.86) (2.56) (2.22) 0.28 0.29 0.26 Tools (3.13) (3.20) (2.99) Dummy cocoa pro- 0.62 0.58 0.60 ducers (4.37) (4.13) (4.31) -0.57 -0.30 -0.61 Dummy 89 (-3.14) (-1.71) (-3.30) -0.25 -0.40 -0.51 Dummy size A' (-1.23) (-2.02) (-1.66) -0.25 -0.46 -0.86 (-1.32) Dummy size B2 (-1.50) (-2.79) Village dummnies (several) no no yes N 139 139 139 R 2 0.46 0.40 0.46 R 2 0.43 0.36 0.42 F 9.60 10.70 10.20 1. Dummy size A is equal to one for extremely small farm size (less than 5 acres) and zero otherwise. 2. Dummy size B is equal to one for extremely large farm size (greater than 100 acres) and zero otherwise. Note: a. t-statistics are in parentheses. b. Double-log specifications were used. c. The dependent variable is the log of real agricultural output per household. 42 VOLUME II Table 2-2: Land Equation, Western Ghana, 1988-89 (double-log specification) (1) (2) 7.10 8.85 Constant (2.66) (4.0) 0.37 Family members (2.33) -1.01 -1.26 Wage/agricultural output price (-2.19) (-3.13) 0.27 0.42 Tools (1.94) (3.32) Percentage of fallow land over -0.19 total land (-0.91) 0.89 0.86 Dummy for cocoa producer (4.37) (4.34) -0.57 -0.69 Dummy for Ashtanti-Bron (-2.44) (-2.97) 1.49 1.44 Dummy 89 (5.77) (6.57) N 139 139 r>2 0.54 0.51 Rm2 0.51 0.49 F 21.67 27.94 - = Endogenous variables. Note: 1. t-statistics are in parentheses. 2. The dependent variable is the log of the area cultivated by each household. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 43 Table 2-3: Net Effect of Increasing Cultivated Land by 10 Percent (percent) Measures based on OLS Measures based on instrumental estimates' variables/village dummy estimates2 On output Level 0.5 0.2 On net farm income3 4 -0.4 -0.65 1. Obtained using estimates in Table 2-1, Column 1. 2. Obtained from estimates in Table 2-1, Column 3. 3. Net farm rzvenue is defined as value of output net of land clearing costs. It was assumed that the share of land clearing costs in net revenues was 8 percent (this is the absolute lowest boundary according to field estimates in western Ghana). 4. A 10 percent discount rate is assumed. Table 2-3A: Extent of Overcultivation and Income Losses (percent) Based on Based on instrumental variable/village OLS estimates dummy estimates Excess land cultivated 26 45 Income loss 0.96 2.7 Table 2-4: Basic Parameters and Elasticities Used in the Simulations Agricultural sector Rest of the economy SL = 0.25 alnx/alnp = 1.26 aInLN/alnq = 0.40 B = 0.66 S, = 0.27 alnx/alnw = -1.26 aInLN/alnw = -0.40 YA/Y* = 0.48 So = 0.17 alnL/alnp= 1.36 LN/L = 0.33 t = 0.17 SC = 0.08 alnLiaInw = -1.36 e = 1.14 SLN = 0.30 T = 0.05 dlnLJdlnO = -1.45 LNILe = 1.33 Y;VY* = 0.25 Note: All parameters for the agriculture sector, with the exception of r, are derived from the estimated production function and land cultivated equation. Parameters for the rest of the economy are obtained or derived from various World Bank sources. The implicit ad valorem net tax rate for agriculture (r) was estimated as a weighted average rate. Table 2-5: Effects of Various Policies (percent) Net reduction of agricultural Decrease tariff protection to Decrease government implicit tax' industry2 employment` Decrease public sector wages4 Harris-Todaro Competitive Harris-Todaro Competitive Harris-Todaro Competitive Harris-Todaro Competitive Model model Model model Model model Model model National 0.21 -0.02 -0.17 -0.02 0.07 0.20 0.22 Income (0.27) (0.91) (-0.13) (0.91) (0.10) (0.03) (0.24) Agricultural Wages 3.36 4.55 -0.99 -1.341 -0.71 -0.83 -0.56 Environmental Resource Level -3.03 -1.67 -1.83 -2.45 -1.32 -1.52 -1.05 - = not applicable. 1. A 5 percent increase in agricultural prices due to the elimination of implicit tax to agriculture. 2. A 5 percent decrease in nonagricultural sector prices due to halving the level of tariff protection. 3. Five percent reduction in public employment. 4. Five percent cut in public wages. Note: Numbers in parentheses correspond to the effect of the policies on national income when the environmental distortion is ignored. 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 45 Annex 2-A: Derivation of a Land Demand Equation If farmers maximize profits, the level plies full account of the externality and, thus, of cultivated land and the amount of community efficiency. Since Fe is mostly an labor would be chosen so as to satisfy external effect, the value of A will depend on (ignoring by the time the discount rate) the community controls over individual land the following: cultivation decisions. Note that since FL(L,X,z,O)=w/P (a) < 0, a higher A implies that farmers cultivate less land. Equations (a) and (b) solve simultaneously Fx(L, x, z, 0) + A. Fe =c/p (b) for L and x, where L = L z,,z) and ax p p where L is labor, x is area cultivated, H is x = x( p , z). Note that land clearing village biomass, w is the wage rate, p is agri- costs are mostly labor and hence, c can be cultural output price, c is the cost per acre of written as a function of the wage rate. If y clearing land, and z is other input endow- units of labor are needed to clear an acre of ments. Subscripts in the F( ) function reflect 1 a n d, t h e n c = y w a n d marginal products of the respective input. The w w parameterO < Ilisanefficiencyparameter L = L(-; z), x = x(-; z). Table 2-2 that reflects the extent to which the effect of p p reports the estimates of x based on this specification. land cultivation decisions on blomass are taken into consideration. A value A = 1 im- 46 VOLUME II Annex 2-B: Derivation of a Labor Demand Specification The first order condition for labor aInL 1 - SX alnxlalnw - se ain oIalnw allocation is -S aInw SL1 pFL (L, x (p,W), 0, z) = W, The estimates of the production function where x(p,w) is evaluated at the estimated directly yield the values for s, s, SL. More- ' . . . . . ~~~~over, from the estimate of the land equation level. Totally differentiating this with respect w e evtiuate ofnthenland equation to w, explicitly using the fact that x depends we canuals measure on w as well, we obtain aIn 0/l In w = dlnO 8lnx aL 1 - pFL, axlaw - pFLe a0/aw dlnx alnw 8w pFLL where Fij indicates second order derivative with respect to arguments i and j. The above equation can be represented in log form, 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 47 Annex 2-C: Data Sources and Definition of Variables The data used come from the Living consumption, and home production. Standards Survey (LSS) conducted in Ghana in 1988 and 1989 (World Bank 1993). These Wage/Agricultural Output Price panel data include a wide range of information Average wage per village earned per day such as labor force activity, individual charac- by a male working in the fields, divided by teristics, consumption, and production at the the average price of agricultural output at household level, as well as wages at the vil- the village level. lage level. The sample used consists of all observations repeated in 1988-89 for which Family Members information on agricultural revenue, land Number of family members older than ten cultivated, hours worked by family members years and younger than sixty years of age. and hired labor exists. Statistics on fallow, forest, and agricul- Total Labor tural land and biomass density for each village Total number of days worked in the field were provided by a special study done for this by family members or hired labor. project by EARTHSAT, based on satellite images that cover the western region of Ghana Land Cultivated per Household for the years 1988 and 1989. The total number Land cultivated by each farmer (in acres). of villages considered was sixteen-all in the western region of Ghana. The villages are Tools New Bansakrom, Yankye, Akantamwa, Total number of tools owned by each Tanoso, Sireso, Bibiani, Asunsu, Doduoso, farmer. Nsuatre, Susuanso, Dormaa, Apronsie, Kenyasi, Ewisa, Boasi, and Wenchi. Percentage of Fallow over Total Land The Living Standards Survey (LSS) and LFALLOW/(LFALLOW + LAGRIC), the EARTHSAT data were matched to obtain where LFALLOW is total land under a unique data set that includes the usual indi- fallow in each village and LAGRIC is vidual characteristic variables with informa- total land cultivated in each village. tion on natural resources at the village level. Source. EARTHSAT. The total number of households considered is 139. Village Biomass LFALLOW*iFALLOWI(LFALLOW + Variables LAGRIC), where WFALLOW is the aver- age biomass density in each village. Real Agricultural Output per Household age AR SiTy g Total Output in Cedis 1988. It is the sum of total sales to the market, payments in kind to hired labor, seeds kept, home 48 VOLUME II Annex 2-D: Effects of Policy Adjustment oni National Income Policy simulations in a Harris-Todaro economy dlnY aInL x I +r ~~- L_~+ fs -sC+ Reducing taxation to agriculture dlnp LInp - x r +xlx As indicated in the main text, lowering s dln0 amnx [ alnL is equivalent to increasing p, keeping p* con- dlnx) alnp lnw stant. Thus, the effect of such policy on Y can be estimated by differentiating totally Y* in + x 1 +r alnO Equation (7) with respect top, (maintainingp' +p X c r+x/ s alnx constant): alnx dlnw YA adln w dlnp Y * (13) dY P F aL(p,w) +p *(F2 -yF1 + lF3 dp Iap I+rA r x(p,w) + *F aL(p,w) n+xlx) ap aw where = F1L aInF alnF n+ where ~~~~~SL = ' sx = a L F aInL I alnx +p (F, - yF. +rTF, lx a inF , 2Y,+P3n + xht.>) se, _ a ln , and YA = p*F(-) is agricultural ax(p, w) dw output valued at world prices. From the mi- aw dp (12) cro analysis of the previous section, we ob- tain estimates for sL, sx, SO, , Thus, the effect of p on Y* can be decomposed alnx aInp' into three effects: (a) an increase in p causes alnx Also the effects of p and w on em- greater use of farm labor, i.e., a rise in L alnw which, in turn, increases agricultural output; (b) a higher p also leads to an expansion in ployment are implicit in the production cultivated areas, which will affect agricultural function. The effect of agricultural output production negatively if an environmental prices on the wage rate w, however, has to distortion exists (i.e., if the term in parenthe- be obtained by solving the general equilib- ses is negative); (c) the rise in p also induces rium model. Logarithmically differentiating higher wages which, in turn, affect both the Equation (10) we obtain level of agricultural employment and culti- vated land. The first right-hand term in Equa- dn w alnL(p,w)/alnp(14) tion (12) corresponds to effect (a), the second dlnp B - 3lnL/Olnw one to effect, (b) and the third one to effect (c). where - L-L is the ratio of urban labor Expressing Equation (12) in percentage or L logarithmic form yields force relative to rural employment. The effect of p on w is positive, given that L is increasing in p and decreasing in w. Note, however, that this effect is always less than one. Thus, the microestimates of the previous 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 49 section, Equation (14), and information on the employment is entirely attained by reducing share of agricultural income in national in- the level of unemployment or underemploy- come evaluated at world prices (YI/Y*), are all ment in the urban centers. Thus, what reduc- the information needed to evaluate the effect ing agricultural taxation does is to induce a of decreasing agricultural taxation on national better utilization of the pool of labor in the income. Similarly, one can obtain an expres- economy and, at the same time, a more inten- sion to evaluate the general equilibrium effect sive utilization of land. The tradeoff, however, of raising agricultural prices on the environ- is that it also causes a reduction in the envi- mental resource. From Equation (1'), it is ronmental resource, which was already at clear that 0 = O[x(p,w),x]. Thus, logarithmic levels below the social optimum. Which one differentiation of 0 with respect to the price of of the two forces dominate is an empirical the agricultural goods yields matter depending on the actual values of the various parameters that conform Equations (12), (14), and (15). dln 0= dlnO alnx(p,w) dlnp dlnx alnp Trade liberalization via reduced protection to urban activities alnx(p,w) dlnw + lnw dln (15) Decreasing trade protection to urban -ln w dlnp .1 activities implies a reduction in tariffs t, or equivalently, a reduction in q for a given level d-nO of q*. Logarithmic differentiation of Y1 as where dln0 is defined in Equation (4). The defined by Equation (7) with respect to q dlnx (keeping q* constant) yields first term in square brackets corresponds to the partial equilibrium effect, while the second reflects the indirect wage effect. Given that dlnY* _ A J alnL dlnw dlno), (1 + r)xlx s dlnO) alnx dlnp r + x/x ° dlnx) alnw while the indirect wage effect is positive, >0. ____~~~~~~~ alnLN given that aInx <0 and dlnw Since dlnw LNw (I -r (IN (16) dlnw dlnp dlnq L l+t Olnq J d ln w < , however, the direct negative effect dlnp must dominate the indirect effect. Thus, Equa- where e -L is the inverse rate of urban tions (12), (14), and (15) provide the basis for LN +Lg the evaluation of the effects of reducing taxa- employment. We note that in deriving Equa- tion to agriculture on national income, agricul- tion (16), we have used Equations (8(i)), (9), tural wages and the environmental resource and (10). Combining these three equations, we stock. get G(-) =e ) I - F Note also that we Note that, since in the context of a Harris- 1 + t Todaro model, the agricultural sector does not have normalized the world prices to one, i.e., directly compete with the urban sector, there p+ = q* = 1, which implies no loss of general- is no real reduction in output in urban activi- ity, given that these prices are exogenous (in ties. Indeed, the expansion of agricultural fact, we can always pick the units of measure- 50 VOLUME II ment of the agricultural and nonagricultural goods so that each unit is worth US$ 1.00). The effect of changes in the protection dnO _ dInO [lnx(p,w) alnw I level of the nonagricultural goods can be dlnq dlnx alnw alnq (18) decomposed into three effects: (a) an increase in q will reduce employment in agriculture An increase in q will induce higher agricul- because of its wage effect thus negatively tural wages which, in turn, reduce cultivated affecting agricultural production and national land and thus generate a positive effect on the income (first right-hand term in Equation level of the environmental resource. Thus, (16)); (b) the increase in agricultural wages Equations (16) through (18) allow us to evalu- induced by increasing q also leads to a fall in ate the effect of import liberalization (i.e., of cultivated land which, given an environmental reducing q) on national income, agricultural distortion, will have a positive effect on agri- wages, and the environmental resource, given cultural output and national income (second information on the relevant parameters and right-hand term in Equation (16)); and (c) the elasticities. increase in q causes an expansion in employ- ment in the protected sector (third right-hand Wage and public employment policies term in Equation (16)). This effect is also A decrease in government intervention in positive to the extent that it decreases unem- urban wage settlements (i.e., more lax regula- ployment. However, its magnitude will cru- tions on wage indexation, reducing public cially depend on the degree of relative protec- wages, discarding minimum wages) can be tion of the nonagricultural sector ( 1 represented by a reduction in W. A change in W 1 +t) on Y* can be analyzed directly from Equation The larger is the tariff protection level t and (7) by differentiation: the agricultural tax a, the smaller will be effect where S1N is the share of labor costs in the (c) in this paragraph. Indeed, this reflects a private urban sector. An increase in W will tradeoff between decreasing unemployment, __a n_p_w dlw which is positive for national income and for dlnY L alnL(p,w) dlnw increasing employment and output in the dlnw L Olnw dn) protected sector, which is detrimental for (I +r) x/x dlnO national income. + s -s + xse d In order to evaluate Equation (16) we need x r+x/x dlnx to determine the effect of changes in the price alnx(p, w) dlnw * of the nonagricultural goods on the agricul- alnw dln/v YA tural wage rate, dlnw From Equation (10) din q aInLN we get that S N N (19) dlnw alnLNIalnq (17 change w (not necessarily in the same direc- dlnq aI - OlnL/lnw tion, as we shall see) which, in turn, will affect agricultural employment (first right-hand side Of course, if labor in the urban sector is a term in Equation (19)) and land cultivated normal input (67nL,,lnq > 0) we have that (second right-hand side term in Equation Equation (17) is necessarily positive. (19)). Also, the higher W reduces employment Finally, the effect of q on the environmen- in the private urban sector, which is detrimen- tal stock comes entirely through its effect on tal for national income (third right-hand term cultivated land: in Equation (19)). 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 51 The effect of the regulated urban wage on lar way. Actually, the effect of Lg on agricul- the agricultural wage can be derived directly ture takes place eneirely via the level of the from Equation (10): rural wage: dlnw 17LN 1 +alnN/alnv ( _____ dlnwA dlnw wLd-nY* aln(pw) dlnw din iv wL 0-alnLIaInw (20)dlfnLg a ( lnw dlnLg) Clearly the sign of Equation (20) will be + (Is_sc+ (r)xlxsOdlnO determined by the elasticity of labor demand xc r + xlx dlnx in the urban sector. If demand for labor in the nonagricultural private activity is inelastic alnx(p, w) dlnw YA (i.e., alnN/alnW is less than unit elastic), the alnw dlnLg y agricultural wage will increase; but, if it is elastic, it would cause a fall in the agricultural + aL /Y' (22) wage. The intuition behind this is clear: An 8 increase in W reduces employment in the A change in Lg does not have any direct effect urban sector which, in turn, reduces the rate of on the urban sector to the extent that there is urban employment rate, or equivalently, re- urban unemployment. It does affect the agri- duces the probability of finding employment cultural wage, however, by changing the rate in the urban sector. The lower probability of of urban employment. The effect of this on finding employment has a negative effect on agriculture is captured by the first and second rural wages because fewer farm workers right-hand term in Equation (22). The third would be willing to migrate. But the increase right-hand term reflects the fact that workers in W has the opposite effect. Which effect in the public sector are productive (possibly domninates depends on the extent by which the with a very low productivity), which has an rate of urban employment falls. If the demand effect on national income. Thus, in the context for labor in urban activities is more than unit of a fixed urban wage, the tradeoffs of a gov- elastic, the probability of finding an urban job emient employment policy occurs between will dominate and vice versa if the demand for the public sector and the agricultural sector labor is inelastic. rather than the urban activities. Similarly, the effect of W on the environ- From Equation (10) it can be shown that mental resource can be represented through the effect of Lg on rural wages is the effect on cultivated land: dlnw wLg 1 dlnO = dinO ralnx(p,w) dlnw 1 2- dnLg _W anL (23) dlnfv dlnx alnw dlnw' 0 dnw which is positive, as could be expected. Fi- Thus, if an increase in w increases, the agri- nally, the effect of Lg on the environmental cultural wage land cultivated will decline, and factor is the environmental resource will improve. In dInO = dlnE alnx(p,w) dlnw general, however, the effect of regulated urban dlnL dlnx alnw dlnL (24) wages on the rural environmental stock would g g be ambiguous. The sign of Equation (24) is necessarily posi- The implications of changes in public tive given that the agricultural wage rate sector employment can be analyzed in a simi- increases. Thus, public sector employment 52 VOLUME 11 expansion is beneficial for the rural environ- is identical, but of opposite sign to the effect ment if urban wages are fixed. of p, i.e., dln )/dlnq = - dln /dlnp. Now there are only two distortions, namely, the trade Environmental distortions in a tight and distortion associated with the tariffs and competitive labor market export tax and the environmental distortion. ..dfntooftenational income Y Thus, the effect of changing p or q can also be ThEq definition ofa the profitmaximdecomposed into two effects: (a) The changes cond(tions (7)an the rofit mang in p or q induce a reallocation of labor be- conditions Equati (8xrmazing uonc ane tween the two sectors (first right-hand side while the profit maximizing condition Equa- terms in Equation (25)). This change causes tion (9) for the urban sector changes in a incme incEaseowhen pTis raise cause trivial manner, income to increase when p is raised because the agricultural sector is initially discriminated against due to both the agricultural tax and the (1 + Gt) q G,(-) = w (9') tariff protection to the nonagricultural sector (that is, agricultural production is below the where the only change (with respect to Equa- scal pia ee n h oarclua tio .9) a ent s nta fW o socially optimal level and the nonagricultural tion (9')) has been to use w instead of w . Now sector is producing above such level). When q wages faced by both sectors are equalized. increas the effec suunabiguusl nega- Since competitive labor markets imply full iveases, the effect iS unambiguously nega- employment, Equation (10) is now replaced by The endirightan effect is captured byth lbo mrkt lerig oniton by the second right-hand side in Equation by the labor market clearing condition: (25). This effect is due to the fact that changes in p or q affect the level of cultivated area. L = Lg+LN(( + t)q ,w) To be able to evaluate Equation (25), we + L(-p*wneed to provide an expression for the wage +L((1 - r)p *, W) (10') effects. From the labor market clearing condi- tion it follows that Differentiating Equation (7) using Equa- tions (8), (9'), and (10'), we can now obtain dlnw aInL(p,w)/1lnp expressions for the effect of changes in agri- = -- (26)(i) cultural taxes on national income: dlnp LN) ainn L nL.w) dlnY* YA J + L alnw 3lnw = E sL dlnp y *(I-T)(1+0 dlnw alnLN(q,w)/alnq alnL(p,w) alnL(p,w) dlnq LN 3lnLN(q,w) alnL(p,w) alnp alnw L ainw 3mw dlnw +(s s + (1 +r)x/x dlnp) r + xC x (26)(ii) s dlnO (lnx(p,w) Under the assumption that labor is a nor- 8 dlnx) Inp mal input in both the rural and urban sectors we have that both an increase in p and q in- + 3lnx(p,w) dlnw' | (25) duce higher wages. alnw dlnp) The effects of government employment It can be shown that since wages are flexible, policies can be analyzed in a similar way, by the Lemer's symmetry condition applies. This differentiating Equation (7): implies that the effect of q on national income 2. Evaluating Economywide Policies with Agricultural Environmental Externalities: Ghana 53 tion of cultivated area which has a positive (27) rural environmental effect. Also, government din Y YA t + Tr a in L (p, w) services increase. Given that the marginal dInY =_A_t SL contribution of public employment to national dlnLg Y- (1 -t)( +t) alnw income is possible very low, the net effect of dlnw / (1 + r)x/x increasing L, is likely to be negative. dl + SI -S+ / Also from Equation (10'), it is clear that dInLg dlnO 6 aInx(p,w) dlnw d dlnx alnw dlnLg dlnw 1 gJ dInL, LN) N+ L )aInL (8 (-) _____ 3h- L (28) aLg YN Lg alnw Lg alnw +-_ - S* LN Y Y* *which is positive given that demand for labor The increase in government employment in both sectors is downward sloping. Finally, induces higher wages which, in turn, cause a the effect of Lg on the environmental resource fall in employment in both the urban private is identicail to the Harris-Todaro case, except sector and the agricultural sector. At the same that now we use Equation (28) instead of time, the increased wage rate leads to a reduc- Equation (23) to evaluate dlnw/dlnLg. 54 VOLUME 11 Bibliography African Development Bank (ADB), Economic Glantz, M. 1977. Desertiflcation: Environ- Commission for Africa (ECA), and Orga- mental Degradation in and around Arid nization of African Unity (OAU). 1984. Lands. Boulder, Colo.: Westview Press. "The African Food Crisis and the Role of Larson, B., and D. Bromley. 1990. "Property African Development Bank in Tackling Rights, Externalities, and Resource Degra- the Problem." A paper presented jointly by dation." Journal of Development Econom- the staff of the ADB, ECA, and OAU to ics 33:235-62. the Symposium on the future of ADB and L6pez, R. 1993. "Resource Degradation, the Food Crisis of Africa. Tunis, Tunisia, Community Controls and Agricultural May. Productivity in Tropical Areas." Univer- Allen, J. 1985. "Wood Energy and Preserva- sity of Maryland, College Park, unpub- tion of Woodlands in Semi- and Develop- lished. ing Countries: The Case of Dodoma Re- L6pez, R., and M. Niklitschek. 1991. "Dual gion, Tanzania." Journal of Development Economic Growth in Poor Tropical Ar- Economics 19:59-84. eas." Journal of Development Economics Dasgupta, P., and K.-G. Maler. 1990. "The 36:189-211. Environment and Emerging Developing Perrings, C. 1989. "An Optimal Path to Ex- Issues." World Bank Annual Conference tinction? Poverty and Resource Degrada- on Development, Washington, D.C. tion in the Open Agrarian Economy." Deacon, R. 1992. "Controlling Tropical De- Journal of Development Economics forestation: An Analysis of Alternative 30:1-24. Policies." Working Paper No. 1029. Riveros, Luis A. 1992. "Labor Costs and Country Economic Department, World Manufactured Exports in Developing Bank, Washington, D.C. Countries: An Econometric Analysis." Feder, G., T. Onchan, Y Chalamwong, and C. World Development (U.K.) 20:991-1008. Hongladarom. 1988. Land Policies and Sinn, H. 1988. "The Sahel Problem." Farm Productivity in Thailand. World KIKLOS, May, 41. Bank, Johns Hopkins University Press, Todaro, M. 1989. Economic Development in Baltimore, Md. the Third World, 4th edition. New Food and Agriculture Organization (FAO). York: Longman. 1986. "Natural Resources and Human World Bank. 1993. "Ghana Living Standards Environment for Food and Agriculture in Survey (GLSS), 1987-88 and 1988-89: Africa." Environment and Energy Paper Basic Information." World Bank, Poverty No. 6. Rome. and Human Resources Division, Washing- ton, D.C. Economic Policies for Sustainable Resource Use in Morocco Ian Goldin and David Roland-Host WATER IS A DISTINCTIVE RESOURCE because it constitutes a direct input to almost every economic activity, drawing a continuous thread from ordinary use in agriculture, through manufacturing, and into the myriad of service sectors. Despite a long economic history, water allocation stands out as one of the most significant cases of market failure in both developing and developed economies. The institutional arrangements that have governed water allocation, from prehistory to the present day, have fostered serious resource misallocation, technological choice that is neither statically nor dynamically efficient, and an array of negative economic and environmental externalities that propagate through downstream linkages to the rest of the economy. This chapter is based on a paper presented in a conference on Sustainable Economic Development: Domestic and International Policy, convened by the Centre for Economic Policy Research and the Organisation for Economic Co-operation and Development (OECD) Development Center, May 24-25, 1993, Paris. 55 56 VOLUME II Historically, the relative abundance and conclusion that increased marketability of regenerative nature of water resources have water can realize substantial static and dy- fostered inefficient water use.' In modern namic efficiency gains in agriculture, includ- times, the clear delineation of regional bound- ing greater conservation. The development of aries limiting migration, population growth, an appropriate incentive framework, which and rising living standards have combined to includes the implementation of water tariffs intensify water use in agriculture and else- that more closely approach marginal costs, is where, and the sustainable practices for utiliz- expected to contribute to demand management ing this essential resource are receiving greater and conservation in irrigation where subsidies scrutiny. It is only a matter of time until the are greatest. The adjustment of relative prices reforms in water allocation, already initiated in favor of sustainable resource use will be in a number of countries, are more widely greatly facilitated by trade reforms. These applied. The essential issue is how to devise currently provide protection to agriculture; in reforms that have the efficiency and incentive the absence of trade reform, higher water properties to be economically and politically charges would be passed through to output sustainable. To answer this question, which is paces, seriously limiting efficiency gains. a pressing concern for many developing coun- In order to focus discussion of the general tries, we consider the case of Morocco. Our economic principles of sustainable resource choice of Morocco has been facilitated by the use, we have developed a computable general fact that it possesses relatively sophisticated equilibrium (CGE) model of Morocco. This data resources and shares many common captures sufficient detail in agriculture to characteristics with water-scarce economies. evaluate alternative water allocation schemes. We thus anticipate results from this case to The model provides a convenient vehicle for have wider applicability. identifying the components of domestic eco- In this chapter, we draw on Morocco to nomic adjustment in response to various illustrate the linkages between trade and resource management policies, and allows us macroeconomic policies and sustainable to assess the implications for Morocco's trade resource use.2 To this end, we evaluate the position. Conversely, the CGE model is also current state of water distribution in Morocco, used to evaluate the implications of alternative distinguishing between relatively arid and trade policies for agriculture in general and moist farming regions and between rural and water use in particular. Our results indicate urban households. The available evidence that the static benefits of more efficient water suggests that the existing allocation system allocation would be considerably magnified if fosters suboptimal and unsustainable patterns - the country pursued more outward-oriented of water use and that inefficiencies arising trade practices. from inappropriate pricing at the micro level The next section presents an overview of have been compounded by distortions in Moroccan water use and policy. The section, relative prices resulting from macroeconomic Methodological Issues, contains a general and trade policies. Our analysis supports the conceptual discussion of efficient and sustain- able water allocation. The section, The Moroc- can CGE Model, presents the Moroccan CGE 'Relative abundance is of course a de facto result of model and the data that were used to calibrate human settlement where traditional farming practices it. This is followed in the section, Simulation were sustainable. Results, with a series of simulation results 2The analysis of Morocco presented in this chapter is illustrative and preliminary and is an initial stage of a major research project which includes construction between water policies and trade policies. The and estimation of a detailed database on the country's last section is devoted to concluding remarks. economy and water use. 3. Economic Policies for Sustainable Resource Use in Morocco 57 Practical Issues lead to overall water shortages in the medium in Moroccan Water Allocation term. In addition, serious water management problems are already evident in some parts of Owing to a limited resource base and recur- the country. These include a severe deteriora- rent droughts, water is a scarce commodity in tion in water quality, depletion of nonrenew- many parts of Morocco. The main sources of able groundwater resources, the increasing fresh water are found in the mountains, at exploitation costs for marginal water sources, considerable distances from urban and irri- and the failure of the existing system to supply gated areas. This geographic situation necessi- more than 10 percent of the rural population tates storage dams and long transmission with potable water. Despite the fact that nearly facilities, with high average cost and marginal all urban dwellers have safe water supplies, cost rising steeply as increasingly sparse water half the country's population is rural. Thus, resources are developed. The country's popu- over 45 percent of the population lack secure lation is growing at more than 2.6 percent water supplies. annually and is expected to exceed 43 million By the year 2020, assuming current trends by the year 2020. Rapid increases in water continue, water will still be in surplus in the demand by agriculture, industry, and urban Loukkos, Sebou, and Mediterranean basins in households are anticipated by the government. the north, but all other river basins will have Urban migration and rising incomes are ex- negative balances. Moreover, without signifi- pected to lead to a 4.2 percent annual growth cant water demand management, the overall in urban household demand, and rapid eco- water balance will become negative (in excess nomic growth is predicated on an anticipated of 200 million cubic meters annually before 5 percent annual increase in industrial de- 2020), which means that even nationwide mand. Meanwhile, the National Irrigation water transfers (assuming they are economi- Program aims to extend the area under irriga- cally feasible) could not meet regional water tion by 30 percent in the next decade. Irriga- deficits. The case for timely and judicious tion currently accounts for 92 percent of the government intervention is therefore compel- demand for water, urban households 5 per- ling. On the demand side, this might include cent, and industry only 3 percent. policies to promote conservation and im- The availability of adequate water resources proved demand management in irrigation and is a prerequisite for meeting the country's among urban and industrial consumers. On the economic targets. Water resources are already supply side, greater attention needs to be paid becoming scarce in certain areas, and initial to water quality and development of additional projections for the year 2020, based on ex- water resources. The latter possibilities in- pected economic growth, current relative clude improved secondary water capture, prices, and related water demand, clearly point treatment, and recycling, decentralization of to the emergence of an economywide water brackish and sea water, and sustainable deficit. While the water balance today is groundwater mining. largely positive in most of the north (Loukkos, Appropriate pricing and incentive policies Mediterranean, Sebou, and Oum Er R'bia are the keys to resolving Morocco's water basins), it is generally negative in the south dilemma. A continuation of current practices (Souss, Zia, Rheris, and Guir basins). Al- implies not only growing distortions in the though there is currently an annual overall allocation of an increasingly scarce resource, water surplus of 1.1 billion cubic meters but also an increasingly unsustainable fiscal (amounting to some 15 percent of all renew- burden. Public investments in the water sector able water resources), the anticipated annual already account for more than 25 percent of growth in demand of 4 percent is expected to the government's investment budget and, by 58 VOLUME II 2020, are expected to account for up to 60 prices are being affected by macroeconomic percent, reflecting the escalating costs associ- reforms and trade liberalization. This calls for ated with interbasin transfers and exploitation the analysis of water pricing issues in a frame- of marginal resources.3 Meanwhile, recurrent work which simultaneously includes possible costs are mounting with the aging of the changes in output and other relative prices. existing water supply, irrigation, and sanita- Significant structural reform is currently tion systems. Irrigation water charges cur- under way in Morocco. This is expected to rently cover less than 10 percent of the long- lead to the gradual reduction in import protec- run marginal cost of delivering water to agri- tion, which currently maintains the prices of culture, while urban water tariffs contribute sugar, cereals, oilseeds, meat, and dairy prod- less than half the cost of urban and industrial ucts well above their opportunity cost on supplies.4 world markets. Gradual and broad-based Subsidies have had a particularly agricultural trade reform is envisaged in the distortionary effect in the agricultural sector. context of structural adjustment agreements Low water charges (coupled with low effec- with the World Bank. These reforms would be tive collection rates on these charges) have accelerated following a General Agreement on artificially promoted production of water- Tariffs and Trade (GATT) agreement. The intensive crops. Sugar cane (which uses ap- simultaneous introduction of trade and water proximately 10,000 cubic meters per hectare) pricing reforms would imply increased input and lucerne (15,000 cubic meters per hectare) prices and a decline in output prices. The have particularly low value added per unit of impact on the highly subsidized sugar cane, water use. The existing incentive framework cereals, cotton, and oilseeds subsectors could does not encourage production of crops with be particularly serious as factor price increases high returns on water use. These include are passed through to agricultural commodi- wheat (4,000 cubic meters per hectare) and ties. The government budget is affected negli- vegetables (3-4,000 cubic meters per gibly. Thus, the static effects of the policy hectare).' appear to be detrimental, but a silver lining The consequences of higher water charges appears in the results for water use as the on patterns of production and the rural-urban economy moves decisively toward greater terms of trade are examined in this chapter. sustainability. In the rural sectors, water use Our primary focus is on the general equilib- falls by about one third, and even urban water rium effects of water policy on relative prices use is also driven down slightly by declining and resource allocation between agricultural aggregate demand. Trade liberalization has and nonagricultural activities. In Morocco, more salutary effects on the static efficiency of these sectoral effects will be intertwined with resource use. those of macroeconomic adjustment policies that are being pursued concurrently. In partic- Methodological Issues ular, water prices are expected to be increased at a time when output prices and other input In this section, we discuss the Moroccan water distribution question from three per- spectives.6 The first of these concerns static 3Authors' estimates based on current trends. efficiency of water allocation in Moroccan 4Authors' estimates based on average current water charges. 5These estimates of water use are derived from field 'As was emphasized in the previous section, trade estimates in Morocco. They are simply indicative reform also plays an important role in our empirical averages. In practice, water use varies from area to area policy analysis. The theory and methodology on this and intensity of production, and inter alia, depends on topic are well known, however, and we omit reviewing the rainfall, soils, and time of planting. them in this section. 3. Economic Policies for Sustainable Resource Use in Morocco 59 agriculture. We argue that a publicly managed arguments for improving static efficiency rest system of water distribution, if it is coupled on this basic productivity disparity; what with economic incentives, could significantly remains is the incentive properties of different improve the efficiency of water use in the allocation schemes. country, increasing real output and incomes in Assume for convenience only that water the process. Second, we discuss how reform- can be transferred without conveyance losses ing agricultural water distribution can improve or quality degradation. If water is simply the intertemporal efficiency or sustainability appropriated and redistributed, M and A are of water use and expand the growth prospects still unlikely to use the resource efficiently for Moroccan agriculture. Third and finally, even though aggregate output rises. If, on the we discuss the important issue of how to other hand, water were bought from M and allocate the country's water resources between sold to A, any price would begin to exert rural and urban areas. Guaranteeing equitable rational efficiency considerations on their quantities and qualities of water for household production decisions. Water now has a real and economnic use in both parts of the country opportunity cost for M, who might change the will inevitably determine the political mix of other inputs or even invest to innovate sustainability of Moroccan water policy. and reduce water use with minimal effects on marketable output. By the same token, a price Static efficiency for water would also allow A to expand factor Historical water rights conventions repre- use in the appropriate mix to most profitably sent one of the most serious market failures in use the newly available water. traditional agriculture. As Easter (1986) and Clearly, water reallocation can increase others have observed, the water rights mecha- output, and market pricing of water can im- nism results in a queuing allocation, where prove efficiency. What then is the role of the those with the original endowment use the public sector, apart perhaps from providing resource at zero marginal cost and pass the education and initial financing. The failure of remainder on to successive rights holders at water markets is not simply due to ignorance, zero marginal revenue. Thsie rgsersof a but results from significant problems of en- market for water thus removes any incentive forcement. Any distribution scheme of reason- for conservation or transfer activity which able scope will cover many private holdings might increase overall output and income. It is and perhaps even regional or national jurisdic- now widely acknowledged that some kind of tions. It iS quite difficult to guarantee the institutional intervention, between the two integrity of transfer agreements over such extremes of public appropriation with uniform distances and with so many intervening third redistribution and completely privatized water parties. Thus, the public sector must assume a redistibutio and c l pr lasting supervisory role to make such an sales, can provide the missing incentives to a s wee use water more efficiently. allocation scheme work. As we shall see Consider a simple case with two agricul- below, public oversight may also be necessary tural producers, identical except for their to correct for other externalities arising in water endowments, calling one arid (A) and water use and allocation. the other moist (M). Under the traditional scheme, M uses free water until its marginal Dynamic efficiency product equals zero, while A is constrained by Water is a renewable resource, but its insufficient water. Since A's marginal product recurrent supply is relatively constant over the for water is inevitably higher than that of M, long term. Once the recurrent supply is fully any reallocation of water between the two utilized, the only potential for growth in the would raise national agricultural output. All economy will come from innovations which 60 VOLUME II reduce the water intensity of agricultural agriculture.8 Agriculture has been the major production. To avoid a period of stagnation consumer under the traditional rights system and forced adjustment, Morocco could begin and the current administered system of water now to foster the innovations that will reduce distribution at controlled prices. On the other the intensity of its water use and raise its hand, agricultural water use diminishes not growth potential. only the quantity, but the quality of water The literature on economic growth is re- available for other uses. Water use in farming plete with examples of how the correction of occasions a number of negative externalities, market failures can guide resource allocation including loading of water with high mineral and use to more sustainable paths. Dasgupta concentrations, fertilizers and their organic and Heal (1979) and, more recently, Grossman derivatives, and pesticide residues. These and Helpman (1991), are two prominent potential shortcomings must be weighed examples of approaches that seek to elucidate against the economic benefits of increased sufficient conditions for improving dynamic farm output and exports. efficiency. Both emphasize the importance of From the perspective of the urban sector, market incentives to invest and innovate as a water is already an economic commodity, means of raising the sustainable trajectories of although its price rarely measures its produc- output and real income growth. It is clear, for tion or opportunity cost with any meaningful example, that the reforms proposed above to degree of precision. Urban households in improve static efficiency of water allocation developing countries also have incomplete would also improve dynamic efficiency by information about water quality, and interven- promoting both conservation and shifts to tion is usually necessary to protect the public other inputs, including new technologies. interest. These considerations argue for a Bringing water "into the market" will also carefully designed and comprehensive ap- facilitate intertemporal appraisal of its asset proach to national water policy, one which value and contribute to dynamic efficiency. relies on a balance between regional interests, as well as between market forces and regula- Rural-urban water allocation tory responsibility. The growth of urban areas has occasioned water reallocation since well before Roman The Moroccan CGE Model times. In the last century, however, the prob- The conceptual approach of the previous lems of urban water insufficiency have inten- section was intended to clarify some basic sified as these areas have rapidly grown in size issues. In this section, we provide a tool for and density and become focal points for indus- policy analysis in the form of an empirical trial activity. In terms of public health and model of the Moroccan economy. As an economic moderization, assuring adequate example of its use, we report some simulation quantity and quality in water supplies is one of results in the section, Simulation Results, the most serious challenges facing some .7 which evaluate the kind of water reforms developing countries.d In Morocco, the challenge of balancing dus above. water use between urban and rural constituen- Our Morocca model is tyc of mot cies is complicated by the rapid growth of 8Half the country's population lives in rural areas and these are generally the poorest segment of the population. Agriculture constitutes 15 percent of gross 7Ninety percent of all fatal childhood diseases in domestic product (GDP), 40 percent of employment, Africa are waterborne. and 30 percent of export earnings. 3. Economic Policies for Sustainable Resource Use in Morocco 61 ture. It simulates price-directed resource rural and urban areas, is assumed for our allocation in commodity and factor markets, simulations to be effected at prices fixed by with endogenous determination of domestic the government, and that water redistribution consumption and production decisions. We is costless. To model market-directed water assume that Morocco is a small importer and transfer between agricultural activities and exporter relative to the rest of the world, and also between rural and urban activities, a two- thus world prices are taken as exogenous to stage transformation frontier might be speci- the model. In a given sector, demand for fied between Rainfed and Irrigated and be- domestic goods and imported substitutes is tween rural and urban, respectively. The characterized by a standard constant elasticity nonmarket allocation mechanism used here of substitution (CES) specification of product most closely resembles current Moroccan differentiation. Likewise, domestic producers policy, however, so we focus on this in the supply output to constant elasticity of transfor- simulations below. '0 mation (CET) differentiated domestic and Like the social accounting matrix (SAM) to export markets. which it is calibrated, the CGE model distin- The formal structural equations of the guishes three types of final demand: house- Moroccan CGE model are presented in Annex hold, investment, and government demand. In 3-A. They are typical of one-country models all the simulations reported below, we assume of this type, except for the treatment of re- that the latter two are held constant in terms of gional and sectoral factor allocation.9 The the numeraire. We report a variety of aggre- agriculture sector is decomposed into two gate and household-specific results to give an (generic) regions, and Irrigated, according to indication about how real output, resource use, method of water use. Each agricultural and general purchasing power are affected by subsector chooses its factor mix and output the policies considered. level individually, subject to endogenous Both increased water prices and reduced domestic prices for output, intermediate in- tariffs have direct fiscal implications, of puts, labor, and property. Property or capital is course. In the case of water, we specify that assumed to be immobile within and between the net change in water revenue is redistrib- agricultural sectors and mobile between other uted to households in proportion to their sectors. This leads to imperfect factor price incomes. Otherwise, we have chosen to hold equalization across sectors, but we assume the numeraire levels of government consump- that land is essentially dedicated to agricul- tion constant to minimize distortions in final tural activities, and that its irrigation status demand. This leaves the government financing remains constant over the time horizon of the gap endogenous, and we assume this would be simulations. Firms are assumed to be perfectly reconciled with nondistortionary taxes or competitive, with average cost pricing under transfers. constant returns to scale. The model is calibrated to a 1985 social Water is a factor of production in this accounting matrix for Morocco, and herein lie model, and its use is determined by neoclassi- two of its distinguishing features. First, the cally derived factor demand criteria. We Moroccan SAM distinguishes between urban assume that water is costless in Rainfed agri- and rural income groups. This is essential to culture (see the social accounting matrix in Annex 3-B). Water transfer between Rainfed an Irgte agiulue as well as between '"The results of most of our simulations entail water savings (less transfer) and increased water revenue vis-a-vis the status quo. This mitigates the severity of the water mobility assumption and means the 9See, e.g., de Melo and Tarr (1992) for a more fiscal implications of the policies examined are gener- detailed exposition on models of this type. ally positive. 62 VOLUME II evaluate the issue of social efficiency in water moval of nominal Moroccan tariffs, which in distribution. Second, the SAM details value 1985 averaged 21 percent across the economy added for three factors of production, labor, and 32 percent in agriculture. The third and capital, and water. This decomposition of final experiment combines the water price and factors is particularly important to the model- liberalization policies of the first two simula- ing of agricultural sectors. The original SAM tions. detailed twenty agricultural sectors, but for the The aggregate results indicate that, other present simulations, we have chosen a three- things being equal, reforming water prices in sector aggregation to facilitate more intensive the agriculture sectors will have a contradic- discussion of the general policy issues. The tory effect on the economy. This is to be three-sector SAM is presented in Annex 3-B. expected, since the price increase is basically To completely calibrate the model, the taking the form of a distortionary tax against above data on observed economic activity a leading sector of the economy. Incomes and must be combined with estimates of structural real consumption of both rural and urban parameters. These include elasticities of de- households decline slightly, and real con- mand (Equation 3.5 in Annex 3-A), factor sumption falls somewhat more as increased substitution (Equation 3.7), and transforma- factor prices are passed through to agricultural tion between domestic and export markets commodities. The government budget is (Equation 3.9). Table 3-1 details the values affected negligibly. Thus, the static effects of used in the all the simulations reported in the the policy appear to be detrimental, but a next section. Even in an aggregate model such silver lining appears in the results for water as this one, results are sensitive to these pa- use as the economy moves decisively toward rameters, and we report on this in Annex 3-B. greater sustainability. In the rural sectors, water use falls by about one third and even Simulation Results urban water use is also driven down slightly by declining aggregate demand. In this section, we report on the results of a Trade liberalization has more salutary few policy simulation experiments with the effects on the static efficiency of production Moroccan CGE model. These experiments are and real incomes. Economywide real GDP intended to illustrate simulation methodology rises only slightly since the main factors, labor in the context of two important general eco- and capital, are fixed in total supply. Despite nomic issues currently facing the country, this, household incomes and real consumption trade orientation and sustainability of water post significant gains substantial import barri- resources. In particular, we conducted three ers are reduced, domestic purchasing power experiments to evaluate generic policies of rises, exports become more competitive, and trade liberalization and water price reform. resources are allocated more efficiently across We find that the results of each policy are the economy. While this typical neoclassical interesting in their own right, but that when result supports greater initiative for Moroccan they are implemented in concert, the economy trade reform, two drawbacks are readily appar- reaps the benefits of each policy without the ent. First, the government has foregone an serious negative side effects of either. important sources of revenue by eliminating Table 3-2 presents the aggregate results of tariffs, and a gap of about 35 percent emerges the three experiments. In Experiment 1, prices in the public budget. Second, the expansionary for rural irrigation water, which cover 92 influence of liberalization has increased do- percent of the country's marketed water use, mestic water use substantially. Thus, the are doubled from 8 to 16 percent of their economy is on a more growth-oriented, but urban counterpart. Experiment 2 is a trade less sustainable, trajectory. policy simulation, entailing a complete re- 3. Economic Policies for Sustainable Resource Use in Morocco 63 The last experiment combines the first two, constant. Rainfed agriculture manages to and it is apparent that combining trade and attract labor and expand production in both water reform would confer substantial advan- water price experiments, but loses a little tages on the economy, both in the medium and ground in the balance between expanding long term. The expansionary effects of trade exports and increased import penetration in liberalization are largely retained, but reform- the trade reform only experiment. ing water prices still induces substantial re- The other two sectors behave in predictable ductions in agricultural (and economywide) ways. Increased water prices and reduced real water use. Although this conservation is partly incomes reduce demand for Manufactures. offset by expanding demand in urban sectors, Since Manufactures are more intensive in the net result for the economy is less water agricultural intermediates and therefore expe- consumption. Thus, the higher growth path rience more water price transmission, both under the combined policy is more sustainable their domestic and external demand are re- than was the status quo. The government duced. Services thus become relatively more budget still declines appreciably with tariff competitive, but the effect is barely signifi- revenues, but this might be offset by alterna- cant. When Morocco eliminates tariffs in tive, nondistortionary source of revenue. Experiment 2, the more tradeable sectors More detailed results for sectorai adjust- expand output, the less so (Services) contracts ments in the three reform experiments are in the presence of economywide constraints given in Table 3-3. Note first that agriculture on labor and capital. In the combined experi- in general and irrigated agriculture in particu- ment, the results of larger magnitude in the lar suffer output declines in the two water individual experiments prevail. In other price increase experiments. In Experiment 1, words, Agriculture and Services contract for agriculture obviously suffers from factor the different reasons, and Manufacturing taxation, but in Experiment 3, the expansion- expands. Indeed, the results of the two poli- ary effects of trade reform fail to offset this cies for nonagricultural sectors are almost because agriculture is the most protected of additive. the three sectors. Irrigated agriculture, where the water price increase is directly incident, Conclusions contains both the most protected (e.g., sugar) and most export-oriented crops (e.g., tree Like many other developing countries, crops), and in this case, the export expansion Morocco faces serious water resource con- offsets most of the combined contractionary straints. The continuation of current water use effects of tariff removal and higher water practices into the next century threatens to taxes." Irrigated farming is induced to cut sharply attenuate the development process. A water use almost 50 percent in the first experi- more sustainable approach to development ment, and more than one third in the com- therefore requires that appropriate mechanism bined trade and water reform experiment. In for water production and allocation be intro- the case of Experiment 2, the benefits of duced. In this chapter, we evaluate direct and export expansion outweigh the costs of in- indirect economic policies to influence water crease import competition, and irrigated agri- use patterns, including changes in water: culture expands slightly when water prices are pricing and trade reform. Our results indicate that there is considerable scope of substitution between water and other factors of production "It would be desirable to disaggregate this and and that economic incentives to promote this other sectors to trace the more detailed effects of these policies. This is the subject of work in progress, which can move Moroccan resource use decisively includes the estimation of a 125-sector SAM for the toward a more sustainable path of economic country. development. 64 VOLUME II To assess the country's water use policy in Moroccan trade reform, the medium-term an economywide setting, we chose a CGE effects on incomes would be more than offset, model. The modeling methodology is still with rural, urban, and aggregate real income under development, and the data we have been rising substantially while still achieving sub- able to assemble thus far are highly aggregated stantial water savings. Thus, the combined and preliminary; our results should be inter- policies move the economy onto a path that is preted accordingly. Despite this, the conclu- at once more prosperous and sustainable. sions that emerge from the general equilib- While more precise quantitative estimates rium analysis are robust to reasonable parame- await refinement of the model and more ter variations. Most importantly, it is apparent intensive data gathering, our preliminary that increases in water paces, particularly in results demonstrate that a combination of agriculture, can realize substantial resource economywide and sectoral policies are neces- savings. sary to secure a sustainable basis for the coun- The general equilibrium results suggest that try's future. Our results indicate that piece- agricultural water use, which constitutes 92 meal approaches are unlikely to achieve the percent of Morocco's total use, could be combined objectives of static efficiency and reduced by more than a third if rural water dynamic sustainability. The potential effi- prices were doubled (even though in this case ciency gains and resource savings from inte- they would still only reach about 16 percent of grated, economywide policy reform are sub- urban water prices). Taken in isolation, such a stantial and, despite the short-term adjust- policy would secure a more sustainable basis ments they might occasion, they can increase for future income growth, but would reduce medium-term real incomes and facilitate medium term real rural incomes and Moroc- sustainably high real growth in the future. can GDP. If water price reform were under- taken in concert with more comprehensive 3. Economic Policies for Sustainable Resource Use in Morocco 65 Table 3-1: Estimates of Structural Parameters Agriculture Manufactures Services CES demand elasticity 0.9 1.4 1.4 CET supply elasticity 1.4 1.4 1.4 CES factor elasticity 0.6 0.9 1.1 Table 3-2: Aggregate Simulation Results (percentage changes from base) Experiment I Experiment 2 Experiment 3 Real GDP -.65 .79 .11 Rural Income -.92 8.69 8.48 Real consumption -1.21 8.97 8.23 Urban Income -.92 8.58 7.77 Consumption -1.02 7.07 Government balance* -.59 Water use Rural -34.28 1.64 -34.48 Urban -1.23 9.20 8.02 TOTAL -28.66 -4.12 * Change in the government budget balance as a percentage of the overall budget. 66 VOLUME I Table 3-3: Sectoral Simulation Results (percentage changes) Experiment I Output Exports Labor Capital Water Imports Agriculture Rainfed 23.84 17.36 28.23 Irrigated -25.40 -29.30 -22.76 -49.33 TOTAL -3.24 -8.30 .19 .27 -34.28 1.75 Manufactures -.59 -.63 -.60 -.47 -1.46 -.55 Services .01 .64 00 .16 -1.05 -.67 Experiment 2 Agriculture Rainfed -.18 12.66 -.78 Irrigated .34 13.24 -.77 3.65 TOTAL .10 13.19 -.78 1.64 17.46 Manufactures 4.01 23.25 4.18 2.96 11.23 9.32 Services -1.05 11.27 -.95 -2.38 7.30 -12.96 Experiment 3 Agriculture Rainfed 1.03 7.22 4.57 Irrigated -7.36 -1.69 -4.82 -34.75 TOTAL -3.58 -.79 -.60 -15.64 18.71 Manufactures 3.54 22.76 3.65 2.65 9.80 8.73 Services -1.03 12.16 -.97 -2.13 6.26 -13.70 3. Economic Policiesfor Sustainable Resource Use in Morocco 67 Annex 3-A: Model Equations Formal Structure of the Morocco Model Domestic demand c Th Qj =LES (PQ, Y,-)=Y-+ MiPQ1y_) 3.1 PQi 0i G=Si G -~G 3.2 Qi =s; Q3.3 QI=S 3.3 QiV=En= a jX. 3.4 Demand Allocation Between Domestic and Imported Goods Qi=CES(Di,Mj,Ai), Q=QiC+QiV, 3.5 Di [( (l-aj)j( PM. 3.6 Production Technology and Factor Demands Xk= CES (LLk,Kik ,Wik 'Pik) 3.7 X -k (l -v,*) TC.k Pk , L K F,k=-- F=L,K,FW 3.8 ~AXi ' pWi 68 VOLUME II Supply Allocation by Destination of Commodities and Services Xik= CET(Sik' Eik' tlik) 3.9 sik[( (I ik)( PEikJ 3.10 Composite Domestic Prices PQi Q=PD, D +PMj Mj 3.11 PSi Sj=Ek PSik Sik 3.12 Foreign Prices PMj=(l +tMd) PWMj R 3.13 PEi =(1 +tE ) PWEjR 3.14 Domestic Market Equilibrium Sik =Dik 3.15 PS ik=PDi 3.16 3. Economic Policies for Sustainable Resource Use in Morocco 69 ins ` k Fjk=Fv, F=L, K 3.17 ivw=ww 3.18 k -k Ki =IYi, i=Agriculture 3.19 Cost Tcik aik lxik E bK, F'k w Ik +En, a PQjX 3.20 WFik=WF WFik i*Agriculture, F*W 3.21 Pricing PXik =ACik 3.22 Private and Public Income Y = E hiFkWFJkFik k+R +Yc QG 3.23 FYL,K, W YG=I- Ek tikPX ikX 1k +RE7.[CMI PWMi Mi EktEikPWEEiEk] g,lOiWEk WWikWik 3.24 70 VOLUME 1I Balance of Payments I- [Ek PWEiEik -PWMjM,.]=h 3.25 Numdraire ., PDiDi Fn,p, PDjD 3.26 Variable and Parameter Definitions ACik Average cost of firm type k in sector i B Net foreign savings Di Total domestic demand for output in sector i E; Total exports in sector i Eik Exports of firm type k in sector i Fi Total demand for factor F in sector i Fik Factor F demand by firm type k in sector i FCik Fixed cost of firm type k in sector i Fs Economywide factor F supply Mj Total imports in sector i Nik Number of firms of type k in sector i PDi Domestic demand price for output in sector i PEi Export demand price for output in sector i PEik Export demand price for output of firm type k in sector i PMi Import price in sector i PQj Composite demand price in sector i Psi Domestic supply price in sector i PSik Domestic supply price for firm type k in sector i PWEik World export demand price for output of firm type k in sector i Pxi Ouput price in sector i PXik Output price of firm type k in sector i PWMi World import supply price in sector i Qic Composite (domestic and imported) domestic demand for output of sector i 3. Economic Policies for Sustainable Resource Use in Morocco 71 Qi Composite consumption demand for output of sector i Ql Aggregate real investment demand QG Aggregate real government demand Qiv Composite intermediate demand for ouput of sector i R Exchange rate rh Remittance income accruing to household h Si Total domestic supply in sector i Sik Domestic supply of firm type k in sector i tEik Export subsidy rate for output of firm type k in sector i tMi Tariff rate in imports in sector i TCik Total cost of firm type k in sector i tik Sectoral producer or indirect tax rates VCik Variable cost per unit output WF Economywide average price of factor F WFik Price of factor F paid by firm type k in sector i Wik Water demand by firm type k in sector i Xi Total ouput in sector i xik Output of firm type k in sector i Xi Output of sector i Y Aggregate domestic income YG Government income Yh Household income Parameters Pi CES elasticity of substitution between product varieties in sector i °ik Base share of exports in value of output of firm type k in sector i pik Factor substitution elasticity of firm type k in sector i Ogiw Share of water revenue going to government 0hiF Share of factor F income from sector i going to household h aj Base share of imports in domestic demand for sector i ;L CES elasticity of substitution between imports and domestic goods T)i Marginal budget share for consumption of good i Yi Subsistence consumption of good i tik CET elasticity of transformation between domestic and export markets PFik Factor F share in value-added of firm type k in sector i ADi Calibrated intercept parameter for demand 72 VOLUME II a,j Intermediate demand share for good i by sector j AXik Calibrated intercept parameter for production Subscripts i Sectors of production-Agriculture, Manufactures, Services k Subsectors, in the case of Agriculture only, where the two subsectors are Rainfed and Irrigated F Factors of production-Labor, Property, and Water h Households-Rural and Urban Note: Variables are denoted by English letters, and structural parameters by Greek letters. An overstrike denotes a base value in the case of a variable and an exogenously specified value in the case of a parameter. More than one pricing rule is specified above, although they are mutually exclusive in the simulations. Social Accounting Matrixfor Morocco, 1985 (millions of current DH) AgRahr-Fed Agirrig Manufact Services Labor Cetral Water Tariff Cp Tax IncfTax HHRural Hurban Govt CapAcct Rou Total 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 o t* 1.AgRainFed 199 244 2829 15 0 0 0 0 0 0 3062 4277 0 1430 134 12190 -- O 2.Agirtig 348 298 3458 18 0 0 0 0 0 0 3062 5227 0 1748 613 14771 3.Manutact 1318 1611 47376 22625 0 0 0 0 0 0 11074 23273 0 13273 12186 132737 4. Services 525 642 9516 11542 0 0 0 0 0 0 7591 24851 16399 23500 3785 98352 I'S S Labor 1555 1901 11219 43258 0 0 0 0 0 0 0 0 0 0 0 57933 6. Capital 5125 6264 11599 14437 0 0 0 0 0 0 0 0 0 0 0 37425 ;t 7. Watet 590 722 1497 1863 0 0 0 0 0 0 0 0 0 0 0 4672 6 Tariff 598 731 7719 0 0 0 0 0 0 0 0 0 0 0 0 9047 9.ExpTax 9 10 314 0 0 0 0 0 0 0 0 0 0 0 0 333 10 IndTax 55 68 2683 463 0 0 0 0 0 0 0 0 0 0 o1 3269 11.HHRural 0 0 0 0 7782 12832 0 0 0 0 0 0 306 0 85S9 29509 12.HHUtban 0 0 0 0 50151 24593 0 0 0 0 0 0 436 0 1200 76379 13 Govt 0 0 0 0 0 0 4672 9047 333 3269 583 6420 31 806 663 25823 14. CapAcct 0 0 0 0 0 0 0 0 0 0 4137 12331 3020 0 21269 40757 15. ROW 1867 2281 34527 4131 0 0 0 0 0 0 0 0 5631 0 0 48437 ToTAL 12190 14771 132737 98352 57933 37425 4672 9047 333 3269 29510 76379 25823 40757 48437 0 N 74 VOLUME II Annex 3-C: Results of Sensitivity Analysis Because the Morocco model is calibrated would expect, the magnitudes of the results to an equilibrium SAM dataset, most of its are monotone in the elasticity, indicating that information requirements are met from direct greater flexibility in factor substitution yields observation of the economy under study. more dramatic adjustment. Despite this, the Although this small model is rather parsimo- essential endogenous variables summarized nious, a number of its structural parameters here vary much less than the elasticity parame- must still be specified with indirect informa- ter. From these comparisons we infer that our tion, using econometric estimates from other general conclusions about the three policies sources and judgment about values which are robust. In particular, the efficacy of com- might reasonably be expected to apply. Of bined trade and water reforms, in terms of particular interest in the present simulations is water conservation and real income growth, is the elasticity of substitution between produc- consistently supported. tive factors in the agricultural sectors. This Another way to appraise the sensitivity of CES parameter (p in Equation 3.7 above) is a the model is to experiment with a wider range key determinant of the adjustment in irrigated of exogenous shocks. Table 3C-2 reports the water use resulting from a price increase. results of a variety of alternative water price Table 3C-1 summarizes the results of nine increases, including increases of 1.5, 2, and 4 simulations we conducted to assess the CGE times current rates for irrigated agriculture. model's sensitivity to this parameter. We The last case corresponds roughly to an equal- replicated each of the policy scenarios re- ization of rural and urban water prices. There ported in section five (experiments 1-3) with are six replications, three each for experiments three alternative values to bracket the experi- 1 and 3, experiment 2 being omitted since it ment outcomes. Substitution elasticities in entails only tariff reductions. The qualitative agriculture are generally conceded to be rather results are fully consistent within and across low, and we chose a central case of .6 as a experiments, and their magnitudes vary in the reference.'2 expected direction and by about the same Note that, with only two exceptions in relative magnitude of the shock. From this we thirty-six, the results are qualitatively consis- infer that water price reforms of any reason- tent, and the exceptions are within a reason- able degree would yield significant conserva- able neighborhood of zero in any case. As one tion, and that a concerted effort to reform water policy and trade would achieve agricul- tural water saving with increased rural income. '2Recall that, in light of capital immobility in agriculture, this elasticity is actually restricted to substitution between labor and water. 3. Economic Policiesfor Sustainable Resource Use in Morocco 75 Table 3C-I: Resultsfor Varying Factor Substitution Elasticity Experiment I Experiment 2 Exneriment 3 lOw Central High Low Central High Low Central High .3 .6 1.2 .3 .6 1.2 .3 .6 1.2 All agricultural output -2 -4 -5 .1 . 1 .3 -2 -4 -5 Irrigated ag output -5 -8 -12 -.1 .3 .8 -5 -7 -11 Irrigated ag water use -22 -37 -58 2 4 8 -20 -35 -55 Rural income -.1 -.1 .3 9 9 9 8 9 9 Table 3C-2: Resultsfor Varying Water Prices Exneriment I Exveriment 3 Low Central High Low Central High 1.5x -2x 4x 1.5x 2x 4x All agricultural output -2 -4 -8 -2 -4 -8 Irrigated ag output -4 -8 -17 -4 -7 -17 Irrigated ag water use -24 -37 -67 -21 -35 -60 Rural income 0 0 0 8 7 6 76 VOLUME I Bibliography Arrow, K. J., and R. C. Lind. 1970. "Uncer- Gibbon, Diana. 1987. The Economic Value of tainty and the Economic Evaluation of Water Washington, D.C.: Resources for Public Investment Decisions." American the Future. Economic Review 60:364-78. Grossman, G. M., and E. Helpman. 1991. Boggess, W., R. Lacewell, and D. Zilbeiman. Innovation and Growth in the Global 1993. "Economics of Water Use in Agri- Economy. Cambridge, Mass.: MIT Press. culture." In M. Osborne (ed.), Agricultural Hotelling, H. 1931. "Economics of Exhaust- and Environmental Resource Economics. ible Resources." Journal of Political Clarendon, U.K.: Oxford University Press. Economy 39:137-75. Burness, H., and J. Quirk. 1979. "Appropria- Nobe, K. C., and P. K. Sampath. 1986. Irriga- tive Water Rights and Efficient Allocation tion Management in Developing Coun- of Resources." American Economic Re- tries: Current Issues and Approaches. view 69:25-37. Boulder, Colo.: Westview Press. Caswell, M., E. Lichtenberg, and D. OECD. 1989. Water Resource Management: Zilberman. 1990. "The Effects of Pricing Integrated Policies. Paris: OECD. Policies on Water Conservation and Drain- Parlin, B., and M. Lusk. 1991. Farmer Partic- age." American Journal of Agricultural ipation and Industrial Organization. Economics 72:883-90. Boulder, Colo.: Westview Press. Chakravorty, U., and J. Roumasset. 1991. Pearce, D. W., and R. K. Turner. 1990. Eco- "Efficient Spatial Allocation of Irrigation nomics of Natural Resources and the Water." American Journal of Agricultural Environment. Baltimore, Md.: Johns Economics 73:165-73. Hopkins University Press. Clarke, C. W. 1976. Mathematical Postel, Sandra. 1992. Last Oasis: Facing Bioeconomics: The Optimal Management Water Scarcity. New York: Norton. of Renewable Resources. New York: John Reigner, Marc. 1990. Overtapped Oasis: Wiley and Sons. Reform or Revolution for Western Water Dasgupta, P., and G. H. Heal. 1979. Economic Washington, D.C.: Island. Theory and Exhaustible Resources. Cam- Reinert, K-A., and D. W. Roland-Host. 1991. bridge, U.K.: Cambridge University Press. "Structural Parameter Estimates for Trade de Melo, J., and D. Tarr. 1992. A General Policy Modeling." Discussion Paper, U.S. Equilibrium Analysis of U.S. Foreign International Trade Commission, Wash- Trade Policy. Cambridge, Mass.: MIT ington, D.C. Press. Tuluy, H., and B. L. Salinger. 1989. "Trade, Dinar, A., and D. Zilberman. 1991. Econom- Exchange Rate Policy, and Agricultural ics of Management of Water and Drainage Pricing Policies in Morocco." Compara- in Agriculture. Norwell, Mass.: Kluwer tive Studies in the Political Economy of Academic Publishers. Agricultural Pricing Policy, World Bank, Easter, K. W. 1986. Irrigation, Investment, Washington, D.C. Technology, and Management Strategies World Bank. 1989. "World Bank Experience for Development. Boulder, Colo.: with Irrigation Development," Volume III: Westview Press. Morocco, Report 7876. World Bank, Fisher, A. C. 1981. Resource and Environ- Washington, D.C. mental Economics. Cambridge, U.K.: Cambridge University Press. 4 Poverty, Migration, and Deforestation in the Philippines Wilfrido Cruz, Herminia Francisco, and Gregory Amacher IT IS NOT AN ACCIDENT that assessments of problems of environ- mental degradation invariably bring up concerns about poverty. In many instances, the worst effects of environmental pollution or natural resource degradation are borne by the poor. In both urban and rural areas and in various occupations, they are the ones who can least afford to protect themselves from environ- mental degradation. The poor spend long hours in polluted factories; they are exposed to agricultural chemicals; and services that are taken for granted by those who are better off, such as clean water and waste disposal, are normally unavailable in slums and rural areas (World Bank 1992; WCED 1987). 'rhe authors are grateful to Ma. Concepci6n J. Cruz, Jeremy Warford, Gershon Feder, Stein Hansen, William Hyde, lmelda Zosa-Feranil, and Shreekant Gupta for their comments and suggestions. Murty Kamilbady provided valuable research assistance in implementing the migration model. The authors are, of course, solely responsible for the views in this chapter. 77 78 VOLUME II Poverty also often turns out to be an under- study. Even with this information it was lying reason for households or communities to necessary to substantially simplify the applica- degrade resources. For example, short tion of the migration model; future work is decisionmaking time horizons, associated with clearly required in this area. In this context, poverty, prevent many farmers from investing the policy implications identified in the last in soil conservation techniques that may be section should be viewed as a limited, though necessary for sustainable agriculture but important step to improve current understand- whose payoffs are not immediate (Mink ing of poverty, migration, and deforestation 1993). interactions. While the links between poverty and envi- ronment are many, this chapter has the limited Deforestation, Upland Population scope of focusing solely on one particular Growth, and the Role of Policy component of the poverty problem: the large Failures number of rural poor It also focuses attention on a particular aspect of environmental degra- In this section, we describe the deforesta- dation: the conversion of forest land to unsus- tion trend in the Philippines, focusing on land tainable agriculture. Both these issues consti- use changes. We then argue that, together with tute pressing problems for developing coun- commercial logging, the growth of upland tries. In the rural areas of many countries of population has been a key part of the process the developing world, too many people de- of deforestation since increasing conversion of pend on too few resources for their livelihood, land to agriculture has effectively prevented resulting in persistent rural poverty and creat- forest renewal. Finally, we discuss the range ing pressures to exploit even fragile resources, of sectoral and macroeconomic policies (or such as forest lands. policy failures) that have contributed to the The goal of this study is to examine the role incentives motivating this pattern of migration of poverty and population pressure in defores- from lowlands to forest lands. Because these tation through the conversion of forest lands disincentives to environmentally sound behav- to agriculture. A second, related objective is to ior arise not from externalities but from mis- determine the potential for combining poverty guided policies, their effect has been referred alleviation reforms with environmental goals to as "policy failures" (Pearce and Warford in the area of forest land management. To 1993).1 accomplish this, the study builds on previous work on estimating the extent and causes of Deforestation increasing population pressure on Philippine uplands (M. C. Cruz, Zosa-Feranil, and Goce Recent assessments have documented the 1988). An empirical migration model is then rapid decline of forest resources in the Philip- constructed, based on the most recent Philip- pines, the growth of forest land communities, pine census, to focus explicitly on the role and the expansion of cultivated areas in mar- played by economic, environmental, and ginal areas. Between 1970 and 1985, the demographic factors in motivating migration Bureau of Forest Development (1985) re- from lowland communities to forest lands. ported a deforestation rate of 85,000 hectares The limitations of this modeling exercise must a year. A higher estimate of 95,000 hectares be noted. The lack of economic and demo- per year in the same period was cited in FAO graphic data that differentiate between low- land and forest sites was a formidable obsta- 'In contrast to the term "market failure," which cle, and a major investment in time was neces- applies to the environmental disincentives arising from sary to generate new information for this externalities. 4. Poverty, Migration, and Deforestation in the Philippines 79 (1983). More recently, the Philippines Natural ally forested). With this classification, about Resource Accounting Project (1992) in the 14.1 million hectares, or close to half of the Philippines has estimated an average reduc- land area of the whole country, are forest tion in forest cover of 210,000 hectares per lands and under the jurisdiction of one gov- year for 1970-89. This process of deforesta- ernment department. Private users are for- tion and the subsequent conversion of forest mally excluded from these resources or al- land into nonsustainable agriculture are per- lowed access only as concessionaires, under haps the most costly forms of resource degra- government jurisdiction. The pattern of timber dation in the Philippines. felling and the permanent conversion of forest Actual forest cover declined from more lands to agriculture is directly affected by this than 34 percent to about 24 percent during the resource management approach. This ap- last two decades. Table 4-1 provides the cur- proach makes the assumption that centralized rent land use in the Philippines, based on the regulation of such vast areas is feasible. How- estimates made by the Swedish Space Corpo- ever, the incentives governing the behavior of ration. By 1988 the forest cover had declined forest users undermine this management to only 24 percent with areas devoted for approach. extensive and intensive cultivation comprising Both the commercial loggers and the farm- 72 percent of land cover. The areas under ers who have settled illegally on forest land extensive cultivation are made up of planta- contribute to the deforestation process. With tion areas and grasslands, which are usually respect to the logger's structure of incentives, deforested lands and are degraded by slash- low timber cutting charges make him under- and-burn farming. value the true worth of the forest as a renew- able resource. He therefore worries only about Upland population and conversion of maximizing the returns to his investment from forest land to agriculture harvesting the resource. This perspective leads Soils and the natural resources directly him to harvest more than if he were made to associated with land use can be generally appreciate the much greater value of the for- classified as land resources. This incorporates est. The latter would be possible only if both various characteristics, such as fertility, topog- the market and environmental value of the raphy, and vegetative cover that result in trees were charged to him and if a long substantial land quality and productivity enough planning horizon is provided so that distinctions. In most countries, the lands he may appreciate the long-term returns to suitable for intensive agriculture are generally forest management. As it is, the private logger the lowlands. (Two notable exceptions are the harvests considerably more trees and at an age hill regions of Nepal before the eradication of earlier than would be appropriate from a social malaria in the Terai and the Altiplano of decisionmaking perspective. Bolivia.) From the perspective of upland farmers, the Boin .) t continuing conversion of public lands to In the Philippines, the term lowlands gener- arcluei nieial euto h ako ally corresponds to land officially classified as agriculture IS an inevitable result of the lack of "alienable and disposable"-that is, suitable alternative livelihoods in lowland agriculture for privatization. The technical criterion used and in industry. Although crops can be grown for alienable and disposable lands is whether in marginal or even in forest lands, doing so their slope is less than 18 percent. Lands with generates substantial environmental externali- 18 percent or greater slopes are deemed map- ties. For example, soil erosion can damage propriate for agriculture and are kept in the irrigation and power facilities downstream in public domain. They are referred to as uplands the "lowlands," reduce water quality, and orforest lands (whether or not they are actu- aggravate seasonal flooding. Erosion also 80 VOLUME II reduces soil productivity in the uplands. In information is classified according to adminis- this case, agriculture is not sustainable or can trative jurisdictions (with municipalities as the be maintained only with increased inputs of basic unit for data processing) and not accord- capital and labor. ing to environmental boundaries. To make The decisionmaking perspective of the such information relevant for environmental upland farmer explains this propensity for assessment, it is necessary to reclassify admin- degradation. For him, the absence of secure istrative categories into environmental ones. claims to his upland plot creates a Cruz et al. (1988) have used the official classi- decisionmaking perspective where the long- fication system for "forest land" (based pri- term worth of the land is disregarded in favor marily on slope) to devise a conservative of harvesting or growing as much as may be method for identifying municipalities and the possible in the short term. The primary eco- proportion of their population that inhabit nomic constraint that he perceives is the forest lands. This system identifies 709 mu- amount of labor that his household can gener- nicipalities as forest land municipalities, ate to exploit the land. At the same time, the representing 47 percent of all municipalities upland farmer is not concerned with the and more than half of the country's total land downstream externalities generated by his area of 30 million hectares. activities. The results are on-site land resource Using this method and census data from degradation and off-site damages through soil various years, M. C. Cruz et al. (1992) have erosion. estimated the increasing number of people There have been attempts to distinguish inhabiting forest lands from 1950 to 1990. between deforestation from commercial log- (see Table 4-2). It was estimated that forest ging and fiom encroachment, and to identify land population has been increasing and ac- the "primary" source of deforestation (e.g., counted for as much as 30 percent of total Kummer 1990). This distinction, however, is population in 1980 (M. C. Cruz, Zosa-Feranil, not very fruitful for resource management and Goce 1988). Together with population concerns. Deforestation is often associated increase in forest lands, recent assessments of with both commercial logging and encroach- Philippine land use conclude that large areas ment of migrants into the uplands. These two previously forested have already been con- are interrelated in the sense that commercial verted into farms. Extensive agriculture con- logging by building roads and reducing forest tinues to expand in forest lands, resulting in clearing costs often paves the way for mi- soil erosion (World Bank 1989a). In the 1960s grants who engage in shifting cultivation, cultivated area in the uplands amounted to Since logging by itself will not necessarily only 10 percent of lowland cropped area. By lead to a permanent change in land use, it is the 1980s it had increased to about 40 percent. the growth of forest land population that is the This means that the growth of upland culti- effective cause of land conversion. vated area had averaged 7 percent or more per The difficulty, of course, is in estimating year for the past three decades (W. Cruz and the population in forest lands. Both because of Repetto 1992) (see Figure 4-1). the geographic extent of the problem as well as its lack of official recognition until re- What Motivates Increasing Upland cently, there is no direct source of information Population? on rmigration to forest lands. We therefore use Our analysis of the link between poverty the methodology developed by M. C. Cruz, and deforestation focuses on the role of up- Zosa-Feranil, and Goce (1988) to modify land migration. This extends work done in census categories of residence according to three important areas: (a) analyzing the sys- environmental maps. Conventional census tematic relationship among problems of popu- 4. Poverty, Migration, and Deforestation in the Philippines 81 lation pressure, agricultural stagnation, and cycle of poverty, population pressure, and environmental degradation (what has been environmental degradation. referred to as the population-agriculture-envi- ronment nexus); (b) the identification of The role of internal migration migration to forest lands as a direct mecha- While absolute increases in population nism relating poverty in rural lowlands to the have crucial long-run implications, the imme- exploitation of marginal resources; and (c) the diate problem of population pressure on de- recognition of the contribution of macroeco- graded resources is directly related to the nomic policy distortions to the problem of geographical distribution of population. Tradi- persistent rural poverty and subsequently to tional assessment of migration has focused on upland migration and deforestation. the attraction of cities and patterns of rural to urban movements (Yap 1 977; Williamson The population, poverty, and degradation 1988). Howver, ecnoi stagnationi nexus 1988). However, economic stagnation in industry and traditional agriculture has led Various studies have considered the sys- growing numbers of households to migrate tematic relationship between rapid population from traditional, lowland agriculture to mar- growth, rural poverty, and environmental ginal forest lands in search of alternative degradation. Boserup (1965) identified the livelihoods. Thus, internal migration, from problem of population pressure leading to lowland to upland sites, has been the main intensification of agriculture. Although she mechanism linking conditions in lowland recognized the questionable impact of intensi- communities to the problem of rapid popula- fication on quality of life as more labor is tion growth in uplands (M. C. Cruz, Zosa- utilized per unit of land, she did not explicitly Feranil, and Goce 1988). address the environmental degradation associ- ated with the decline of the fallow and the The role of macroeconomic and sectoral increased dependence on chemical inputs. policies More recent concern with environmental The factors that lead to pervasive problems aspect of increasing population pressure on of rural poverty, as well the factors that en- limited agricultural resources has led to a r J courage migrants to occupy (and overexploit) broader appreciation of the nature and the forest lands are all affected by economywide extent of the problem. Cleaver and Schreiber and sectoral policies. W. Cruz and Gibbs (1991), assessing the conditions of agriculture (1990) focus on tenurial insecurity regarding in Sub-Saharan Africa, focused on what they resources as a major source of inadequate called the "nexus" effect, linking land degra- investment. Poverty from lack of access to dation with high population growth, limited agricultural resources and unemployment lead agricultural lands, and the difficulty of adapt- landless rural households to exploit marginal, ing traditional techniques and institutions to aess resouses.lds to the laco rapidly changing conditions. Shifting cultiva- open access resources. In turn the lack of tpio y chandgring actiities, appropriatelinvlo property rights leads to overexploitation, tion and grazing activities, appropriate in low reue vnoe,ad essetpvry population density conditions, give way to r i a more intensive land uses. Agricultural intensi- Beyond these sector-specific institutional fication, however, cannot keep pace with factors is a more pervasive but less apparent population pressure, and resources are subse- set of misguided development policies (such quently overexploited and become unproduc- as industrial protection and exchange rate tive. Low productivity leads to economic overvaluation) that has led to economic stag- stagnation, and the resulting poverty delays nation in general and have penalized the the demographic transition, thus closing a agricultural sector in particular (Krueger, 82 VOLUME I Schiff, and Valdes 1991). Such policies create half of the 1980s. In spite of efforts at eco- pervasive distortions that significantly contrib- nomic reforms with the change in administra- ute to economic stagnation and rural poverty. tion in 1986, economic recovery has been Thus, efforts to reduce deforestation pressures quite tentative, resulting in barely positive will require not only sector-specific policy average growth rates for the 1980s. (Refer to changes, but also macroeconomic reforms Intal and Power 1991 for a detailed descrip- (Hansen 1990, W. Cruz, and Repetto 1992). tion of the main phases of economic policies.) Indeed in a recent study on the Philippines and The policy regime also had important Costa Rica, it has been proposed that attempts implications for unemployment. Capital- to address the deforestation problem will intensive industrialization was encouraged entail a combination of demographic, eco- indirectly by the protection system and di- nomic, and resource management policies (M. rectly by the investment incentives program. C. Cruz et al. 1992). During the 1970s, the system of tariffs made capital and intermediate input use relatively The role of policy failures: postwar to cheap. Similarly, Board of Investment subsi- early 1980s dies encouraged the expansion of capital and This section describes how misguided energy-intensive industries, such as pulp and policies-from the 1950s until reforms were paper, petrochemicals, and steel and copper. implemented in the 1980s-have led to eco- While the trade and investment policies pro- nomic stagnation and pervasive poverty. moted capital-using activities, foreign ex- Absolute poverty measures based on mini- change policy discouraged the development of mum food requirements or basic needs show labor-using export industries. The result was that there has been some progress in alleviat- a long-term lagging capacity to generate ing poverty in the Philippines. In the early to employment. The expected process of industry mid-1970s, 50-60 percent of the population siphoning off labor from the agricultural fell below the poverty threshold (World Bank sector was arrested. The average annual 1985). By the early 1980s the percentage fell growth of the labor force from 197088 was to about 40 percent. However, throughout the 3.6 percent, but total employment for the same 1970s and 1980s, a greater proportion of rural period grew by only 3.3 percent (NEDA households fell below the poverty threshold 1990). (World Bank 1988). Table 4-3 shows that Beyondthegeneraleffectofsluggisheco- poverty incidence, defined as the number of nomic growth on poverty incidence and unem- households falling below the income needed ployment, some studies have analyzed the households falling below the income.needed specific impacts of government policies on for basic consumption, is substantially larger agriculture. These have proposed that the net in the rural sector. inthe muracroectnomi sourceofpersistent effect of government policy in many poor The macroeconomic source of persistent conre hs bee to pealz agiutrei pvryin the Philippines was the generally countries has been to penalize agriculture in poverty istoricalipine of the enomy order to subsidize an inefficient industrial poor historical performance of the economy, sector. Due to specific subsidies in agriculture, from the postwar period until reforms were such as irrigation and credit support, it is often initiated in the mid-1980s. Gross national believed that policies of poor country govern- product (GNP) grew at respectable rates of ments generally favor agricultural activities close to 8 percent annually during the early mndtsenefra favor agicltra aties 1950s, but growth in the 1960s and 1970s and therefore incomes. However, Krueger, declined to less than 6 percent per year. There Schiff, and Valdes 1991 have demonstrated in was some improvement in the 1970s, but this a comparative study Involving 18 developing was undermined by the economic crisis that countries that government policies, in fact, resulted in net economic decline in the first generally penalize agriculture, while protect- 4. Poverty, Migration, and Deforestation in the Philippines 83 ing the industrial sector. data.2 To analyze the mechanism linking The studies showed that the direct incentive economic conditions in lowland areas to the effect for the key agricultural exports of these process of environmental degradation in forest 18 countries, measured as the proportional lands, we focus on migration streams (M) difference between a commodity's producer from lowland origins (i) to forest land destina- price and border price, was negative, averag- tions (j). ing -8 percent during the last 25 years. On top Migrants in lowland urban or rural areas of this incentive bias versus agriculture, the face a large set of potential upland (forest indirect effects of the overvaluation of the land) destinations. To capture the decision- local currency and the impact of industrial making process implicit in this process, we protection on the price of agricultural relative utilize an econometric procedure that accom- to industrial goods was even greater, averag- modates the many choices facing the potential ing -20 percent. Thus, macroeconomic poli- migrant: a multinomial discrete choice frame- cies that were designed to promote industrial- work. In this approach, it is explicitly recog- ization had the unanticipated effect of under- nized that the individuals' decision to move mining agricultural incentives and therefore involves a complex comparison of (known) rural incomes. utility the individuals receive at their origin, This intercountry finding has been studied and the utility they expect to receive at poten- in the detail in the Philippines (David et al. tial destinations. 1986; Intal and Power 1991). Focusing on the Since each individual can physically move four most important agricultural crops in the to only one site at one time, the potential Philippines, Intal and Power found that the destinations form a discrete set of alternatives. total nominal protection rates for these crops, Each migrant in the same location faces the from 1960-86, were mostly negative (refer to same alternative set. The aggregation of simi- Table 4-4). lar choices made by individuals at an origin makes possible the application of the discrete Migration from Lowlands choice framework to typical data sets, where to Forest Lands migration streams (e.g., the number moving from one location to another location) rather This section builds on the discussion of than individual data are available.4 Also, note historical patterns of increasing upland popu- that the model does not address the initial lation and the role of macroeconomic and decision on whether or not to move. Incorpo- sectoral policies in motivating migration to rating this aspect would require much more forest lands. The role of economywide deter- data than are available. (Refer to Annex 4-A minants is analyzed together with environ- for the formal presentation of the migration mental and demographic factors in upland model.) migration. This allows us to undertake an In the Philippine upland migration case, the empirical assessment of the key variables affecting migration to forest lands, and to identify relevant areas for policy interventions. 2See also M. C. Cruz, Zosa-Feranil, and Goce 1988 and M. C. Cruz et al. 1992. Migration model 3Although the migration streams are derived from While there have been many propositions municipality-to-municipality data, they are aggregated regarding the factors affecting population into provincial streams for the purpose of this analysis. movements to forest areas, this represents only one of a small number of efforts to undertake 4That is, the polychotomous choices faced by each individual are translated into similar "choices" for a quantitative analysis, using actual migration migration streams. 84 VOLUME II multinomial discrete choice model is essential are the key factors that could affect rmigration for two reasons. First, there are over 70 prov- into forest lands. These include timber conces- inces spread over a large geographic area. sion rules and delineation of government Each province provides a potential opportu- forest reserves. In addition to forest sector nity for upland exploitation by willing mi- factors, economic incentives and a rapidly grants in any other province. Second, forest growing population constitute external factors lands vary widely in quality and suitability for that contribute to pressures for migration into agriculture across provinces, making the forest lands. The economic factors affecting decision to move to a particular province migration include poverty, income and em- dissimilar with the decision to move to any ployment, and agricultural tenure conditions other province.5 that influence economic security. Finally, demographic variables (e.g., population den- Migration measures and determinants sity and conditions of urbanization) also play important roles in the migration decision. Usn the 190cnu,mgrtotem As was the case with upland migration from lowland to forest lands, M1i were deter-Aswstecewihulnmgrio from lowland tofretstream, information on the determinants for mined based on whether current (1990) resi- the regression model is not readily available. dents in forest land municipalities had resided One of the most difficult to obtain was infor- in lowland municipalities five years before (1985). The basic data source for the Ma'ston on income, whwch had to be different- .1985). The basic data source for the M swas ated between lowland and forest land areas in a municipality-by-municipality migration the regression model. The use of average matrix, obtained from the census office. The household income in the origin and destina- migration streams into the 709 forest land tion provinces (available from the Family municipalities from lowland municipalities Income and Expenditure Surveys) cannot be were aggregated into provincial lowland to interpreted as indicative of the level of low- forest land streams. This resulted in 77 desti- land and upland income. This is because nations (the forest land municipalities grouped average income measures in any given prov- into the 77 provinces in the country) and 77 ince include both the income of households in origins (lowland municipalities grouped lowland and forest land municipalities (within similarly by province). The 77 X 77 matrix the province). yielded 5,929 observations on Mij's. The The data used to construct the variables for frequency distribution of these streams is lowland and forest land average household presented in Table 4-5. Note that most of i w b I i,...r ncomes were based on a national '"bench- these Mij's were either 0 or less than 500. mark" survey undertaken by the Institute of Determinants of upland migration Agrarian Studies (LAST) of the University of the Philippines (IAST 1990a,b). This survey, As indicated in the previous sections, involving 8,935 households was commis- migration to forest lands constitutes an impor- sioned by the Department of Agrarian Reform tant factor leading to deforestation. In this and had a sampling frame that was explicitly section, we describe the various factors affect- designed to capture lowland versus upland ing forest land migration. Within the forestry socioeconomic differences. The villages (or sector, conditions of access to forest resources barangays) in the study sample were grouped by ecological zones (lowland, forest land, and coastal). The major limitation of the survey 5The use of a geographically broad data set also was that not all provinces were included in the allows assessment of economy-wide policies and the role of the environment, unlike other recent studies that sample. Data for the other determinants in the have focused on a small group of migrants, or on a model were taken both from published na- specific urban area. 4. Poverty, Migration, and Deforestation in the Philippines 85 tional surveys and from records of government migration stream, or equivalently, on each departments. Table 4-6 describes in detail the representative individual in a particular origin. definition of the variables, their sources, and Eleven of the explanatory variables are signifi- their hypothesized relationship to the migra- cantly different from zero, including many tion streams, M1j. (See Annex 4-B for a de- economic and environmental quality indica- scription of the data.) tors. Most of the variables have expected signs, except where they are not statistically Data transformation and estimation different from zero. Some transformations of the data were The first set of results, on socioeconomic required to accommodate the econometric determinants in Table 4-7, demonstrates that specification. First, the municipalities com- econorlc incentives exert substantial influ- prising the Metropolitan Manila area were ence in the migration decision. Among the combined into one "province," reducing the socioeconomic factors, the cost of migration number of provinces to 74. Second, 6 of the has the largest effect on choice of destination, 74 provinces do not contain upland forest based on the index of cost of long-range areas, so no migration could be observed into migration. Most of the income-related vari- these provinces. Therefore, these provinces ables are also important, although other re- were dropped as potential destinations; how- lated socioeconomic factors, such as farm ever, they were still included as origins for tenancy rates and provincial unemployment, migration streams. Since the goal of the mi- do not significantly affect the migration deci- gration analysis is to predict lowland to upland sion. migration trends, dropping these provinces is Both average provincial income and the not problematic and is consistent with our incidence of poverty have significant effects model. Some of the data regarding attributes on migration. Higher household incomes of each province were also missing, and af- make upland sites in the province more attrac- fected provinces had to be dropped from the tive to migrants. In addition, the more poverty model prior to computing migration propor- there is (the more households fall below the tions, leaving us with 31 possible origin prov- poverty line in a destination province), the inces and 52 possible destination provinces.6 more migrants want to go to other places. The total number of observations for this Aside from province-level determinants, reduced data set is 3,898. The missing data the economic factors also include more site- problem was particularly important for the specific income variables. The results on these upland and lowland income variables. variables suggest that further refinement of Table 4-7 presents maximum likelihood income variables-so that they directly relate estimates of the effect of the various explana- to upland versus lowland destinations within tory variables on the choice of migration. All a province-can contribute to the analysis of variables were converted into log form. These migration. Our current results suggest that estimates reflect the effect of the explanatory such "targeted" income variables have differ- variables on the destination decision of the ent effects on migration. Average upland income clearly increases the likelihood that the upland site will be chosen by migrants, although its effect on destination choice is not 6Dropping observations effectively means that the model represents fewer choices. However, dropping choices does not affect the estimation with respect to any of the other choices, under the independence of 'The index took the value of 0 when migrants to irrelevant alternatives assumption required of the upland municipalities remained within their origin multinomial discrete choice model (e.g., Amemiya province, or the value of I when they moved to 1985, pp. 300-301). municipalities in other provinces. 86 VOLUmE 11 as large as those of provincial income and very difficult for government to exclude poverty incidence (its coefficient is smaller). settlers from these lands, and in many areas By contrast, if income gains occur in purely they are defacto open access lands. lowland municipalities, these do not signifi- The econometric results show that the cantly affect migration destination. purely physical attributes of forest land re- Environmental variables (including agricul- sources (both extent of land that is forested tural resources) comprise another set of fac- and slope characteristics) are not significant tors important to the migration decision. determinants of migration. By contrast, pro- Arable land and crop land in lowland areas portion of public forest land is significantly serve as indicators of the predominance of associated with upland migration. In addition, lowland agricultural resources in a province. road density ir. forest land significantly con- If the main rural opportunities are dominated tributes to migration. These results imply that by agriculture, these provinces would be less the forest sector factors most relevant to important as upland migration destinations. migration are not the physical conditions of This is supported by our econometric results, uplands. Instead, the focus should be on indi- showing that provinces with larger crop land cators of accessibility to resources: tenure or areas have less upland migration. The size land allocation systems in the case of public alone of arable land is not a significant factor, forest lands and physical accessibility in terms indicating that what is more important is the of road density in the upland sites. ongoing use of these resources. The last group of variables in the model The proportion of cultivated land to total includes the demographic determinants of land area could be positively or negatively migration. Lowland population turns out to be linked to upland migration. As noted in Table a positive influence on migration. Large up- 4-6, a high percentage of land being inten- land population, controlling for upland popu- sively or extensively cultivated could indicate lation density, also leads to more migration. crowding in the province, thus discouraging However, upland population density itself did migrants. Alternatively, it could indicate more not have a significant effect on migration. developed markets and land accessibility, Finally, growth rate of population, as an indi- which means increased opportunities for cator of general expansion of the community, upland migrants. In the estimation results, the was positively linked to migration. proportion of intensively cultivated land leads to more migration, while extensively culti- Conclusion vated land does not appear as a significant factor in the model. Previous studies have already linked the The group of environmental factors in the continuing conversion of forest land to agri- model that are directly related to physical culture with increasing upland population, but characteristics of forest land include the per- the role of economywide factors motivating centage of land actually under forest cover population movements to forest lands has not (forest land), and the proportion which is been previously studied in detail. It had been steeply sloping (with greater than 30 percent suggested that economywide and sectoral slope). Public forest land, however, follows an economic policies (e.g., those promoting administrative classification: not all public logging) and institutional conditions (e.g., the forest land is forested, and not all are steeply open access nature of forest lands) are the sloping. These are lands that have been kept in cause of increasing forest land population. the public domain and are administered by the In this chapter, we have integrated forestry agency. As noted in the discussion in economywide considerations with both sec- the previous section, historically it has been toral and environmental factors, in a model focusing on the choice of upland migration 4. Poverty, Migration, and Deforestation in the Philippines 87 destinations. We hypothesized the role of demonstrates that public forestry management economic factors, specifically poverty and (or lack of it) aggravates migration. Thus, in income, as the key determinants of migration. this case reforms or programs to improve To assess this empirically, lowland to upland upland income reforms or alleviate poverty migration data were constructed, a database on need to be implemented with complementary a range of economic, environmental and forest land tenure and management reforms. In demographic factors was assembled, and the addition, economywide reforms that target migration model was implemented using a more directly lowland income gains, and multinomial discrete choice framework. improved lowland opportunities could lead to The most important result was to demon- income improvements without increasing strate empirically the role of economywide migration incentives. incentives motivating upland migration deci- Finally, focusing on the role of environ- sions. Poverty incidence and perceived in- mental management, policy-relevant instru- come in potential destinations have significant ments need to be developed. While the physi- impacts on migration destination, essentially cal characteristics of forest lands (e.g., slope) supporting the view that improved economic or existing population density are not affected conditions constitute the key attraction to by policy (at least in the short term), the prop- migrate to an upland destination. This sug- erty rights governing forest lands can be gests that economic incentives also operate on modified substantially. Regarding forest land the origin side, in a complementary man- management reforms, our results support ner-i.e., higher poverty incidence motivates current discussions on the need to devolve out-migration.8 forest land management, from central govern- In addition to this result on economywide ment to community or user groups. This economic factors, location-specific income would change the open access conditions in data allow an assessment of the role of more many areas and would contribute to reducing "targeted" policy instruments. From the mi- migration pressures on these lands. This gration model results, improving incomes in analysis is relevant for articulating an agenda existing upland communities (a good goal) for effective reforms. The results confirm that leads to more upland migration (an unwel- forest sector-specific efforts to create alterna- come effect). To respond to this apparent tives to ineffective centralized management dilemma, it is important to incorporate the and to provide technological support in stabi- environmental factors in the evaluation of lizing agriculture in steep lands should con- economic policy options. For example, the tinue to be pursued. However, because of the analysis of environmental factors in the model extent of external pressures on forest lands, economic incentives-both economywide and sectoral-underlying migration pressures on forest lands must similarly be addressed. 'Because our migration model focuses on the choice among alternative upland sites (and not on whether or not to migrate), the results do not allow us to make direct conclusions on the role of poverty and income in the decision to migrate. 88 VOLUME 1I Table 4-1: Land Use Classification in the Philippines, 1988 Land use/cover Area (000 ha) Percent of total A. Forest 7,226 23.92 Pine 81 0.27 Mossy 246 0.81 Dipterocarp 6,629 21.94 Closed 2,435 8.06 Open 4,194 13.88 Mangrove 149 0.49 Other 121 0.40 B. Extensive cultivation 11,958 39.58 Open in forest 30 0.09 Grassland 1,813 6.00 Mixed grass, brush plantation, etc. 10,114 33.49 C. Intensive cultivation 9,729 32.58 Plantation 5,336 17.67 Coconut 1,133 3.75 Other 91 0.30 Coconut, cropland 3,748 12.40 Other cropland 365 1.20 Cropland 4,392 14.54 D. Fishponds 205 0.67 Fishpond from mangrove 195 0.64 Other fishponds 10 0.03 E. Nonvegetated Areas 101.4 0.34 Eroded areas 0.7 0.002 Quarries 8.0 0.02 Riverbeds 81.0 0.27 Other barren land 10.0 0.03 F. Other 439 1.45 Built-up areas 131 0.43 Marshy areas 103 0.34 Lakes 205 0.68 G. Unclassified 546 1.80 TOTAL 30,205 99.96 4. Poverty, Migration, and Deforestation in the Philippines 89 Table 4-2. Philippine Forest Land Population, 1950-85 Forest land population Annual growth rate between Census years (thousands) centsus years (percent) 1950 5,868 - 1960 8,192 3.39 1970 11,169 3.15 1975 12,702 2.61 1980 14,440 2.60 1985 17,513 3.39 Source: M. C. Cruz et al. (1992). Table 4-3: Incidence of Poverty in the Philippines, 1971-88 (percent) Year Urban Rural 1971 38 58 1985 42 58 1988 48 60 Note: The income threshold for the mid-1980s was estimated to be P4,764 per person per year in 1978 prices. Source: 1971 and 1985 estimates from World Bank 1988; 1988 estimate from the Family Income and Expenditure Survey (1988). Table 4-4: Nominal Protection Rates for Main Agricultural Commodities, 1960--86 Averages (percent) Crop Direct NPR Indirect NPR Total Rice 8 -25 -17 Corn 39 -33 6 Sugar -18 -19 -37 Coconut -12 -21 -33 Note: Direct nominal protection rate (NPR) measures the proportional difference between a commodity's producer price and the border price due to direct government price subsidies or taxes. Indirect NPR measures the difference due to the indirect effect of industrial protection and local currency overvaluation. They then estimate the implication of total disincentives to total real income effects on the households dependent on these crops, for 1971-84. This indicator measures the impact of output and input price policies on value added of a sector, also incorporating the effect of price policies on cost of living. For the four crop sectors, the real income effects were all negative (even for corn-producing households), ranging from -2.49 to -3.61 percent. These were negative and large since agricultural disincentives meant that farmers not only earned less, but also had to pay relatively more for protected industrial goods (Intal and Power 1991). Thus, the effects of policy failures have meant pervasive poverty and economic stagnation. In particular, lowland agriculture was penalized heavily, causing a distinct rural bias in poverty incidence. Traditional agricultural communities suffered most. Coupled with rapid population growth, these conditions of poverty and unemployment have contributed to incentives for migrants to move to forest lands, in search of alternative livelihoods. 90 VOLUME 11 Table 4-5: Frequency Distribution of Migration Streams, M5j's, 1990 Census Cumulative Cumulative MA, Frequency Percent frequency percent 0 2,638 44.5 2,638 44.5 1-500 3,128 52.8 5,766 97.3 500-900 94 1.6 5,860 98.8 > 1,000 69 1.2 5,929 100.0 Note: Mij's refer to individuals in forest land municipalities in 1990 who resided in lowland municipalities in 1985. Table 4-6: Description of Migration Determinants and Hypothesized Link to Migration Description of upland migration factors Expected sign of migration effect Socioeconomic factors: Average household income. Mean household [+] Higher provincial income levels tend to income in province. Source: NEDA 1988. make a destination more attractive to upland migrants. Poverty incidence (%). Proportion of households [-] The more poverty there is, the more likely falling below poverty line. Source: National Sta- upland migrants will choose other locations. tistical Coordination Board 1985. Average upland income. Mean income of house- [+] Higher income in upland sites attract holds in upland municipalities in the province. more migrants. Source: LAST 1990a,b. Average lowland income. Mean income of house- [-] Higher income in lowland sites in the holds in lowland municipalities in the province. province makes upland sites less attractive. Source: LAST 1990a,b. Unemployment rate. Percentage of labor force [-] More unemployment will discourage up- that is unemployed in the province. Source: land migration. NEDA i985. Percent of population employed in agriculture. [+/-] The importance of agricultural employ- Source: NCSO 1980. ment may affect migration destination. Farm area tenancy/lease rate. Percent of farm [-] High rate of tenancy or leasing indicates area under tenancy or lease. Source: Bureau of tenurial insecurity and will discourage upland Agricultural Economics 1980. migration. Index of cost of long-range migration. Index = 1 [-] Within province, movements are less if migration is to municipality in other province, costly; this encourages migration to upland 0 otherwise. Source: NCSO 1990. sites within the province. Urbanization rate (%). Percent of Urban to Total [+] Urbanization is associated with greater Population. Source: NCSO 1980. economic opportunities and social services, thus attracting migration. 4. Poverty, Migration, and Deforestation in the Philippines 91 Description of upland migrationfactors Expected sign of migration effect Literacy rate (%). Percentage of population that [+] High literary rate is an indicator of socio- is literate. Source: NEDA 1990. economic well-being and increases likelihood that site will be chosen by migrants. Environmental factors: Cropped area (sq km). Total cropland in province [-] Indicator of lowland agricultural orienta- in 1987. Source: NEDA 1987. tion of province, thus less attractive as upland migration destination Arable land (sq km). Total arable land in prov- [-] Indicates importance of lowland resources ince, 1987. Source: NEDA 1987. versus uplands in the province. Percent of land under extensive cultivation. Ex- [+I-] A high percentage of extensively culti- tensively cultivated land includes: cultivated and vated land could indicate crowding in the other open areas, grass land, cultivated area province, thus discouraging migrants, in gen- mixed with brush and grass land. Source: World eral. Alternatively, it could indicate more Bank 1989b. developed markets and land accessibility, which means increased opportunities for up- land migrants. Percent of land under intensive cultivation. Inten- [+1-] Presence of intensive cultivation could sively cultivated land includes: coconut and other discourage more migrants. However, it could plantations, cereal crop and sugar lands, combi- also mean more accessibility and developed nation of crop land and coconut/other plantations. markets. Percentforested land. Proportion of provincial [+] Greater proportion of land under forest area under forest cover. Source: World Bank will attract more upland migrants. 1989b. Land area with greater than 30 percent slope [-] Steeply sloping lands will tend to be less (%). Calculated from data provided by the Bureau productive and will discourage upland mi- of Soils and Water Management. grants. Publicforest land (% of provincial area). In- [+] Proportion of land area governed as pub- cludes all land, whether forested or not, classified lic forest land. Because of ineffective public as public forest. Source: Forest Management management, such lands are mostly de facto Bureau 1989. open access and attract migrants. Road Density in Upland Area (km/sq km). Kilo- [+] Presence of roads in forest land indicates meters of roads per square kilometer of area of easier access and improved profitability of upland municipality. Source: Department of Pub- farming, thus attracting migrants. lic Works and Highways, National Statistics Of- fice 1987. Demographic factors: 1980 lowlandpopulation. Provincial population [+/-] Having large lowland population may less upland population in 1980. Source: 1980 increase attraction of upland destination, but Census and estimates of 1980 upland population it could also have a negative crowding effect. from M. C. Cruz, Zosa-Feranil, and Goce 1988. 92 VOLUME!! Description of upland migration factors Expected sign of migration effect 1980 upland population. Population in munici- [+] Large upland population indicates pres- palities classified as upland. Source:M. C. Cruz, ence of established communities, which at- Zosa-Feranil, and Goce 1988. tract migrants. Upland population density. Upland population [-] High upland density would limit land per square kilometer of upland municipalities, available to new migrants. 1980. Source:M. C. Cruz, Zosa-Feranil, and Goce 1988. Population growth rate 1980-90. Calculated [+] Perceived growth in potential destination from 1980 and 1990 Censuses. would attract migrants. 4. Poverty, Migration, and Deforestation in the Philippines 93 Table 4-4 ultinomial Discrete Choice Estimates of Determinants of Upland Migration Migrati Aactors (expected link to upland migration) Coeff icients Standard errors Economic factors: Average household income 3.97** 1.465 Poverty incidence (%) -3.74** 1.216 Average upland income 0.208** 0.058 Average lowland income -0.038 0.067 Unemployment rate (%) 18.87 20.68 Percent of population employed in agriculture -1.966 2.137 Farm area tenancy/lease rate 0.11 0.26 Index of cost of long-range migration -33.03** 1.01 Urbanization rate (%) -0.32 0.427 Literacy rate (%) -11.53 20.49 Environmental factors: Cropped area (sw km) -1.54* 0.922 Arable land (sq km) 0.519 0.614 Percent of land under extensive cultivation 0.494 0.416 Percent of land under intensive cultivation 1.081 ** 0.438 Percent forested land 0.906 0.77 Land area with greater than 30 percent slope (%) 0.352 0.255 Public forest land (% of provincial area) 1.05** 0.322 Road density in upland area (km/sq km) 1.27** 0.468 Demographic factors: 1980 lowland population 3.55* 1.929 1980 upland population 0.78** 0.302 Upland population density -1.63 1.20 Population growth rate 1980-90 1.37** 0.34 * denotes significance at 0.10 Ilevel. ** denotes significance at 0.05 level. Notes: Log likelihood is -292.39; F statistic is 0.0033. All variables refer to destination province, unless noted otherwise. Index of Long-range Migration = l, if upland migration is to another province; 0, if within province. 94 VOLUME II Figure 4-1: Expansion ofAgriculture in Lowlands and Forestlands, 1960-87 14 12- 10- F~ D * 3Forest lands F Lowland 4- 2- 1960 1971 1sso 1987 Year of inventory 4. Poverty, Migration, and Deforestation in the Philippines 95 Annex 4-A: Multinomial Discrete Choice Model of Upland Migration Our discussion of upland migration suggests that individuals in either lowland urban or lowland rural provinces face a large set of potential upland forest destination provinces. However, some migration occurs between upland provinces for households displaced from their illegal land claims, or that leave after their land becomes unproductive. In this section, we describe an econometric procedure that accommodates the many choices facing the potential migrant, and we show how the model can apply to observed migration streams. An individual's decision to move is a complex comparison of (known) utility the individual receives at his origin, and the utility he expects to receive at potential destinations. Since each individual can physically move to only one site at one time, the potential destinations form a discrete set of alternatives. Each migrant in the same location thus faces the same alternative set. The aggregation of similar choices made by individuals at an origin makes possible the application of the discrete choice framework to typical data sets, where mnigration streams (e.g., the number moving from one location to another location) rather than individual data are available.9 Assume there is a representative individual, person i, residing in province i, who receives the following utility: v(X;Qj)= V:-- (1) where Vio is individual i's known utility at his/her location, Xi is a vector of variables specific to province i, and D, are demographic characteristics of individual, as well as important environmental variables that affect utility. We assume aV(X1;i)/dXj 2 0 and eV(Xj;Qj)/d0X7 0. This ith individual will migrate from province i to province j only if j is preferred to all other m alternatives: 10 T,/(Xj;Q) > EtV*(Xm.;d2)j, Vm ,j. (2) where E[Vi(Xm;7Qm)) represents utility the ith individual expects to receive at site m, and the total number of possible destinations is denoted by N. If the individual chooses to stay in location i and not move, then Vj(Xj;Qi) > EjVj(X,;Qk)j, Vk ,i. 3 If the choices in Equations (2)-(3) are observed for each individual in destination i, then a "migration stream" can be defined that describes aggregate migration. The migration stream from origin site i to destination site j,denoted M1j, is the number of individuals in province i for whom condition (2) is true. By aggregating these individuals into a common stream of migrants, we necessarily assume that they are "identical"; that is, each migration stream can be described by the decision facing a representative individual. This representative individual assumption is conventional 9That is, the polychotomous choices faced by each individual are translated into similar "choices" for migration streams. "'The assumes the individual moves to only one location at a time. 96 VOLUME II in discrete choice models like (1)-(2), where individual-based data is not available (Ben-Akiva and Lerman 1985)." Any environmental, income, or demographic variables that make an individual's expected utility higher at a given location will determine their destination site choice. Moreover, if each individual moves only once in the time frame covered by the model, then the arguments of the utility function can be defined so that they vary only over destinations. In other words, for each M,i the origin variables become constant for all individuals composing Mi, thus allowing aggregation. To obtain an estimable form of the migrant's decision, we must convert expected utilities into random utilities. Using random utilities accommodates the fact that each migrant's expectations of their utilities at the destination as well as their origin are known to the individual but not to the researcher. Random utilities are obtained by appending error terms to the site-specific utility functions in (I)-(3). Following convention, we assume that random utility for each individual is monotonic and separable in its variables (McFadden 1987; Hausman and McFadden 1984), so that for any potential destination m for the ith individual: V/Xm;Qm) + em = ao + A3X,m + + BkXkt + YD,] + ... + rkfk,(5) where X,, are destination specific variables, e, is a utility-specific error term with probability density function, f,, £ are demographic indicators, and A, 6 k,,k yi,... yk are constants to be estimated. Note that by appending the demographic variables in a linear fashion, we approximate their effect in the utility function, as others have suggested (e.g., Pudney 1989). Using (5) in (I)-(3), the probability that migration from i to j occurs for the ith individual is py = pr[V,(CX;D,) + e, > V,(Xi;Q,) + 'e (6) This probability can be estimated using either individual data, where choices are represented as (0,1) binary variables, or using sample proportions.'2 Sample proportions are especially well suited for large-scale population studies where individual decisions are not observed or aggregated (Ben-Akiva and Lerman 1985, other cites). In the above model, the sample proportion defines the frequency distribution of migration patterns. 3 The proportion is computed for each origin i and destination j: Pj = M/E1Mij. (7) Using pj, the choices in (6) are represented as frequencies that must sum to one when added over choices for each migration stream. "The representative individual assumption is essential in this chapter given the scale of the problem we investigate and the availability of data. For example, we later estimate migration for over 5,000 possible choices and 10,000 migrants. Like other countries where such data are compiled, the census data available for such a study in the Philippines is not individual-based. '2The assumption required is that individuals making the same choice can be summarized by a representative individual. '3This assumes, as is the case for our data, that observations are collected randomly. 4. Poverty, Migration, and Deforestation in the Philippines 97 The error in (6) has an extreme value distribution. Assuming Ei are iid normal random variables, a maximum likelihood procedure can be used to determine how the probability of observing a migration pattern from site i to destination j depends on the vector of explanatory variables in the utility function, (5). From the maximum likelihood estimates, one can then determine how migration patterns are affected by the explanatory variables using the standard marginal effects computation. 14 The random utility decision together with (1) and (4) comprise a "multinomial discrete choice model" (Amemiya 1985). An important feature of discrete choice models is that they accommodate decisions made simultaneously about multiple destination sites. Statistical tests can also be conducted regarding the relevance of explanatory variables that determine the observed migration patterns. "In particular, cp,/cXj = ...-a.. where A is the ML estimate of the coefficient associated with Xk, from (5). 98 VOLUME II Annex 4-B: Descriptive Statistics of Explanatory Variables in the Migration Model Table 4B-1: Descriptive Statistics of Explanatory Variables in the Migration Model Variable Xi MEAN STD DEV MIN MAX Percent upland migration Y 0.23 0.17 0.04 0.79 Upland population density Xlj 38.41 6.74 6.60 73.96 Arable land in i (sq km) X2i 1,148.44 467.25 247.00 2,363.00 Percentlandareawith30per- X3j 38.70 18.14 5.79 69.19 cent slope in provincej Land area classified as public X4j 328,097.00 216,105.00 40,422.00 1,041,850.00 forest (ha) in j Percent land under extensive X5j 38.85 12.43 16.00 66.00 cultivation inj Percent of total farms in i under X6i 24.75 11.37 2.48 50.43 tenancy/leasehold Unemployment rate in province X7i 6.16 2.75 2.00 11.90 i(%) Percent literacy rate in i X8i 89.60 8.30 59.77 97.84 Urbanization in j X9j 0.32 0.22 0.06 0.99 Lowland income in i XlOi 40,078.00 20,126.00 11,722.00 94,763.00 Upland income inj Xllj 25,919.00 12,382.00 6,518.00 60,042.00 4. Poverty, Migration, and Deforestation in the Philippines 99 Bibliography Amemiya, T. 1985. Advanced Econometrics. Rural Sector. Manila: Philippine Institute Cambridge, Mass.: Harvard University for Development Studies. Press. Family Income and Expenditure Survey Ben-Akiva, M., and S. Lerman. 1985. Dis- (FIES). Manila: National Economic and crete Choice Analysis: Theory and Appli- Development Authority. cation to Travel Demand. MIT Press Food and Agriculture Organization (FAO). Series in Transportation Studies. Cam- 1983. Tropical Forest Resources Assess- bridge, Mass.: MIT Press. ment Project Forest Resources of Tropical Boserup, E. 1965. The Conditions of Agricul- Asia. Rome: Food and Agriculture Organi- tural Growth. London: Allen and Unwin. zation. Bureau of Agricultural Economics. 1980. Forest Management Bureau. 1989. Philippine Agriculture Census. Manila: Bureau of Forestry Statistics. Quezon City: National Agricultural Economics, Department of Economic and Development Authority. Agriculture. Hansen, S. 1990. "Macroeconomic Policies Bureau of Forest Development. 1985. Manila: and Sustainable Development in the Third Department of Environment and Natural World." Journal of International Develop- Resources. ment 2(4). Cleaver, K. and G. Schreiber, 1991. The Pop- Hausman, J., and D. McFadden. 1984. "Spec- ulation, Environment and Agriculture ification Tests for the Multinomial Logit Nexus in Sub-Saharan Africa, Africa Model." Econometrica 52(5):1219-1240. Region Technical Paper, World Bank, Intal, P., and J. Power. 1991. "Trade, Ex- Washington, D.C. change Rate, and Agricultural Pricing Cruz, M. C., C. A. Meyer, R. Repetto, and R. Policies in the Philippines." In A. Woodward. 1992. Population Growth, Krueger, M. Schiff, and A. Valdes, The Poverty, and Environmental Stress: Fron- Political Economy of Agricultural Pricing tier Migration in the Philippines and Policies. Costa Rica. Washington, D.C.:World Institute of Agrarian Studies (IAST). 1990. Resources Institute. Terminal Report: Benchmark Survey of Cruz, M. C., I. Zosa-Feranil, and C. L. Goce. Potential CARP Areas. Los Bahos, 1988. "Population Pressure and Migration: Laguna, Philippines: University of the Implications for Upland Development in Philippines at Los Banios. the Philippines." Journal of Philippine . 1990. Unpublished information based Development XV(I). on a survey conducted in 1990. Los Cruz, W., and C. Gibbs. 1990. "Resource Bafios, Laguna, Philippines. Policy Reform in the Context of Popula- Krueger, A., M. Schiff, and A. Valdes. 1991. tion Pressure: The Philippines and Nepal." The Political Economy of Agricultural American Journal of Agricultural Eco- Pricing Policies. Baltimore: Johns nomics 72(5). Hopkins University Press for the World Cruz, W., and R. Repetto. 1992. The Environ- Bank. mental Effects of Stabilization and Struc- Kummer, David. 1990. "Deforestation in the tural Adjustment Programs: The Philip- Post-War Philippines." Ph.D. diss., pines Case. Washington, D.C.: World Boston University. Resources Institute. McFadden, D. 1987. "Method of Simulated David, Cristina, W. Cruz, and associates. Moments for Estimation of Discrete Re- 1986. AgendaforActionfor the Philippine sponse Models with Numerical Integra- tion." Massachusetts Institute of Technol- 100 VOLUME HI ogy, Department of Economics. Working ton, D.C., and New York: World Bank Paper No. 464 (August), pp. 1-47. and Oxford University Press. Mink, S. 1993. "Poverty and the Environ- Pudney, S. 1989. Modelling Individual ment." Finance and Development 30(4): Choice: The Econometrics of Corners, 8-10. Kinks, and Holes. Oxford, U.K., and New National Census and Statistics Office York: Blackwell. (NCSO). 1980, 1990. Census of the Popu- Williamson, J. 1988. "Migration and Urbaniza- lation of the Philippines. Manila: National tion." In H. Chenery and T. N. Srinivasan Economic and Development Authority. (eds.), Handbook of Development Eco- . 1985. Barangay Census. Manila: nomics, Vol. 1. New York: Elsevier Sci- National Economic and Development ence Publishing Co., pp. 424-65. Authority. World Bank. 1985. The Philippines: Recent National Economic and Development Author- Trends in Poverty, Employment and ity (NEDA). 1985, 1987, 1988, 1990. Wages. Washington, D.C.: World Bank. Philippine Statistical Yearbook. Manila: . 1988. The Philippines: The Chal- National Census and Statistics Office. lenge of Poverty. Washington, D.C.: National Statistical Coordination Board World Bank. (NSCB). 1985. Philippine Statistical . 1989a. Philippines: Environment and Yearbook. Manila. Natural Resource Management Study, The Philippines Natural Resources Account- Washington, D.C.: World Bank. ing Project. 1992. Technical Resources . 1989b. SPOT Satellite Surveys 1987 Program, Project No. 498-0432. Prepared of the World Bank. by the International Resources Group, . 1992. World Development Report Ltd., in association with the Mandala 1992: Development and the Environment, Agricultural Development Corporation New York: Oxford University Press. and Edgevale Associates under Contract World Commission on Environment and No. 492-0432-C-00- 1015-05 with USAID Development (WCED). 1987. Our Com- Philippines, Manila. mon Future. Oxford, U.K.: Oxford Uni- Pearce, D. W., and J. J. Warford. 1993. World versity Press. Without End: Economics, Environment Yap, Lorene. 1977. "The Attraction of Cities: and Sustainable Development. Washing- A Review of Migration Literature." Jour- nal of Development Economics 4:239-64. 5 Economywide Policies and the Environment: A Case Study of Poland Robin Bates, Shreekant Gupta, and Boguslaw Fiodor Overview IN EXAMINING THE LINKAGEs between economywide policies and environmental degradation in formerly planned economies, we focus on Poland for two reasons. First, Poland is representa- tive of the challenges that all former socialist countries of Central and Eastern Europe (CEE) face as they attempt to restructure their economies. Not only was it one of the pioneers in the dramatic changes we now witness in CEE, but Poland also has now had substantial experience with the transition toward a market economy. CEE comprises Bulgaria, the Czech and Slovak Federal Republic (CSFR), the former German Democratic Republic, Hungary, Poland, Romania, and the former U.S.S.R. In the subsection, Energy intensity and the environment in Poland, references to CEE exclude the German Democratic Republic. 101 102 VOLUMEII In addition, Poland demonstrates clearly the We believe that the Polish experience disastrous environmental legacy of central contains important lessons for all centrally planning: in 1988-89 it was the third largest planned economies that are making the transi- emitter of sulfur dioxides (SO2) and nitrogen tion toward a market economy. It also rein- oxides (NOJ) in Europe (after the former forces the more general lesson-that competi- Soviet Union and Germany); with regard to tive markets and clear property rights can lead particulate emissions it ranked second to substantial environmental benefits, (Nowicki 1993); and SO2 and NO, emissions per dollar of GNP were highest among all Environmental Effects of a countries of CEE, including the former Soviet Centrally Planned Economy: Union (WRI 1992). A Retrospective View With respect to air quality, the concentra- tions of S02, NO,, and particulates were General features of a socialist economy significantly higher than permissible standards in all big Polish cities (Nowicki 1993). In the Until the 1980s, the Polish economy was heavily polluted Katowice region, for in- dominated by central planning, and shared the stance, in 1989 average annual concentrations main features of other socialist economies: of SO2 and NO, exceeded permissible stan- extensive state ownership of the means of dards by a factor of two, while particulates production, and a centralized mechanism for exceeded standards by a factor of four (Bates, the allocation of scarce resources. Such econo- Cofala, and Toman 1994)-well beyond the mies have been analyzed by Kornai and oth- guidelines proposed by the World Health ers,2 who argue that these economies are Organization (WHO).' resource constrained, in contrast to capitalist The primary focus of this case study is on economies, which are demand constrained. the impact of energy production and consump- Resource constraints arise due to the simulta- tion on air pollution, although more efficient neous existence of taut plans and soft budgets. production and consumption of energy will A taut plan implies physical targets that are have a beneficial impact on other media as rigid or binding enough to force the firm to hit well. The section, Environmental Effects of a a resource constraint. Thus, in the Kornai Centrally Planned Economy: A Retrospective framework, a firm would produce more output View, examines historically the ways in which if additional inputs were available. At the central planning led to environmental degrada- same time, budgets are soft, i.e., if inputs are tion in Poland, in order to isolate the policy available, funds to buy them can usually be changes that are necessary in the future. Given obtained. Any financial losses are covered the fundamental role played by a perverse through subsidies, and firms are never bank- incentive system, the section, Environmental rupt. Effects of the Transition to a Market Econ- The softening of budget constraints implies omy: Challenges for the Future, discusses a basic change in the system of managerial how Poland is restructuring economic incen- incentives, compared with a market-based tives in general, and analyses the impact on economy, since profit-seeking and cost-mini- the energy sector and air pollution in particu- mizing behavior are not rewarded. Two conse- lar. quences in particular are underlined here: the lack of responsiveness to input prices, and the 'The permissible annual average concentration in Poland for SO2 is 32 pg/rn3, and for NO, and 2Kornai's analysis was articulated in Economics particulates, it is 50 pg/m3 each. The WHO guidelines of Shortage (1980). It has since been refined and suggest upper limits for SO2 and particulates of 60 and restated by Kornai and others. For a review, see Kornai 90 gg/m3, respectively (UNEP 1992). (1992). 5. Economywide Policies and the Environment: A Case Study of Poland 103 creation of excess demand. With respect to the pute various measures of material intensity for former, a downward-sloping demand curve by eleven capitalist and four socialist countries the firm for its inputs presupposes the exis- and find the latter to be considerably more tence of a hard budget constraint. However, material intensive, in particular, more energy the softer the budget constraint, the weaker the intensive.3 In another cross-country study, compulsion to adjust demand to relative Slama (1986) also finds higher energy inten- prices. In the extreme case of a perfectly soft sity of production in socialist countries. More budget constraint, own-price elasticity of recently, Moroney (1990) simultaneously demand is zero: the demand curve is vertical, considered energy and physical capital as i.e., determined by other explanatory variables aggregate inputs. He concluded that CMEA and not by the price. It is important to note countries as a group consumed, on an average, that this is not merely a matter of a firm react- approximately twice as much energy per unit ing less elastically to some particular price or of capital and per unit of real GDP as the the other. "The softness of the budget con- economies of Western Europe. Further, while straint blunts responsiveness to every price" both sets of countries relied on capital accu- (Komai 1992, p. 146). For example, if one of mulation and increased energy consumption as two inputs that are perfect substitutes becomes sources of economic growth, the CMEA cheaper, it is far from certain that the firm will countries relied to a greater extent on energy. consider this in selecting its input mix: it is In the case of Poland, the material intensity not price, but other considerations, such as inherent in all socialist systems has been output maximization, that drive the demand exacerbated by abundant and cheap availabil- for inputs. The implication of this in the ity of coal, resulting in particularly high levels context of excessive energy use in Poland is of energy intensity. In contrast, in a market that price reform in itself may not achieve a economy like Australia's, which exports three reduction in energy use: it has to be accompa- times as much coal as Poland (U.S. EIA nied by a hardening of the budget constraint, 1990), coal accounted for only 44 percent of i.e., enterprise reform or privatization. primary energy supply (British Petroleum Second, a combination of soft budgets and Company 1992). Thus, it appears that an taut plans results in persistent shortage or abundant energy supply is necessary but not excess demand. Firms in a resource sufficient for high energy intensity: a constrained economy are not cost minimizers, resource-intensive economic system has to but rather output and resource-use exist as well. maximizers. If the budget constraint were hard, demand for inputs would be limited, Energy intensity and the environment in since expenditures on purchasing inputs Poland would be conditional on revenues generated The energy sector plays a major role in the by the sale of output, which in turn is con- strained by the demand for the firm's output. e nomy of Pln Witrim energy Further, output maximization is the firm's sequvalent e s 1990, mt was the second largest paramount goal, since it is usually the only equivalent i10tw the seon lEst criterion~~~~~~~ bywihispromnei.ugd energy sector among the countries of CEE, critrio bywhih is prfomane i juged after the former Soviet Union. It iS also one of Thus, it tries to stockpile inputs in order to be able to produce without interruption. The only constraint is the actual availability of inputs. Empirical evidence on the material inten- 3Aggregate energy consumption per unit of value added in the industrial sectors of the seven European sity of socialist economies is provided by CMEA countries was found to be 40-134 percent Gomulka and Rostowski (1988). They com- higher than that in the 15 Organisation for Economic Co-operation and Development (OECD) countries. 104 VOLUME II the largest sectors of the Polish economy. In Note that the figure for Poland is close to the 1990 it contributed 13 percent of gross domes- CEE5 mean of 2.0. What is notable about tic product (GDP), employed 5 percent of the Poland, however, is the dominance of coal labor force, and between 1986 and 1990 (often referred to as the "coal monoculture") accounted for 30 percent of the total invest- in the consumption of primary and final en- ment in industry (World Bank 1993b, Vol. 1). ergy. In addition, coal accounted for over 95 While several cross-country comparisons percent of domestic primary energy produc- bear out the energy intensity of the Polish tion in Poland in 1990. Finally, Poland was economy (defined as energy consumed per the world's fourth largest exporter of hard unit of output),4 two problems with national coal, and coal exports accounted for 12 per- income data need to be kept in mind. The first cent of export earnings in 1990 (World Bank is that official figures for centrally planned 1993b, Vol. 2). Thus, the major environmental economies suffer from conceptual limitations problems in Poland are those connected with and poor accuracy. Second, since official energy in general, and the excessive use of exchange rates for these economies are arbi- coal in particular, which is a dirty fuel to mine trary, comparable estimates of national in- and to burn. come are calculated either by using commer- Energy production and use in Poland in cial exchange rates, or by using a Purchasing 1990, by stationary sources (mainly for pro- Power Parity (PPP) approach. Since the latter ducing electricity and heat), accounted for results in income figures that are higher, using 70-80 percent of the emissions of particulate PPP measures of national income reduces matter (PM) and SO2, and nearly 50 percent of energy intensity statistics. the emissions of NO, (Bates, Cofala, and Table 5-1 reports energy intensity of na- Toman 1994). Within the energy sector, coal tional income, based on PPP. In addition to contributed 82 percent of PM, 90 percent of the level of energy intensity, an important SO2, and 46 percent of NO, emissions (Bates, aspect is its behavior over time. Thus, while Cofala, and Toman 1994). An important energy intensity in Poland has declined, the source of this contribution to PM is the "low- rate of decline has not been as rapid as in stack" sector, i.e., households, commerce, and other countries. Moreover, the rate of decline services, which, in 1987, relied on coal for has slowed down over time (Table 5-1).5 more than 60 percent of its total fuel use With respect to energy intensity, it is im- (Hughes 1991). This figure is high, not only in portant to note, however, that Poland is not an comparison to Western European countries, outlier compared with other parts of CEE. but also other parts of CEE (see, for instance, Table 5-2 shows energy intensity per unit of Hughes 1991, Figure 7). output for CEE (relative to OECD) and also The low-stack sector includes district for the five smaller CEE countries as a group, heating6 enterprises, which "use coal as their excluding the former Soviet Union (CEE5). primary energy source, consuming 24 million tons of coal products-13 percent of the coun- try's output-in 1989, and creating 11 percent 4See for instance, Bates and Moore (1992), Table 4, .of tional sulfu ode *emsins lIper- and IEA (1990), Figure 2. of national sulfur dioxide emissions, 10 per- cent of particulate fallout, and 8 percent of 5This is confirmed by fitting a quadratic time trend to annual energy intensity data from 1960-91. While the coefficient on the trend variable was negative and 'District heating is formally defined as "a heating significant, the coefficient on the square of the trend system that transmits and distributes heat from one or variable was positive and significant. Thus, energy more energy source to, among others, residential, intensity has declined, but the rate of decline has slowed industrial and commercial consumers for space heating, down over time. Results are available from the authors cooking, hot water and for industrial purposes" (IEA on request. 1992, p. 378). 5. Economywide Policies and the Environment: A Case Study of Poland 105 nitrous oxide emissions. Because so many of these categories is whether an economic entity the sources of that pollution are low-stack, is public or private. The crucial difference small boilers in urban areas, the subsector has between the two is that the latter have rela- a disproportionate impact on local air quality" tively hard budgets. From this perspective, the (World Bank 1993b, Vol. 5, p. 2). Localized nature of economic activity undertaken by emissions from the low-stack sector cause these entities is not particularly relevant. Thus, greater health and property damage than households and private firms are both classi- power station emissions, which are dispersed fied as private entities, since both face hard from high stacks. budgets. In socialist countries, however, Other pollutants related to the mining and private entities contribute a relatively insignif- use of coal include solid waste from coal icant share of total output: in Poland, the mines and fly ash disposed on land, and salted public sector, broadly defined to include mine water discharged into rivers. Other government, accounted for more than 80 ecological impacts of coal mining consist of percent of national output in 1985 (Milanovic depletion of ground water, subsidence of land, 1989). Also with respect to employment, the and loss of topsoil through strip mining. public sector accounted for more than 70 percent of total employment in Poland in 1985 Patterns of energy use (Lipton and Sachs 1990). Using 1990 data, the primary energy supply Public entities and energy intensity in Poland went to the following uses (see Figure 5-1): 32 percent went to the transfor- Public entities in Poland that exceeded their mation and energy sector, where it was con- budgets could expect to receive regular exter- verted into heat and electricity. Direct con- nal assistance. Some discussion of the ways in sumption of primary energy by final consum- which this can occur are in the section, Envi- ers accounted for 62 percent, and the remain- ronimental Effects of the Transition to a Mar- ing 6 percent was nonenergy use (based on ket Economy: Cliallenges for the Future, but World Bank 1993b, Vol. 1, Fig. 2).7 The final the mechanism is not relevant here. What is consumers of primary energy, in turn, were important, however, is the consequences, dominated by industry (38 percent), and com- namely, firms had no incentive either to mini- mercial and agricultural uses (50 percent), mize input use for a given technology, or to with the residential and transport sectors use more efficient technologies in the produc- accounting for 6 percent each. Of the total tion process.8 These points were empirically primary energy used by the transformation and validated in a World Bank study that surveyed energy sector, the breakdown was (a) district the likely responses of industrial consumers of heat 15 percent, (b) combined heat and power energy in Poland, to, inter alia, an increase in generation 56 percent, (c) own use apd losses energy prices. The results suggested that "in 25 percent, and (d) gas and oil production 4 the short to medium term, these changes percent. would have little impact on either the pattern For an understanding of the causes of of industrial fuel use or the general level of energy intensity in Poland, the breakdown of fuel efficiency" (World Bank 1991, p. 16). In energy use is better grouped into two analyti- retrospect, a perverse justification for low cal categories. The key attribute for defining input prices in a "cost plus" pricing scenario was precisely this: increases in prices of key inputs, such as energy, would have been 7In addition to direct consumption of primary energy, final consumers also receive energy, mainly heat and electricity, from the transformation and energy 'See the subsection, General features of a sector. socialist economy, above. 106 VOLUMEII passed on via higher output prices, causing cally enriched, mainly for coal exports.'2 inflation without diminishing energy Technological improvements were particularly intensity.9 limited in activities that related to the environ- In general, the Polish development process ment: fluidized bed combustion of coal was from the 1950s to the 1980s was dominated by not started until 1989, no flue gas extensive rather than intensive growth. In desulfurization facility is operational to date, other words, increases in output were accom- and only one hard coal mine is equipped to panied by increasing use of factors of produc- desalinate mining water. tion, particularly energy, and not by increasing An additional influence on productivity is productivity. The fact that extensive growth worth noting in the case of public entities that was the dominant form of output growth in are suppliers as well as consumers of energy, Poland has been noted by several studies, such because of the revenue implications of low as Fallenbuchl (1977) and Kemme and Crane energy prices. Given the financial constraints (1984). The latter use a Cobb-Douglas produc- imposed by low prices, energy suppliers tend tion function to estimate the relative contribu- to concentrate their limited resources on tions of capital, labor, and imported inputs to quantity throughput, rather than investing in the rate of growth of industrial output in maintenance and other measures that reduce Poland from 1971 to 1981.10 They find that costs and/or increase efficiency. Hence, pro- "the rapid expansion of output in the 1970s ducer price controls result in greater input use was the result of extensive growth in factor and reduced maintenance, further aggravating inputs, especially capital services"" (Kemme high energy intensity." We provide illustra- and Crane 1984, p. 35). tions here from the coal mining and energy In terms of specific examples of the ab- conversion sectors. sence of intensive growth, in the 1960s, only Although the share of coal in primary 4.2 percent of the coal extracted was mechani- energy consumption declined from 95 percent in 1960 to approximately 78 percent in 1991, it is still the dominant category. 14 Against this 9Artificially low prices do matter, however, from an backdrop of "coal monoculture," the coal overall macroeconomic perspective, since they result in industry itself was characterized by declining subsidies that create budgetary pressures. Thus, higher labor p tivit an gro rter ial energy prices would have eased budgetary pressures labor productivity and growing raw material and/or some of these revenues could have been used to and energy intensity. The energy intensity of alleviate environmental problems. Another area where coal extraction (hard coal and lignite) in- low energy prices would matter is energy consumption creased by approximately 31 percent between by private entities, discussed in the subsection, Private 1978 and 1985; and the self-consumption of entities and energy intensity, below. coal increased by about 20 percent over the "Quarterly data is used in the study, from the first same period (Bates, Gupta, and Fiedor 1994). quarter (Ql) of 1971 to mid-1981 (Q2). This period is also divided into two distinct subperiods: 1971Q1 to mid-1978 and mid-1978 to mid-1981. This division is 2This is a relatively simple and inexpensive made because, not only does the rate of growth of method of coal purification that increases its calorific output fall in the later period, but output actually value. declines. "3See Bates (1993) for a detailed discussion on "Of the 2.9 percent rate of growth in industrial the consequences of producer price controls. output between 1971 and 1978, about 1.6 percent was due to growth in services. Capital utilization in turn '4See Bates, Gupta, and Fiedor (1994), Table 5, depends, inter alia, on total energy consumption. The for the structure of primary energy use. The share of residual that measures productivity (among other coal in primary energy produced is greater than its share things), was consistently negative over the entire period in primary energy used, due to exports of coal and of the study. imports of oil and natural gas. 5. Economywide Policies and the Environment: A Case Study of Poland 107 Hard coal and lignite extraction, however, Poland is one of the most intensive users of increased by less than 7 percent during district heating in the world, and accounted for 1978-85.'5 Thus, the coal industry was no 27 percent of final energy consumed in 1990, exception to the hypothesis that soft budgets including industrial uses (lEA 1992). In terms lead to increased use of inputs, including of specific examples of inefficiency in the energy. district heating sector, there are significant The financial impact of the decline in labor water losses in many district heating systems productivity was exacerbated by increasing in Poland. It is estimated that water losses are wages. Between 1965 and 1989, wages in hard three to five times higher than those in West- coal mining increased continuously in real em Europe,"9 largely due to poor maintenance terms, and at a faster rate than other sectors of and corrosion (water is generally not deminer- the economy (Syzdlo 1990). Coupled with a alized or treated). Further, comparisons of policy of low coal prices, this implied a grow- heat sales with heat dispatches show theoreti- ing discrepancy between costs and retail cal heat losses of 10-45 percent, "a range that prices.'6 Between 1970 and 1990, the ratio of reflects the shortfall in metering equipment costs to prices in the coal industry increased that would enable distributors to identify from 90 percent to 132 percent; while the cost corroded and poorly insulated mains that leak of producing a ton of coal went up forty-three heat and water" (World Bank 1993b, Vol. 5, times between 1980 and 1989, the retail price p. 2). only increased thirty times.'7 This implicit subsidy contributed to budgetary deficits. Private entities and energy intensity Although we would argue that an increase in Since private entities in socialist economies coal prices would not have materially reduced face hard budget constraints, their behavior coal consumption, it would have reduced the should be similar to that of private agents in deficit; moreover, the excess revenues could market economies, except that they also face have been available to finance environmental quantity constraints and fixed prices; goods protection activities.qatt osrmsan ie rcs od protenergycconversion activitors. useathrd are rationed and sold at fixed official prices, The energy conversion sector used a third which are below market-clearing levels. of the primary energy produced in 1990. In Hence, a more rational energy use on the part this sector again, as in the coal sector, we of prnvate consumers could have been induced notice inefficiency in production: own use, of pri ce distcons. Re ened and transmission and distribution losses by reducing price distortions. Retail energy amounted to 25 percent in 1990. Further, prices, however, were consistently too low amouned to25 pecent n 199. Furher, relative to costs. Further, during the 1980s, between 1978 and 1985, the rate of increase in while retail prices for consumer goods in- final energy production was about half of that creased by 30-70 times, retail energy prices for primary energy.'8 increased only by about 9-10 times (Bates, Gupta, and Fiedor 1994). In any case, in an important part of the retail market, namely district heating, there '5Bates, Gupta, and Fiedor (1994), Tables I I(a) and were (and still are) technical impediments to I 1(b). using pricing incentives: residential consump- '6We refer to financial costs, as data on economic tion of secondary energy (heat and hot water), costs are not available. 17Bates, Gupta, and Fiedor (1994), Tables 12 and '9The figure for Poland is 100_150 m3 per 13. terajoule (TJ) of heat produced, compared with only 20-50 m3/TJ in Western Europe (World Bank 1993b, '8Bates, Gupta, and Fiedor (1994), Table 9. Vol. 5). 108 VOLUME Il showed the features of a public good. Resi- private entities in 1990 were at most 31.4 dential use of heat accounted for 10.6 percent percent of total final energy consumed.22 of final energy and 11.5 percent of primary energy consumption in 1990.20 Of the 7 mil- The pattern of industrial development lion dwellings in urban areas, 70 percent are suppiedwithhea forcenral eatng ad 5 isThe energy intensity of the Polish economy IS not only due to the general resource inten- percent with hot water (EA 1990). Nationally, sity of a socialist economy, as outlined above, 93 percent of the heat sent out by district but is also related to the strategy of industrial heating enterprises (DHEs), IS supplied to coatinsumnterspviasetwors, with trrests development. According to Kharas (1991), the consumers via networks, with the rest sup- shrofidtyinPld'GD in18 pliedby lcal oiler feeing ndiviual uild share of industry in Poland's GDP in 1989 was 51.7 percent, while the average for coun- ings. The heat distribution systems are con- . . . ' tries in similar income brackets and degree of stant vlm,full flow (i.e., the same volume ..23 stantvolume, fullflow(i.e.,thesamevolumurbanization was 39 percent. Given the of water is pumped irrespective of the heat emphasis on rapid industrialization in general, load). The flow temperature of the major heat and development of heavy industry in particu- sources is, however, adjusted according to lar, it is not surprising that four weather conditions by the heat despatcher sectors-energy, metallurgy, chemicals, and (World Bank 1993b, Vol. 5). minerals-accounted for over half of total Charges for heat and hot water were (and fixed capital stock in the entire industrial still are) proportional to the floor area of, and sector between 1960 and 1991. These sectors the number of persons in an apartment, re- also dominated industrial output, contributing spectively. It is obvious that this lump sum approximately 40 percent of the gross output pricing method did not encourage rational use i of~~~~~~~~~~~ ~ hetadhtwtr.urhr teaiiyo over the same period.24 They accounted for of heat adhtwtrFutr70-80 percent of energy used by industry consumers to control their heat consumption (Bates, Gupta, and Fiedor 1994, Table 17), is at present very limited" and "individual and their energy intensity was higher than control of heat consumption is generally industry as a whole (Bates, Gupta, and Fiedor limited to the use of manual radiator valves" (Word Bak 193b,Vol.5, p33)21 Finally, 1994, Table 18). As a result, they were the (World Bank 1993b, Vol. 5, p. 33)F most polluting sectors in industry: their share in addition to all these factors, apartments in particulate and SO2 emissions ranges from were poorly insulated (as discussed below). 80 to 94 percent (Tables 5-3 and 5-4, respec- Hence, it was only for the residual portion of tively). energy directly purchased by private entities, that price signals would have mattered. While these data are not available, an upper bound heavy industry is common to most centrally on them can be inferred from energy balance planned economies, in the case of Poland it statistics: direct purchases of final energy by 22For difficulties in measuring private sector energy consumption, see IEA (1992), p. 380. 20To arrive at the share of residential heat in primary energy, an efficiency factor of 0.62 was used 23Even the highly industrial middle income (personal communication from A. Umer, Polish nations of East Asia had only a 43 percent industry Academy of Sciences). share (Kharas 1991). 2"In contrast, in an experiment reported by 24Their share in net output, however, was 30 Wilczynski (1990), when households were provided percent or less (Bates, Gupta, and Fiedor 1994, Tables with measuring devices, hot water use declined by 15 and 16, respectively). This implies a larger than 50-60 percent, and power consumption for heating average share in depreciation, or a more antiquated (and purposes was lowered by 18-25 percent. polluting) capital stock. 5. Economywide Policies and the Environment: A Case Study of Poland 109 exacerbated the consequences of energy inten- We have argued that the "logic" of the sity. Other sectors also favored energy-inten- system created a demand for energy that was sive development, such as housing construc- constrained only by resource availability, but tion, where there was an emphasis on physical the ideology of energy security provided it quantities, by way of mass production of with an ideological justification. Since energy apartment buildings. Not only were these security and national security are often viewed apartments poorly insulated (and therefore as synonymous, the former can be an emotive consumed a lot of heat), but they also inten- issue, even in market economies. It was thus sively used raw materials such as steel and imperative to ensure adequate supplies of cement, which in themselves were energy energy for domestic needs which. as we have intensive. Construction of these buildings also seen above, were insatiable. entailed huge transportation costs, since large Finally, the role of grassroots environmen- parts of them were prefabricated in what could tal groups in forcing environmental improve- be called apartment factories. ments is well documented in the West. Given the absence of democracy in Poland, however, The political economy of energy intensity as in other communist countries, this avenue In addition to the causes of energy intensity did not exist at least until the 1980s. With that emerge from this analytical framework, respect to the decisionmaking process, as there were other factors at work that may be Wilczynski (1990) points out, "the centraliza- described under the rubric of political econ- tion of governmental decisionmaking weakens local authorities and enterprises. In environ- omy. They include the role of special interests (such as the Upper Silesian lobby and the coal mental matters this is very Important because and steel lobbies), ideological compulsions for the problems of pollution and degradation are energy security, and the lack of citizen pres- sure for a cleaner environment. With respect to the influence of lobbies, plan allocations were often distorted by spe- Environmental problems in Poland stem cial interest groups in the energy-intensive not only from high energy intensity, but also sectors. Thus, during the 1966-70 Plan period, from the failure of environmental policy. In 40 percent of the investment that was in ex- particular, in a centrally planned economy, the cess of plan outlays went to chemicals, hard efficacy of market-based approaches is seri- coal and metallurgy (Bates, Gupta, and Fiedor ously reduced, given the lack of price respon- 1994). In the 1970s, the concept of an "open siveness on the part of state-owned tirms, and plan" created numerous avenues of influence soft budgets. For example, while Poland for the lobbies to impose new projects, even implemented an apparently sophisticated completely new directions of investment. For system of pollution taxes-known as fines and example, the efforts of the metallurgical lobby fees-their basic rationale was questionable. resulted in construction of the Centrum steel Along these lines, Polish environmental ex- works, with a cost overrun of twenty times perts like Zylicz (1993) have argued that (Bates, Gupta, and Fiedor 1994). Similarly, "financial instruments were doomed to failure the fact that the Upper Silesian lobby held key in an economy where all essential inputs were political positions in the 1970s also spurred allocated administratively, and plant managers the development of the hard coal industry had little incentive to pay attention to price during that period. While initially it may have stimuli. As a result, instead of addressing the been the coal monoculture itself which gave real issues, environmental policymakers were rise to the coal lobby, subsequently, this inter- complaining that fees were too low, and poor est group became self-reinforcing. environmental performance was linked to an 110 VOLUME 11 inadequate level of these fees" (p. 7). Thus, all, investment outlays on environmental the questionable rationale of the whole system protection in Poland, up to 1990, ranged from of ecological fines and fees and other instru- 0.27 percent to 0.7 percent (Poland, Ministry ments of environmental protection in Poland of Environmental Protection, Natural Re- has to be viewed in the context of a lack of sources and Forestry 1992, and Gorka 1991). property rights, as well as the contradictory In 1991, however, this figure reached 1 per- role of the state, acting both as a producer and cent, a level comparable to that for OECD regulator of the environment. countries, and corresponds to an actual expen- One can, however, argue that if properly diture of US$840 million (Poland, Ministry of enforced and collected, fees and fines do at Environmental Protection, Natural Resources least generate revenue, which in turn can be and Forestry 1992). used for environmental protection activities. Thus, fees and fines can be viewed as transfer Environmental Effects payments, and this seems to have been their of the Transition role in Poland. Even viewed in these limited to a Market Economy: terms, environmental policy in Poland cannot Challenges for the Future be judged to have been a success. For in- stance, the expenditure on environmental The section, Environmental Effects of a protection in most industrial sectors amounted Centrally Planned Economy: A Retrospective to only 2-3 percent of production costs in View, argued that a central planning approach 1986; at the same time fines collected from to resource allocation, relying extensively on state-owned enterprises (SOEs) were less than physical planning targets, combined with soft- 0.1 percent of the profits, partly due to poor budget constraints, seriouisly damaged the enforcement (Bates, Gupta, and Fiedor 1994). environment in Poland. Those economywide What is also notable about Polish environ- policies resulted in the overuse of energy in mental policy is the lax enforcement of the general, and coal in particular, with no offset- command-and-control (CAC) measures. For .l.e instance, by the end of the 1980s only a third However, profound changes in economic of the plants enmitting air pollutants had emis- policy, which started to take place after 1988, sion permits. The corresponding figure for are expected to lead to major improvements in plants discharging sewage was 50 percent the environmental situation in Poland, even (Bates, Gupta, and Fiedor 1994). Thus, for without the implementation of specific envi- units without permits, the question of being in ronmental measures. Inter alia, these improve- violation did not even arise. In addition, fines ments will come through a further strengthen- that were imposed were often rescinded, and ing of the enterprise and banking sectors, and the fine collection ratio (fines actually paid the setting of energy prices, which reflect versus fines imposed) ranged around 40 per- economic costs. Not only will such changes cent.25 To make matters worse, standards were frequently "liberalized," which undermined result in the much-needed hardening of budget frequently "libralied,whihuconstraints and the consequent improvement the entire basis of the system of fines and fees. in resource utilization, the privatization of In general, environmental inspection and audit large segments of the Polish economy will capabilities were weak, and state and regional introduce more effective property rights, a monitoring networks underdeveloped. Over- precondition for medium- to long-term pro- ductivity gains. 25The fine collection ratio was 37 percent and 45 percent in 1985 and 1989, respectively (Bates, Gupta, and Fiedor 1994). 5. Economywide Policies and the Environment: A Case Study of Poland 111 The economic transformation program enterprises had been commercialized, and Poland made some progress in privatization SOEs were showing better productivity per- following enactment of the Law on Joint formance under the influence of hard budget Ventures, in December 1988, but the results constraints. Nevertheless, at the end of 1993, were limited. A major step forward occurred Poland had succeeded in privatizing only one- in early 1990, when the goverment initiated quarter of its SOEs, rather than one-half, its Economic Transformation Program (ETP), which had been the original goal. Nearly which was supported by two reform initia- 6,000 enterprises remained in public hands, mainly In the industrial and construction tives: in the enterprise sector, Poland em- 26 barked on an unprecedented privatization sectors. program, while, in banking, the reform pro- The enterprise and bank restructuring gram encompassed a number of policy mea- sures which, inter alia, were intended to program strengthen the governance and institutional While significant progress was made under capabilities of the nine newly-created state- the ETP, the pace of privatization was slowed owned commercial banks (SOCBs). A key in the wake of the recession, which followed component of the ETP was the program (an- its introduction, and the simultaneous collapse nounced in February 1991) to privatize about of trading arrangements, linked to the CMEA. 50 percent of the 9,000 SOEs within a period Furthermore, the process of privatization of three years. A "multitrack" approach was proved to be much more complex and lengthy adopted, although the two main routes have than initially expected. For example, the been "commercialization" and "liquidation." principle of voluntary enterprise transforma- The former case is intended to be transitory: tion, embodied in the privatization process, the SOE is converted to a joint stock (or meant that it could be retarded by recalcitrant limited liability) company, with a view to later workers' councils; and there was insufficient sale or distribution of shares, either through implementation capability in virtually all the "individual" or "mass" privatization. With responsible institutions. The severe financial "liquidation," assets are transferred to a new difficulties of the SOEs, which did not service private company, through their contribution, their debts, transmitted themselves to other sale or lease. The "mass" privatization compo- sectors of the economy, through the buildup of nent is an interesting feature of the ETP, debt-service arrears-in particular, through a designed to enable the public at large to par- deterioration in the quality of the banking ticipate, eventually, in the transfer of owner- system's loan portfolio, and fiscal problems ship from public to private hands. faced by the government. As of September From the beginning of the privatization 1992, SOE loans represented the primary efforts, rapid progress was made in the small exposure of the SOCBs, and SOEs accounted business and retail sectors, although privatiza- for most of the roughly 2,000 enterprises, tion has now been extended to other parts of mainly in heavy industry, not servicing their the economy. By 1993, the share of private debt. The government recognized that the sector activity had reached one-third in indus- future success of the ETP in achieving the try and three-quarters in construction. The objective of private sector-led growth, re- private sector also accounted for most of the quires functioning financial intermediaries. growth in GDP during 1993. Productivity To address the range of problems faced by growth was much slower among SOEs, where the enterprise and banking sectors, the govern- the task of privatization was clearly never going to be easy. Even so, Poland had privat- 'Information in this paragraph is based on World ized over 2,000 SOEs by the end of 1993, 423 Bank in te r 3o ' ~~~Bank (1994), Chapter 3. 112 VOLUME II ment, with the assistance of the World Bank, Energy sector restructuring developed an Enterprise and Bank Restructur- As stated in the preceding section, energy is ing Program (EBRP), to address the inter- related problems of the SOEs, which are not oneo th setors where the g ernm servicing their debts, and of the commercial wishesetoret oershiprin the vediu . . ~~~term. Therefore, restructuring efforts have so banks. The EBRP contains additional mea- sures designed to facilitate the restructuring of far focused more on commercialization of enterprises, using the nine SOCBs as key SOEs and the creation of an institutional, "cniiain l egal, and economic framework that will agents of change (through "conciliation" facilitate competition and greater private procedures). Given the fundamental role to be 27 sector participation in the longer term. assumed by the banks, the EBRP also aims to In thcoalisetor the monopliticm ar stimulate their efficiency, including (ulti- In the coal sectort the monopolistic Hard mately) privatization. A crucial outcome of Coal Corporation was abolished in 1990 and matey) rivaizaion A cucil oucom of replaced by the State Hard Coal Agency, an the EBRP will be further movement from soft advisoy and pang bod y, al to hrd udges, n bth te eterpiseand advisory and planning body, which is also tonhard budgors, ioth the entrprise an supervising the planned closure of unprofit- banking sectors. However, the government has al ie.Aon e needn n identified four sectors in which it wishes to competin es aenvisagedet and retain ownership of selected enterprises in the produtin activiies, with gsi o more mediu ter, whre pivatzatin wil be production activities, with six or more In medium t where p atiztion w l bex wholesale distribution (Poland, Ministry of decided~~~~~~~ on a.aeb-aebss ntenx Industry and Trade 1992; World Bank 1993b, three years. These are the energy, mining, steel . '. and defense sectors, which include about 300 Vol. 1). Operating subsidies to coal mines enterprises accontinforabou20erwere abolished on April 1, 1992, except for enterprises, outing Or abu ente of those mines being closed, although loss-mak- idustrial output. atherwise al rpises ing continues at a high level. Given the fact areoeligible. for privatization as rapidly as that lignite is sold almost exclusively to the Despite these changes, however, it is obvi- electric power industry, lignite mines are to be ous that SOEs will continue to be major play- vertically integrated with the mine-mouth ers in the Polish economy, at least in the power plants to which they are attached, mersum term. In thi context,therei . through one or more holding companies. medium term. In this context, there iS encour-Efottoirduecmttonoth agig vidnc frm rcen srve o 7 Efforts to introduce competition to the agingevidece frm a rcent urveyof '~electric power sector started with the dissolu- large state-owned manufacturing enterprises tiont(in 1990) of ted ignite Cor- (Pinto, Belka, and Krajewski 1993; Hume and tion, which ontroled 10 gnterp Pinto 1993). Contrary to negative stereotypes an .reationo the Polsh Pow erGrideon of SOEs as resistant to change and carrying on and creation of the Polsh Power Grid Com- in a business-as-usual manner, the survey pany (PPGC), along with 34 autonomous shows that even without privatization SOEs generating and 33 distributing companies. While these companies are all public, their are adjusting, restructuring, and increasing moplyowrimuhes.TePG profitability in response to the ETP. Of partic- handle eleric transmissi and ismv ular relevance is the finding that "all firms . e managed to reduce the consumption of materi- ing toward a more stringent system of merit als and energy per unit of sales" (Hume and order dispatching, to minimize system costs. Pinto 1993, p. 19). Specifically, the 31 best companies reduced energy and material con- sumption by 22 percent between 1991 and 77The description of energy restructuring in this sumption by 22 percent between 1991 and section serves to show the direction of change, and is 1992, and the remaining ones by approxi- based on available published sources. The Polish mately 17 percent (Hume and Pinto 1993). government is continuously reviewing the situation, and the actual numbers are likely to change periodically. 5. Economywide Policies and the Environment: A Case Study of Poland 113 Under a restructuring program, prepared with foreign investors are apprehensive about the assistance of the Bank, it is proposed to future uncertainty (including environmental reduce the number of generating companies to policy), and municipalities are reluctant to (a) four to five companies based on hard coal, assume ownership of heat supply sources, nine to ten combined heating and power given existing unprofitable conditions. enterprises, and one to two integrated power- lignite companies. There would be 12-15 Energy pricing reform distribution companies. The PPGC would Aside from the restructuring of enterprises purchase power competitively from alternative v generation sources and resell to the distribu- an bnkig, arcentracomponetco theE was energy pricing reform, which gained tion companies, which would also be permit- momentum in the early part of 1990. As in the ted to buy from industrial generators and small case of prnvatlzation, less successful earlier combined heat and power plants (Gula 1992). attempts had been made in 1988 and 1989. In oil and gas, less progress has been made The benefits of these earlier attempts to raise in restructuring, although the aim is to create two idepedent xploatio and rodution energy prices were largely eroded by inflation two independent exploration and production and exchange rate depreciation: despite the companies from the existing Polish Oil and doubling of energy prices in 1988, and further Gas Company (PGNG). These two companies i would compete with foreign enterprises in the inraein18(nnonltrm)th average energy price in Poland was barely access to new oil and gas deposits. A single one-third that of Western Europe at the end of transmission company would be formned for 1989, and lower than Bulgaria, Czechoslova- gas, selling to six independent regional distri- kia, and Hungary (Hughes 1991, Figure 10). bution entities. The state would retain a con- However, major reforms in energy prices took trolling interest in the distribution of petro- place from the beginning of 1990 (Bates, leum products and in refining, although a Gupta, and Fiedor 1994, Figures 3-6). strong private sector participation would be After eliminating uniform national coal encouraged (Poland, Ministry of Industry and pricing and the subsidies on coal transport, in Trade 1992). Finally, district heating has 1990, coal prices (ex mine) rose from under traditionally been decentralized, with about U fifty independent heating companies supplying (at ter meno18 ton U$ per tee (inMa heat o lage an smal tons. Te lagest (at the end of 1989) to US$37 per tce (in May heating sytoelar anud sm balldtowns Te lrarests 1991). Some erosion took place thereafter, but heating systems could be based on a transmis- sion company purchasing from several differ- coal prices (ex mine) were still at US$33 per ent suppliers and selling to different local tce in January 1992, in line with export parity, distribution companies. Of course, as systems allowing for local transport and other costs; decrease in size, integration of transmission further price increases brought coal to US$45 and distribtion i likeltooccr,andinthe per metric ton a year later. In the gas sector, and distribution is likely to occur, and in the dlvrdpie oidsr n eieta smalestsysems a inge itegate copan delivered prices to industry and residential smallest systems, a single integrated company cnues(nUS olreuvln) n may be justifiable for production, transmis- cnues(nUS olreuvln) n mion,aynb justifiableuforprdtion( d, M stra i o creased by factors of 3 and 46, respectively, Ision and di (ldMirofrom the end of 1989 to the beginning of Industry and Trade 1992). 1992. They stood at US$0.12/m3 and The results of restructuring in the energy US$0.18/i3, respectively, on February 1, sector have so far been limited, with only 1993. Although still lower than Western three large power plants and two large district heating enterprises commercialized. Among other obstacles, workers fear large-scale layoffs, social pressures are exerted by miners, 114 VOLUME II European levels,28 they demonstrate that between consumer categories, according to the substantial progress in energy pricing reform particular supply and demand characteristics. had occurred over a thirty-seven-month pe- Reliable data on the economic costs of energy riod. Similar large price increases were imple- supply in Poland are not easy to obtain, but mented for electricity-by factors of 3.5 for information from Bates, Cofala, and Toman industry and 12 for households, in U.S. dollar (1994) tends to confirm that there is room for equivalent, from the end of 1989 to the begin- further price increases.30 Nevertheless, the ning of 1992. In US$/kWh, they rose from achievements so far should not be underesti- roughly US$0.01 to US$0.04 for industry, and mated: as a crude indication of the priority from US$0.005 to US$0.06 for residential attached to reforms in energy pricing, it can be consumers between December 1989 and noted that the share of household expenditures February 1, 1993. Again, however, these were on fuel and power increased from 1.7 percent below the corresponding average Western in 1989 to 8 percent in the first half of 1992, European prices.29 while the food share fell from 45.8 percent to Following a history of heavy subsidies, 39.1 percent, and the share of clothing and district heat prices for residential consumers footwear fell from 16.9 percent to 8.1 percent were doubled in February 1991; increases, of over the same period (Bates, Gupta, and 60 percent and 84 percent respectively, oc- Fiedor 1994, p. 18). The index of aggregate curred in August 1991 and January 1992. expenditures more than doubled from 1990 to Subsequent increases raised the price paid by the first half of 1992, but the component of the residential consumers to US$5.66 per index related to fuel, heat, and electricity went gigajoule (GJ) by July 1993, compared with up nearly six times (Bates, Gupta, and Fiedor US$0.51/GJ only three years earlier, cutting 1994, Table 26). subsidies to 35 percent. Finally, the prices of liquid fuels have been raised to cover produc- The impact of restructuring and energy tion costs, although the values (US$0.45 per pricing reform liter for gasoline and US$0.32 per liter for While these profound changes in economic diesel oil at the beginning of 1992) were very policy are already in train, and beneficial low by Western European standards, due to results discermible, the timing and speed of the relatively low tax component (Poland, further change are matters of considerable Ministry of Industry and Trade 1992, p. 135). uncertainty in Poland, as in the other econo- The above comparisons between energy mies in transition. The effects of decades of prices in Poland and in Western Europe are central planning will take years to attenuate. intended to be indicative, since cost differ- However, we have developed scenarios to ences can be expected between countries and obtain quantitative indications of the orders of magnitude of the impact of economic policy 28 ~~~~~~~~~~change on the environment (Bates, Cofala, 28For example, natural gas prices for households in andnToman 199).nByrcomparinBthesresultsao Poland were substantially below those in France, andToman 1994).Bycomparmgtheresultsof Ireland, Switzerland, and the United Kingdom in the these scenarios, which we term the IBRD first quarter of 1993. Polish industrial prices were also simulations, with other work, reasonable lower, except in the case of the United Kingdom, where conclusions about likely changes in the future they were comparable (IEA 1993, Part D, Tables 10 and 12). 29Polish electricity prices for industry and 3r'ableA1.4 in Bates, Cofala, and Toman (1994), households were 50 percent or more below those in which calculates fuel prices to final energy consumers, Denmark, Germany, Italy, Portugal, and Switzerland in on the assumption that prices fully cover economic the first quarter of 1993 (IEA 1993, Part D, Tables 8 costs from 1995 onwards, is the basis for this and 9). conclusion. 5. Economvwide Policies and the Environment: A Case Study of Poland 115 environmental situation can be drawn. GDP, the production of the energy sector itself The IBRD simulations use projections of and that of certain heavy energy users, such as GDP and its breakdown by energy-using pig iron and steel, decline in absolute terms activity to find a time-path of final energy after 1990, while the production of most other demands and a least-cost solution to satisfy energy-intensive intermediate products dimin- those demands. The dynamic optimization ishes in relative terms (copper, aluminum, part of the model is carried out using sulfur, ammonia, and cement). Freight and multiperiod linear programming (LP). The passenger transport, along with housing, decision variables in the LP analysis include naturally increases with population and eco- new plant investments and fuel choices (e.g., nomic growth (Bates, Gupta, and Fiedor 1994, coals of different grades, as well as coal wash- Table 28). Furthermore, driven by higher ing). More than seventy technologies and energy prices and competition, energy con- twenty-two primary energy types are covered sumption per unit of value added falls across in the framework, including energy imports all the productive sectors (industry, construc- and exports. Projections are carried out at tion, agriculture, and transport). However, per five-year intervals from 1995 to 2015, inclu- capita energy consumption in the residential sive. The LP objective function gives us the sector is expected to increase due to higher per total cost of meeting final energy demands. A capita needs, including an expansion in space key assumption underlying the model is that heating (Bates, Gupta, and Fiedor 1994, Table energy prices in Poland fully reflect economic 29). costs by 1995, which is plausible in view of These changes in the level and structure of the progress already made (see the subsection, GDP, and in economic activity and energy Energy pricing reform). intensity at the sectoral level, are based upon The model's projections of GDP growth are a mixture of analysis and judgement, using the in line with those recently prepared by the views of Polish sectoral experts, in combina- World Bank (World Bank 1993c). Following tion with past experience, both in Poland and the drop that occurred during 1989-91, GDP other countries. They result in the aggregate is expected to continue growing in 1993 and primary energy demands and energy intensi- 1994 to a level about 5 percent above the 1990 ties shown in Table 5-5. It is noticeable that level by 1995; the growth rate reaches 4.5 the sharp drop in the share of hard coal and percent per annum (p.a.) in the period lignite-from a combined 76.9 percent in 1996-2000, and tapers off to 4 percent p.a. 1988 to 68.5 percent in 2000 and 63.8 percent after 2000. On that basis, GDP reattains the in 2010-is offset mainly by increasing use of 1988 level by 1997 and doubles from the 1990 natural gas and oil. Although GDP in 2000 is level soon after 2010. Higher living standards, 17 percent above the 1988 level, total primary along with market reforms, cause a steady energy consumption actually falls. Hence, reduction of the share of industry in GDP energy intensity drops by nearly 20 percent. from 46 percent in 1990 to 40 percent by By 2010, energy intensity is only 63 percent of 2010, a ratio which is more in line with the the 1988 figure. In line with the experience in developed countries of Western Europe. The OECD countries, electricity consumption shares of agriculture and construction also increases more rapidly than energy consump- decline after 1990, but there are increases in tion overall, so that the decrease in the inten- the shares of transport and (especially) ser- sity of electricity consumption is less than that vices (Bates, Gupta, and Fiedor 1994, Table of aggregate energy. 27), as would be expected in an economy in We have also calculated in Table 5-5 emis- transition toward a market-oriented frame- sions of PM, SO2, NOR, and CO2 that are work. Within the industrial component of implied by the above scenarios, on the as- 116 VOLUME II sumption that no new environmental policies capita in the residential sector increases. are put into place. However, we have allowed While, in the High Scenario of the MIT report, for some improvements in the transport sector, the overall intensity of the economy also falls as new diesel vehicles are purchased, with less with respect to electricity consumption emissions comparable to Western European than with respect to energy as a whole, elec- standards. The decline in the total emissions tricity intensity of GDP rises in the Low of SO2, NOX, and (especially) PM are striking, Scenario, in contrast to our results (Table 5-6). in view of the increase in real income over the Of particular interest are the MIT results for period. Only CO2 emissions increase in total. primary energy demands and emissions, under However, emissions per unit of GDP fall in all the assumption of no new environmental cases, the most dramatic being those of PM, controls (Table 5-6). The combined hard coal which are cut to below 60 percent of the 1988 and lignite share of primaiy energy demand in level by 2000. 2010 is 62-63 percent, almost the same as our We compare the results obtained in the result. Emissions of PM (1.3 million metric IBRD simulations first with those in the report tons), SO2 (3.5-3.8 million metric tons), NO, produced recently by the Polish Ministry of (1.3-1.6 million metric tons) and CO2 Industry and Trade (MIT), using essentially (430-510 million metric tons) are all very the same methodology (Poland, Ministry of close to the results obtained in Table 5-5. Industry and Trade 1992). While the latter are Another set of simulations was conducted not regarded as official government forecasts, by the IBRD staff in the course of preparing they observed, as closely as possible, official an environmental action plan (EAP) for Cen- data and known government policies. The tral and Eastern Europe (EAP 1993). The MIT report incorporates a High and a Low results were submitted to the ministerial Scenario for GDP growth, reflecting annual conference which took place in Switzerland, growth rates of 4.2 percent and 2.3 percent, in April 1993. The EAP methodology, how- respectively. These growth rates imply levels ever, is quite different. Instead of a dynamic of GDP in 2010 which are, respectively, 17 LP approach, the simulations are based on an percent higher and 24 percent lower than the input-output framework, using forty-eight- one we have used, thus providing a useful sector input-output tables for each country for range about our estimate for comparison various years. There are different input-output purposes. In the MIT Scenarios, economic coefficients for new and old capital stock. The restructuring and energy price increases cause former are assumed to be the same as those improvements in energy efficiency, and the prevailing in Western Europe in the mid- shares of industry, construction and agricul- 1980s and do not change. On the other hand, ture in GDP decline to values which are com- the coefficients for old capital stock change in parable to ours, while those of transport and response to increases in efficiency, and to services also rise (Bates, Gupta, and Fiedor changes in relative prices. 1994, Table 32). Within the industrial sector, Five scenarios are examined in the study. the MIT results again parallel ours, with Of these, the first three illustrate the impact of falling or stationary shares for heavy energy differences in the nature and speed of eco- users-such as iron and steel, chemicals, and nomic reform on the environment. The last nonferrous metals-as well as the energy two examine the impact of more stringent sector itself. As in our scenarios, economic environmental policies (as compared with the restructuring and energy price increases lead Base Case). Only the first three EAP scenarios to declines in energy intensity for the economy are discussed here, as both the IBRD and MIT as a whole, by about the same order of magni- simulations that we presented above, which tude (Table 5-6); and energy consumption per assume no new environmental policies, are 5. Economywide Policies and the Environment: A Case Study of Poland 117 put into place. The three EAP scenarios are (a) ing then that emissions of PM, SO2, NOR, and the Base Case, which presupposes a reason- CO2 (Table 5-7), are all much lower than in ably comprehensive price and enterprise the MIT and IBRD simulations (as discussed reform program; (b) Slow/Delayed Reform, below). which assumes that pricing and other market To conclude, all three simulations point to reforms are delayed until 1995/96; and (c) clear environmental gains from restructuring, Accelerated Reform, where the "government combined with energy pricing reform, al- presses ahead with radical economic reforms though we have not attempted to estimate the and strict enforcement of hard budget con- relative impact of each component. The ro- straints for enterprises" (EAP 1993, Annex 5, bustness of the results is evident from the data p. 1). In practice, this implies different adjust- in Tables 5-5, 5-6, and 5-7, in terms of the ment periods for the input-output coefficients four major types of emissions (SO2, NO,, PM, associated with old capital stock. Thus, in the and C02) across the simulations. The reduc- Base Case, improvements in efficiency and tions in emissions are greatest for the EAP the adjustment of energy and other inputs per simulations for all pollutants.32 Comparing unit of output produced using old capital are across emissions, it is reassuring to note that phased over a period of ten years. Under the the biggest drop occurs for particulates, which Slow and Accelerated Reform scenarios, this have the worst effect on health.33 While the adjustment period is twenty years and five reductions are not as dramatic for the other years, respectively, and the latter scenario pollutants, it should be kept in mind that GDP assumes that 20 percent of the old capital increases by 1.5 to 2 times over the same stock is scrapped by 1995, as enterprises period, implying a significant decline in emis- contract or shut down. In addition, under the sions per unit of GDP (see, for instance, the Slow/Delayed Reform scenario, it is assumed results for IBRD simulations in Table 5-5). that new investment has (input-output) coeffi- cients based on typical Soviet technology The impact of changes in environmental rather than Western technology. policy Like the IBRD simulations, the projections In the subsection, Environmental policy in of GDP growth in the EAP scenarios follow Poland, we argued that, while Poland had an current World Bank estimates up to 1995. elaborate system of pollution taxes on paper, Projections of primary energy demand, elec- its implementation left much to be desired. tricity consumption, and energy intensity for Further, we also stated that the efficacy of the three scenarios are shown in Table 5-7. financial instruments itself was questionable, Note the sharp drop in the share of coal in primary energy by 2010. This drop, however, occurs after 2000, unlike in the IBRD and Scenario). MIT simulations. The other notable change is a marked increase in the share of natural gas 32This is mainly due to a sharp increase in the in the demand for primary energy to approxi- share of natural gas in primary energy demand and a big mately 30 percent by 2010. The substitution of decline in energy intensity in the EAP simulations. In fact, of all three simulations, the decrease in energy (clean burning) natural gas for coal in the EAP intensity is greatest for the Base Case in the EAP scenarios, occurs against a backdrop of declin- simulations. This, in turn, is due to the fact that 20 ing energy intensity of GDP.3' It is not surpris percent of the old capital stock is scrapped and replaced by new capital by 2000 (in the EAP Base Case). 33"Particulates in the air (soot, dust, metal dust) 3'ln fact, as discussed below, EAP energy intensity are thought to be the most serious type of pollution, projections show a sharper drop, compared to the IBRD both with regard to health impacts as well as economic and MIT simulations (except under the Slow Reform costs" (World Bank 1992, p. iii). 118 VOLUME 11 given the incentive structure that prevailed. In assumed that after 1995, new diesel engines in light of the ETP, however, and all that it the transport sector will have 20 percent lower implies for firm-level behavior, it is logical to unit emissions of NO. For particulates, while expect greater responsiveness to these instru- there is a substantial decline in emissions in ments in the future. all three simulations, as in the case of S02 In May 1991, the Polish government issued emissions, only the EAP simulations meet the the National Environmental Policy of Poland NEPP target. The IBRD simulations come (NEPP) (Ministry of Environmental Protec- close, with PM emissions at approximately tion, Natural Resources and Forestry 1990), 1.46 mt in 2000 (Table 5-5), but the MIT indicating its environmental goals and priori- simulations fall short, at 1.6 mt (Table 5-6). ties, and the direction that environmental It should also be emphasized that the simu- policy would take in Poland. In particular, lation results with respect to emission levels "the greatest possible advantage will be taken are based on a number of assumptions, all of of market mechanisms" (NEPP 1990, p. 4) which may not come true. For instance, under and the medium-term goal is to follow envi- the Base Case EAP scenario, 20 percent of the ronmental policies that "enable Poland to capital stock is replaced by the year 2000. It move closer to European environmental stan- could be argued that this is too optimistic. dards and allow it to join the EEC" (NEPP Thus, in addition to restructuring and energy 1990, p. 12). With regard to air quality, a pricing reform, Poland will need to go further noticeable reduction in the emissions of and deal with air pollution through a stronger particulates and gases was identified as a environmental policy, if it is to achieve its short-term priority. Specific medium-term medium- and long-term environmental goals. goals include, inter alia, a reduction of SO2 In this context, the sharp increases in nomi- emissions by 30 percent, NO, emissions by 10 nal and real levels of fees from 1990 to 1993 percent, and PM emissions by 50 percent are steps in the right direction (Tables 5-8 and (implying an increase in removal efficiency 5-9). According to Zylicz (1993), the current for large stationary sources from 92 percent to sulfur dioxide rate ($75/metric ton) is "higher 96 percent) by the year 2000 (compared with than almost anywhere in the world" (p. 13). At 1980 levels). Thus, the goal is to reduce S02 the same time, however, he concedes that the emissions from 4.2 million metric tons per SO2 fee is still lower than the marginal cost of year (mt/year) to 2.9 mt/year, NO, emissions a 30 percent reduction, estimated at approxi- from 1.5 mt/year to 1.3-1.4 mt/year (NEPP mately $600 per metric ton. In light of the 1990, p. 13), and PM emissions from 2.82 IBRD simulations, even larger increases in mt/year to 1.41 mt/year.34 fees are necessary to attain plausible emis- From Tables 5-5 and 5-6, it is clear that sions objectives for Poland in the longer term under the IBRD and MIT simulations, the (Table 5-9). The need for increases in NO, medium-term goal for S02 emissions will not charges (from US$75/metric ton to be met. While the IBRD, M1T (Low), and US$2,700/metric ton by 2010) is particularly EAP simulations achieve the goal for NO, great, due to the high unit cost of controlling emissions (Tables 5-5, 5-6, and 5-7), the NO,. Although PM and SO2 fees also might assumptions in the IBRD and MIT simulations have to be subject to significant increases, by should be kept in mind. In both, though no factors of 10-20 (to around US$700/metric new environmental policies are employed, it is ton by 2010). As a matter of interest, the calculated NO, tax of over US$1,000/metric ton in 2000, translates into taxes of approxi- 34Actual emissions of particulates in 1980 are not mately US$0.056/liter on diesel fuel and stated in NEPP, but they are shown as 2.82 mt/year in US$0.025/liter on gasoline. the MIT Report. S. Economywide Policies and the Environment: A Case Study of Poland 119 An indication of future developments in shown in Bates, Cofala, and Toman (1994), environmental policy is also provided in the Tables A2.4 and A2.5, from which it will be government's February 1990 Ordinance on the noted that coal demand under CAC is about 4 Protection of Air Against Pollution, which percent lower than in the Base Case, while gas contains plant-level performance standards for demand is about 8 percent higher. As ex- new and existing large stationary sources pected, environmental standards make a (expressed in emissions per unit fuel input). switch toward lower-emitting fuel more attrac- Existing stationary sources refer to plants tive. Figures 5-2 and 5-3 bring together the started after the Ordinance and completed salient information for emissions of SO2, NOx, after 1994. New source standards are effective PM, and CO2. The actual data for 1988 and from 1995. Existing sources must meet 1990 are also included, as well as the Base weaker interim standards until 1998, after Case as a benchmark. Substitution away from which the standards are tightened, but are still coal causes a large drop in SO2, NO., PM, and less stringent than the new source standards. CO2 emissions as compared with the Base Note that since these are plant-level standards, Case. it may still be possible to treat the plant as a The costs of these improvements, however, bubble and trade off intraplant emissions. are high. The net present value of costs from Another important point to note is that the 1990-2015, at a real discount rate of 12 per- Ordinance does not specify how these stan- cent, amounts to over US$12 billion, compris- dards are to be met. For instance, there is no ing increased cost of energy conversion and stipulation that best available technology emission reduction, loss of producer surplus, (BAT) is to be used. Thus, some flexibility for the cost imposed on final energy demand in the polluters is already built into the regula- the household and transport sectors, and loss tions. of consumer surplus from reduced final energy It is obvious that the 1990 regulations will demand.36 Of these costs, large point source further reduce emissions in addition to what controls and loss of producer surplus account will be achieved through enterprise and price for more than half, whereas transport controls reform. The IBRD simulations, discussed contribute less than half of the total. The loss earlier, tried to quantify the impact of these in consumer surplus and the cost of alterations regulations on emissions, using a CAC sce- in final energy demand by households are nario.5 CAC, however, as specified, also negligible. As Table 5-10 shows, however, includes a number of regulations for the unit costs of reducing emissions vary consid- household and transport sectors which, while erably between stationary and transport con- plausible, have not yet been incorporated in trols, and also across different options within Polish environmental policy, namely, a ban on the latter category. Thus, while low-NO, urban coal use by households and mandatory diesel engines appear to be cost-effective catalytic converters for all cars. These restric- methods for emission reduction, catalytic tions are supposed to be 50 percent effective converters and diesel oil desulfurization are by 2000 and fully operational by 2005. Fur- very expensive. It does not follow, however, ther, CAC assumes an additional 30 percent that these transport controls should be reduction in heavy-duty diesel engine emis- avoided. To make that judgment, we need sions, and a 75 percent decline in the sulfur information on the benefits as well. content of diesel oil (from 0.6 percent to 0.15 Despite the difficulties inherent in the percent) after 2000. The results for CAC are IBRD simulations, one result emerges clearly: "The scenario discussed earlier was the Base Case, 36For details, see Bates, Cofala, and Toman where no new environrental policies are put into place. (1994), Chapters 2 and 3. 120 VOLUME II given the magnitude of costs required to meet yields the same total emissions as CAC, could environmental objectives, and the substantial cut control costs by as much as one-half, competing claims for limited resources in an although the result requires the extension of economy in transition, such as Poland, there is incentive-based policies to decentralized a potentially important role for the application sources. Nevertheless, even with limited of economic instruments to air pollution application of economic incentives to large control. Notably, the IBRD simulations esti- stationary sources, the cost savings are not mate that replacing CAC regulations by a trivial. uniform emissions tax on all sources, which Table 5-1: Energy Intensity of National Income for Selected Countries (national income estimated according to purchasing power parity) Energy intensity in Average yearly kgoe/US$000 (1975 prices) decline (%) Country 1960 1975 1986 1960-73 1973-86 Poland 1,796 1,237 1,136 2.83 0.65 France 468 460 337 0.13 2.33 United Kingdom 745 557 490 2.21 0.97 Czechoslovakia 1,602 1,146 902 2.53 1.82 German Democratic Republic 2,019 1,495 1,055 2.29 2.64 Developed Western countries 540 513 418 0.40 1.57 CMEA countries 1,545 1,212 1,014 1.85 1.37 CMEA = Council of Mutual Economic Assistance. kgoe = kilograms of oil equivalent. Source: Bates, Gupta, and Fiedor (1994), Table 1. 5. Economywide Policies and the Environment: A Case Study of Poland 121 Table 5-2: Energy Intensity Relative to OECD for Selected CEE Countries, 1988 Energy use per unit of GNP; OECD = 1 Total Oil and gas Bulgaria 2.2 1.7 CSFR 1.9 1.0 Hungary 1.3 1.3 Poland 1.9 0.6 Romania 2.5 2.9 U.S.S.R. 2.7 3.1 CEE5 2.0 1.3 All CEE 2.5 2.6 CEE5 = Bulgaria, CSFR, Hungary, Poland, and Romania. CSFR = Czech and Slovak Federal Republic. Source: OECD (1990), Table 20, and OECD (1992). Table 5-3: Poland: Composition of Estimated Particulate Emissions (percent) 1975 1980 1985 1989 1991 Energy 41.9 50.2 53.6 53.3 55.2 Metallurgical 12.5 12.2 13.2 9.8 8.4 Chemical 11.4 10.7 11.6 11.4 9.3 Mineral 24.4 18.7 11.1 9.1 7.6 Subtotal 90.2 91.8 89.5 83.6 80.5 All industry 100.0 100.0 100.0 100.0 100.0 Source: Bates, Gupta, and Fiedor (1994), Table 20. Table 5-4: Poland: Composition of Estimated SO2 Emissions (percent) 1975 1980 1985 1989 1991 Energy 69.5 72.5 76.8 74.0 78.1 Metallurgical 10.6 12.5 7.3 5.0 4.6 Chemical 9.5 6.9 6.2 5.2 4.9 Mineral 3.2 1.8 1.2 1.5 1.3 Subtotal 92.8 93.7 91.5 85.7 88.9 All industry 100.0 100.0 100.0 100.0 100.0 Source: Bates, Gupta, and Fiedor (1994), Table 21. 122 VOLUME II Table 5-5: IBRD Simulations: Scenarios for Energy Demand, Energy Intensities, and Emissions 1988 1990 2000 2010 Primary energy demand (PJ) 5,387 4,223 5,071 5,848 Fuel shares (%) 65.9 61.1 58.4 55.1 - Hard coal 11.0 13.4 10.1 8.7 - Lignite 7.5 8.9 11.2 12.9 - Natural gas 13.7 15.3 17.3 18.8 - Oil 1.9 1.4 3.0 4.5 - Other Gross electricity consumption (TWVh) 146 133 173 230 Energy intensity of GDP (1988=100) 100 89 81 63 Electricity intensity of GDP (1988=100) 100 102 100 92 SO2 - total 3,827 2,832 3,192 3,575 - per unit of GDP ('000 metric 5.587 4.666 4.000 3.027 tons per billion Zl) NO, - total 1,363 1,187 1,179 1,321 - per unit of GDP ('000 metric 1.990 1.956 1.477 1.119 tons per billion ZI) PM - total 2,145 1,468 1,463 1,211 - perunitofGDP('O00metric 3.131 2.419 1.833 1.025 tons per billion ZI) CO2 - total (million metric tons) 454 364 432 490 - per unit of GDP ('000 metric 663 600 541 415 tons per billion Z1) PJ = petajoule. TWh = terawatt-hour. Source: Bates, Cofala, and Toman (1994), Tables A2.1 and A2.3. 5. Economywide Policies and the Environment: A Case Study of Poland 123 Table 5-6: MIT Scenarios for Energy Demand, Energy Intensities, and Emissions High Low 2000 2010 2000 2010 Primary energy demand (PJ) 5,327 6,143 4,783 5,192 Fuel shares (%) - Hard coal 59.3 55.0 56.8 52.0 - Lignite 9.6 8.3 10.7 9.9 - Natural gas 12.0 13.7 11.0 13.1 - Oil 16.1 18.0 18.6 20.2 - Other 3.0 5.0 3.0 4.8 Gross electricity consumption (TWh) 186 .46 161 203 Energy intensity of GDP (1988=100) 76 59 93 72 Electricity intensity of GDP (1988=100) 97 86 114 103 Emissions (thousands of tons) SO2 3,460 3,800 3,290 3,510 NO, 1,380 1,550 1,220 1,280 PM 1,630 1,290 1,600 1,310 CO2 (millions of tons) 455 510 405 429 PJ = petajoule. TWh = terawatt-hour. Source: Poland (1992b). 124 VOLUME II Table 5-7: EAP Scenarios for Energy Demand, Energy Intensities, and Emissions Base case Fast reform Slow reform 2000 2010 2000 2010 2000 2010 Primary energy demand (PJ) 2,474 3,528 2,433 4,364 3,231 4,531 Fuel shares (%) - Coal 72.0 54.2 71.6 56.8 75.6 62.1 - Natural gas 13.9 32.1 14.0 28.7 11.8 26.7 - Oil 14.1 13.7 14.4 14.5 12.6 11.2 - Other 0.0 0.0 0.0 0.0 0.0 0.0 Electricity consumption (TWh) 99.2 177.5 97.6 223.6 127.9 215.5 Energy intensity of GDP 57.3 52.6 49.2 47.0 90.4 89.9 (1988=100) Emissions (thousands of metric tons) SO2 1,381 1,469 1,352 1,880 1,838 2,022 NO, 633 901 625 1,119 823 1,156 PM 821 574 813 858 1,121 891 CO2 (millions of metric tons) 232 308 230 390 308 411 PJ = petajoule. TWh = terawatt-hour. Source: EAP (1993) and Gordon Hughes (1993). Table 5-8: Poland: Increases in Nominal Emissions Fees (Zl/kg) 1990 1992 SO2 4.80 1,100 NO, 4.80 1,100 CO2 0.90 300 Aromatic hydrocarbons 48.00 3,000 Lead 960.00 500,000 Source: Bates, Gupta, and Fiedor (1994), Table 39. 5. Economywide Policies and the Environment: A Case Study of Poland 125 Table 5-9: Poland: Increases in Real Emissions Fees-Actual and IBRD Simulations (US$/metric ton) End of 1990 April 1993 1995 2000 2010 S02 28 75 430 541 780 NO, 28 75 542 1,029 2,716 PM 7 38 176 640 640 Source: Bates, Cofala, and Toman (1994), Table 5.4. Table 5-10: Poland: Costs of Emission Control Measures for Large Stationary and Transport Sources Measure Pollutant reduced Unit cost (US$/mnetric ton) Car catalytic converters NO, 9,700 Low emission diesel engines NO, 810 Diesel oil desulfurization S02 4,000 Selective catalytic reduction, new power plant NO, 1,800-2,000a Desulfurization, new hard coal power plant SO2 510 a. Approximately 35 percent higher for a combined heat and power plant. Source: Bates, Cofala, and Toman (1994), Table A2.12. Figure 5-1: Poland: Energy Sector Balance Energy Supply & Demand Balance Dornestic energy supply 80%o Energy Imnports 20% Domes ic Production Imports - Coal 95% - Coal 2% - Oil & Gas 3% - Oil 68% - Hydro & Other 2% - Gas 27% - Electricity 4% 1 8% of production for exporls |820% of produiction for domestic use Exports Primary Energy Supply - Coal 88/o - Transformation & Energy Sector 32% - Oil 7% - Final Consumption 62ffo - Electricity 5% - Non-energy use 6% Transformation & Energy Sector Final Users - District heat 15% - Industry3 Non-energy use -CHP production 56% -Transport 6% Nneeg s *Gas manufacture 3% -Residential 6% - Owin seafindelroies 25o Comnnercial, Agricultural & Other 50% Source: World Bank (1993b). 5. Economywide Policies and the Environment: A Case Study of Poland 127 Figure 5-2: Emissions under Base Case and CAC Scenarios (IBRD simulations) I Sulphur Dioxide Estimates Base Case 3M| T.< ! i0 Projections r-~ ~ ~ ~ ~~~~~Ya 20 'o P 0 ~ ~ siats Pojections o s01 la -tt Igi-iM Base Case A - - 0 .... 1988 1990 2000 2010 2000 2010 1988 1990 2000 2010 2000 2010 Year Source: Bates, Cofala, and Toman (1994). Figure 5-3: Emissions under Base Case and CAC Scenarios (IBRD simulations) Particulate Matter 2500 Estimates 2000 o .Ba-- ;se case Proebctions 3 l l _ . l - | Carbon Dioxide (miiion tons) 1000I kli . * l l t _AEstimates / \ - l l l - _ ~~~Bassecase CAC 50 191eei9Sa 2000 2010 2000 2010 19SS 1880 2000 2010 2000 2010 Year Source: Bates, Cofala, and Toman (1994). 128 VOLUME II Bibliography Bates, Robin W. 1993. "The Impact of Eco- Papers. Krakow: The Krakow Academy nomic Policy on Energy and the Environ- of Sciences. ment in Developing Countries." Annual Gula, A. 1992. "Energy Policy in Poland: a Review of Energy and the Environment Focus on Institutional Changes and En- 18:479-506. ergy Efficiency." Revue de 1'e'nergie 444 Bates, Robin W., Janusz Cofala, and Michael (December). Toman. 1994. Alternative Policies for the Hughes, Gordon A. 1990. "Energy Policy and Control of Air Pollution in Poland. World the Environment in Poland." European Bank Environment Paper, Number 7. Economy 43 (March): 151-67. Bates, Robin W., Shreekant Gupta, and . 1991. "The Energy Sector and Prob- Boguslaw Fiedor. 1994. "Economywide lems of Energy Policy in Eastern Europe." Policies and the Environment: A Case Oxford Review of Economic Policy Study of Poland." World Bank, Environ- 7(2):77-98. ment Department, Working Paper No. 63. . 1993. Personal communication at the Bates, Robin W., and Edwin A. Moore. 1992. World Bank. "Commercial Energy Efficiency and the Hume, Ian M., and Brian Pinto. 1993. "Preju- Environment." World Bank, Policy Re- dice and Fact in Poland's Industrial Trans- search Department, WPS 972. formation." Finance and Development Bergson, Abram. 1987. "Comparative Produc- (June). tivity: U.S.S.R., Eastern Europe, and the International Energy Agency (EA). 1990. West." American Economic Review "Energy Policies: Poland. 1990 Survey." 77(3):342-57. Paris: OECD. British Petroleum Company p.l.c. 1992. BP . 1992. "Energy Statistics and Bal- Statistical Review of World Energy. Lon- ances of Non-OECD Countries." Paris: don. OECD. Environmental Action Programrnme for Central . 1993. "Energy Prices and Taxes, First and Eastern Europe (EAP). 1993. Docu- Quarter 1993." Paris: OECD. ment submitted to the Ministerial Confer- Kemme, David M., and Keith Crane. 1984. ence, Lucerne, Switzerland, April 28-30, "The Polish Economic Collapse: Contrib- 1993. uting Factors and Economic Costs." Jour- Fallenbuchl, Zbigniew. 1977. "The Polish nal of Comparative Economics 8:25-40. Economy in the 1970s." In East European Kharas, Homi J. 1991. "Restructuring Social- Economies Post-Helsinki. Comp. of Pa- ist Industry: Poland's Experience in pers, Joint Economic Committee, U.S. 1990." World Bank Discussion Paper No. Congress. Washington, D.C.: U.S. Gov- 142. ernment Printing Office. Kornai, Janos. 1980. Economics of Shortage. Gomulka, Stanislaw, and Jacek Rostowski. Two volumes. Amsterdam: North-Hol- 1988. "An International Comparison of land. Material Intensity." Journal of Compara- . 1986. "The Soft Budget Constraint." tive Economics 12:475-501. Kyklos 39(1):3-30. Gorka, Kazimierz. 1991. "Changes in the . 1992. The Socialist System: The Environmental Protection Policy of Po- Political Economy of Communism. Prince- land." In Kazimierz Gorka (ed.), Environ- ton, N.J.: Princeton University Press. mental and Economic Aspects of the In- Lipton, D., and J. Sachs. 1990. "Privatization dustrial Development in Poland: Selected in Eastern Europe: The Case of Poland." 5. Economywide Policies and the Environment: A Case Study of Poland 129 Brookings Papers on Economic Activity nal of Comparative Economics 2:293-341. 10:277--92. Marrese, Michael, and Janet L. Mitchell. Syzdlo, S. 1990. "The Average Wage Level in 1984. "Kornai's Resource-Constrained Coal Mining and its Consequences for the Economy: A Survey and an Appraisal." National Economy." In Rachunek Journal of Comparative Economics Ekonomiczny w Gospodarce Surowcami 8:74-84. Mineralnymi. Krakow: Polish Academy of Milanovic, Branko. 1989. Liberalization and Sciences, pp. 270-82. Entrepreneurship. Dynamics of Reform in Umer, A. 1993. Personal communication. Socialism and Capitalism. Armonk and Polish Academy of Sciences. London: Sharpe. United Nations Environment Programme Ministry of Environmental Protection, Natural (UNEP). 1992. Saving Our Planet: Chal- Resources and Forestry. 1990. "National lenges and Hopes. Nairobi. Environmental Policy of Poland (NEPP)." U.S. Energy Information Administration November. (EIA). 1990. Annual Prospects for World Moroney, John R. 1990. "Energy Consump- Coal Trade 1990. DOE/EIA-0363(90). tion, Capital and Real Output: A Compari- Washington D.C.: U.S. Government Print- son of Market and Planned Economies." ing Office. Journal of Comparative Economics Wilczynski, Piotr. 1990. "Environmental 14:199-220. Management in Centrally Planned Non- Nowicki, Maciej. 1993. Environment in Po- Market Economies of Eastern Europe." land: Issues and Solutions. Dordrecht, the World Bank, Environment Department, Netherlands; Boston and London: Kluwer. Working Paper No. 35. Organisation for Economic Co-operation and World Bank. 1991. "Poland: Energy Market Development (OECD). 1990. OECD Development." Report No. 8224-POL. Economic Outlook 48 (December). Paris: . 1992. "Poland: Environmental Strat- OECD. egy." Report No. 9808-POL. *. 1992. Reforming the Economies of . 1993a. "Poland-Cogeneration Pri- Central and Eastern Europe. Paris: vatization Project." Staff Appraisal Report OECD. No. 11353-POL (Annex 2-3), March 1. Pinto, Brian, Marek Belka, and Stefan . 1993b. "Poland: Energy Sector Re- Krajewski. 1993. "Transforming State structuring Program." Five volumes. Re- Enterprises in Poland: Microeconomic port No. 153/93, Industry and Energy Evidence on Adjustment." World Bank, Department, Energy Sector Management Policy Research Working Paper Series Assistance Program (ESMAP). No. 1101. . 1993c. "The Report and Recommen- Poland, Ministry of Environmental Protection, dation of the President of the International Natural Resources and Forestry. 1992. Bank for Reconstruction and Development Poland's Environmental Investment Ex- to the Executive Directors on a Proposed penditures in 1991 and Their Trends Since Enterprise and Financial Sector Adjust- 1988. Unpublished. ment Loan in an Amount Equivalent to Poland, Ministry of Industry and Trade. 1992. US$450 million to the Republic of Po- Energy Policy of Poland and the Draft land." Report No. P-5940-POL, April 8. Programme to the Year 2010. Warsaw, . 1994. Poland: Policies for Growth November. with Equity. A World Bank Country Slarna, Jiri. 1986. "An International Compari- Study. son of Sulphur Dioxide Emissions." Jour- 130 VOLUME II World Resources Institute (WRI). 1992. shop on Taxation and Environment in World Resources, 1992-93. New York: European Economies in Transition, OECD Oxford University Press. Environment Directorate, CCEET/ Zylicz, Tomasz. 1993. "Environmental Taxes ENV/DAFFE (92)243. in Poland." Paper presented at the Work- 6 Energy Sector Policy and the Environment: A Case Study of Sri Lanka Peter Meier, Mohan Munasinghe, and Tilak Siyambalapitiya THE ENVIRONMENTAL IMPACTS of specific energy investments are fairly well recognized, and environmental impact assess- ments of energy projects in developing countries have become a routine and integral part of the Bank's policies to promote sustainable development. But much less well understood are the linkages between broader sectoral policies and the environment. For example, the environmental consequences of energy sector pricing reforms that have become a quite common component of structural adjustment lending remain largely unexplored. Similarly, effective procedures for incorporating environmental concerns into energy and electricity sector planning are not well established beyond some routinely affirmed general principles about the importance of economic efficiency, and the presump- tion that the environmental impacts of an economically efficient energy sector will be less than one characterized by subsidized prices and technological inefficiency (such as inordinately high transmission and distribution (T&D) losses, poor heat rates of thermal electric generation plants, etc). The authors acknowledge the many contributions from their colleagues in Sri Lanka, especially D. C. Wijeratne and K. Nanthakumar. Others who provided valuable review and comments of earlier versions of this study included Kenneth King and Wilfrido Cruz. 131 132 VOLUMEII The impact of direct environmental con- nants between the Bank and its borrowers that straints on sectoral policy are equally poorly provide for tariff increases to generate suffi- understood, except perhaps in the general cient revenue to meet self-financing targets or sense that if energy costs fully reflect the rate of return on equity have become an in- environmental damages of energy production, creasingly frequent part of Bank power sector optimal resource allocation is ensured. For loans. example, very little work has been done to Many methodological problems arise in analyze alternative policy options for achiev- evaluating the environmental consequences of ing greenhouse gas emission reductions in such sectoral policies. Although the environ- developing countries, again beyond routinely mental benefits of an efficient pricing policy stated assertions of the importance of efficient are perhaps intuitive, in practice the relation- resource utilization, and that renewable energy ships are quite complex, and the magnitude of and natural gas should replace coal. Indeed, the impacts will be dependent upon a variety such studies of this subject as have appeared of specific assumptions. If higher prices, are based on highly aggregated macro models based say on established long run marginal that offer little guidance on the specific mea- cost (LRMC) pricing principles, served simply sures to be undertaken at the sectoral level. to depress demand (relative to subsidized In this chapter we analyze and quantify prices), then merit order dispatch would be representative examples of both types of expected to result in reduced use of older, less linkages using Sri Lanka as a case study. First, efficient thermal plants, and therefore pollut- as an example of a sectorwide policy with ant emissions from thermal plants would potentially broad environmental certainly be lower. Lower demand, however, consequences, we examine electricity price will also affect the least cost expansion path, reform. Second, as an example of a direct such that the addition of more efficient units is environmental constraint with broad energy delayed. This may result in older units being sector and economywide consequences, we dispatched more than would otherwise be the analyze options for greenhouse gas emission case, resulting in an increase in emissions. reductions to meet possible targets externally The extent to which the latter effect is in fact mandated by the World Climate Convention. offset by a lower generation requirement is an empirical one, and can only be answered by Electricity pricing reform looking at a specific case. The economic rationale for efficient pricing Part of the problem in estimating the bene- of electricity is well established, and has been fits of pricing policy in an actual case is also documented in many studies.' Reforms to analytical. Financial models of the electric eliminate subsidies, and to raise tariffs in power sector have long been used in project order to improve the financial health of elec- appraisal work, and it is standard practice to tric utilities, have been proposed and imple- use such models, typically Lotus 123TM 2 spreadsheets, to examine the financial impacts mented in numerous instances . Indeed,cove of alternative tariffs and financial covenants. In the typical case, one examines (either by trial and error, or by an appropriate iterative 'For a recent review, see, for example, Munasinghe computation) what the tariff must be to meet 1990. 2A recent review of such lending found that electricity tariff reform was an extremely common component of adjustment policy packages-see, e.g., Honduras, Panama, Venezuela, Bulgaria, Warford et al. 1992. Between FY88 and FY92, tariff Czechoslovakia, Egypt, Hungary, Poland, Rumania, reforms were part of structural adjustment lending in Benin, Burundi, Ivory Coast, Rwanda, Senegal, and Bangladesh, Laos, Papua New Guinea, Guyana, Zimbabwe. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 133 some specific target for the rate of return on continue to focus on the United States3 and equity or on fixed revalued assets, or for the other Organisation for Economic Co-operation percentage of self-financing of the future and Development (OECD) countries. Model- investment program. These estimates then ing studies fall into two general categories: become part of the financial covenants that are "top-down" equilibrium models (with varying entered into as a condition for lending by an degrees of disaggregation),4 and "bottom up" external donor. models, typically simulation models with Unfortunately almost all the standard finan- much engineering detail. cial models take the technical data as exoge- The main focus of much of this modeling nous inputs derived from other, more detailed effort is to assess the potential economic engineering models. For example, the fuel bill impact of GHG emissions reduction options, in the income statement is taken from a pro- particularly in relation to the commitments duction costing model, the debt service as- made by many developed countries to stabilize sumptions are taken from the capacity expan- GHG emnissions at their 1990 levels. Hourcade sion model, and so forth. It thus becomes et al. (1995) provide a comprehensive review extremely difficult to assess the feedbacks that of this literature in the forthcoming Report of arise from changes in the tariff, because the the Intergovernmental Panel on Climate detailed models that provide the technical Change (IPCC). They conclude that top down inputs are not directly linked (and in many model generally indicate larger economic utilities they are run by different departments impacts than bottom-up models, and that in with little or no attempt to ensure mutual OECD countries the annual cost of stabilizing consistency). emissions at their 1990 levels would typically exceed 1-2 percent of gross domestic product Greenhouse gas emissions (GDP).' They also note that whatever lost-cost With the agreement on the World Climate Convention having been signed by over one hundred countries at the recent earth summit 3Manne and Richels (1990), for the case of the in Rio de Janeiro, the practical problems of United States, Goto and Sawa (1993), for the case of how the obligations to reduce greenhouse gas Japan, or Proost and Regemorter (1992), for the case of (GHG) emissions can be met in the most cost- Belgium, typify the genre: time scales that run in some cases to the year 2100, new technologies whose earliest effective manner requires urgent attention, introduction at a significant level is 2010 and beyond, particularly in developing countries that have and highly aggregated sectoral descriptions of the limited resources. Simply put, there exists a economy. For example, in the case of Global 2100, a wide range of potential policy options, ranging variant of the earlier ETA-MACRO model, the system from direct technology substitution (e.g., has three sectors: goods, electric energy, and delivered replacing coal-buring power plants with wind nonelectric energy. The production function for goods is a hierarchical system with a constant elasticity generating plants) to indirect pricing and combination of Cobb-Douglas value added and energy taxation policies (such as the imposition of inputs. (See also Hogan, 1990, for a succinct review.) carbon taxes on fuels, or the imposition of 4For a systematic review of top-down studies in emissions charges). How this wide range of the United States, see EMF (1993). Well-known options is to be evaluated on a consistent models of this type include Global 2100 (Manne and Richels 1990) and the GREEN model (OECD 1993). bThe results broadly tend to show that the losses assessed in a systemic manner, is still subject to GDP growth resulting from a fairly wide range of to considerable question. carbon emission restriction scenarios are rather small: Although the number of GHG mitigation as noted by Hogan (1990), the Manne-Richels cost studies continues to grow, most studies calculations suggest that stabilizing CO2 emissions at 80 percent of the 1990 level would account for about 5 percent of the total GDP through the next century. 134 VOLUME II emission reduction options may exist for dies, these studies estimate the level of reduc- developing countries, they are likely to be tion in GHG emissions by application of insufficient to offset increasing emissions assumed price elasticities to the difference from economic growth. between the subsidized and unsubsidized In what appears to be the first such analysis price. Burgess (1990) uses the difference of GHG emissions for a developing country, between actual average cost of electricity and Blitzer et al. (1993) apply a general equilib- the estimated long-run marginal cost (LRMC), rium model to Egypt-with six nonenergy applies an assumed long-run price elasticity of sectors, and four energy sectors-crude oil, -1, and estimates the reduction in GHG emis- natural gas, petroleum products, and electric- sions for eleven countries, including the ity. GDP losses-ranging from 4.5 percent for United States, China, India, and some small a 20 percent reduction in annual carbon emis- developing countries, such as Tanzania and sions to 22 percent for a 40 percent reduction Peru. Not surprisingly, the bulk of the total over the base case-are much higher than carbon emission savings of 124 million tons those estimated for the developed countries. per year (mt/y) come from coal fuel savings, Yet even with these higher GDP impacts, of which India accounts for 11.9 mt/y, China there is still a significant increase in actual 26.6 mt/y, and the United States 85.4 mt/y. CO2 emissions. Moreover, these results are for Larsen (1993) does the same analysis, but a country for which the assumed backstop from the perspective of fuel prices, applies technologies-natural gas-powered transport estimated own- and cross-price elasticities for and nuclear power-are feasible options. Sri the different fossil fuels to the difference Lanka, like many low-income countries, has between an appropriately adjusted border no natural gas resources of its own, and nu- price and the domestic subsidized fuel price, clear power is not an option for the foresee- and, more significantly, includes the countries able future. of the formner Soviet Union (FSU) and Eastern Indeed, there is no chance whatsoever that Europe. In this analysis, the former (FSU) a country like Sri Lanka could stabilize its (917 mt/y) and Poland (105.2 mt/y) dominate C02 emissions at the present level, or even at the results; indeed, the combined estimated twice the present level. Given the low existing impact in India (54 mt/y), and China (45.4 level of per capita commercial energy use, mt/y) together is less than that for Poland. future demand growth to support economic Anderson and Williams (1993), using a simple development needs will be high. Therefore, demand model and an assumed -0.5 value for even if electricity demand growth could be the price elasticity of demand, estimate that entirely accommodated by renewable energy total subsidies due lo unrealistic electricity (which, in light of projected electricity de- tariffs in developing countries amount to more mand growth of 7 percent to 9 percent per than US$100 billion, and that improvements year, requires hlroic assumptions about the in price and managerial efficiency would feasibility of building large hydro plants), the reduce pollution in developing countries by up anticipated growth in transportation sector to 40 rercent. -- fuel consumption, even uimder highly optimis- Such studies certainly provide some esti- tic expectations about mass transit and fuel mates of the efficacy of pricing instruments, efficient vehicles, precludes stabilization of but tell us little about the actual impacts, and CO2 emissions in the near future. therefore also the barriers, likely to be encoun- A somewhat different approach has been tered if indeed the policy reforms were to be taken by Burgess (1990), Larsen (1993), and implemented. Clearly, as relative fuel prices Anderson and Williams (1993). Addressing change, hydro-electricity may become more merely the impact of eliminating price subsi- attractive for electricity generation, which has 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 135 its own set of significant environmental im- challenge, insofar as most energy models pacts. Moreover, while shifts from coal to presently available tend to focus on only part natural gas may well be desirable not just of the problem. Most attempts at providing an from the perspective of CO2 emissions, but "environmental impact assessment" capability also from the perspective of reducing SO2, to energy models are little more than add-on NO, and trace metals, that is small comfort for modules to calculate residuals (Markandya countries such as Sri Lanka that have no 1990). In the case of GHG emissions, that natural gas resources of their own. In any such models are limited to residuals is perhaps event, countries for which imported fossil fuel not very serious, but certainly there is very is the least cost fuel for baseload thermal little reason to believe that the actual impacts electricity generation (as is again the case for of other pollutants, such as SO2 and Sri Lanka) have no fossil fuel price subsidies particulates, are linearly related to emissions to eliminate. levels. More serious from our perspective, In fact the most widespread approach to however, is the fact that such models are of exploring policy options for GHG emissions limited use to assess price and tax policies on reductions in developing countries is the a consistent basis, since they generally lack bottom-up approach largely focused on tech- integration with financial models, and energy nology options (see, e.g., Van Dang (1993) for demands tend to be defined exogenously. Thailand; TERI (1992) for India or Moreira et Several additional methodological prob- al. (1992) for Brazil)6 While these provide lems relate to the workings of the Global useful guidance, none compare technology Environment Facility (GEF). This is the fund- interventions with pricing and tax policy ing mechanism, on an interim basis, to pro- options, and most are based on rather broad vide for financial resources needed by devel- scenario exercises rather than the sort of least- oping countries to meet the "full incremental cost planning models that are actually used for costs incurred" in complying with their obli- electricity planning. gations under the Convention on Global Yet any consistent assessment of policy Climate Change. options for GHG emissions reduction-even if limited to bottom-up examinations of tech- (a) If the GEF is to fund the "incremental nology interventions-requires a comprehen- costs" of GHG reduction measures, this sive representation of the technical, environ- presupposes that a baseline can be un- mental, economic, and financial characteris- ambiguously defined, against which tics of the energy system.7 This is a major such incremental costs be defined. One obvious approach here is to define the baseline as the "least cost" solution to 6For a review of the country studies being expansion of the energy system. As is undertaken by the United Nations Environment now fairly well recognized in the case of Program (UNEP), see, e.g., UNEP (1992). A more electric sector expansion, however, the recent status report on country studies is to be found in concept of "least cost" is extremely Fuglestvedt et al. (1994). conce ofs"lat 1989" ieremely 7Particularly in the case of electric sector fragile (Crousillat 1989; Meier 1990). technologies, the proposition that technologies can be Such a solution may be valid only for a evaluated in isolation from the systems context is very narrow band of input assumptions, extremely tenuous: the impacts of different and if these assumptions prove to be technologies cannot be linearly superimposed because different, then an investment program of the interactions that occur through the system load curve. For example, the impact of load management preicated u e "est cost"mpla and transmission and distribution (T&D) loss reduction may ultimately be distinctly nonoptimal. when implemented together is demonstrably less than In short, any deterministic definition of the sum of the impacts of each evaluated alone. 136 VOLUME 11 "incremental cost" will run into a range Again the growing literature on the benefits of of practical operational difficulties. GHG emissions control is thoroughly re- viewed in the forthcoming IPCC report by (b) Equally difficult is the treatment of Pearce et al. (1995). "joint products." Options that reduce GHG emissions may also provide signif- Approach icant changes in other pollutants whose In this chapter we address these various impacts are of a quite different scale. For analytical and methodological issues by the example, the substitution of renewable application of the ENVIROPLAN model, an energy technologies for coal will reduce integrated energy sector model (see Figure not jUSt CO2 emissions, (a global bene- 6-I). At its core is a fairly conventional finan- fit), but also SO2 and particulate emis- cial representation of the electric utility and sions that bring a reduction in local the repry-with the eleic state- environmental damages. On the other the refinery-with the usual income state- enirnd,mtheintal magedus. on then hew e ment, sources and uses of funds, and balance hand, the increased use of renewable sheet. The model allows the user to specify energy technologies (such as hydroelec- one or more balance sheet ratios as the crite- tric generation), which also reduce CO2 nron for setting the level of the tariff-such as emissions, may also impose new and dmifferentslocai environ mentalcls posts (h areturn on equity or assets, self-financing ratio, di1fferent local environmental costs (such an so on h iaca odlihnlne as loss of biodiversity associated with and so on. The financial model is then linked seeviinundation) 8 to technical representation of the energy reservoir sector, with demand and supply modules for Estimates of the importance of joint bene- each major subsector-in the case of the electric sector, for example, including merit fits and costs ary w id (i par ecaus order dispatching and capacity expansion, and valuing the costs an benef t o vi- driven by an econometrically specified de- ronmental impacts may be subject to similar mn oe ta ntr ae rcsfo h difficulties of valuation and scientific uncer- mand model (that in tum takes prices from the tainty as for GHG emissions), but most lie in current tariff in the financial statements). The the range of US$1 to US$10 of additional model is closed by passing the investment requirements for caDacity expansion back to benefits for every dollar of benefit from GHG .qu y en the financial module. Thus, adjustments in the emissionyAlfsereducton. reentad Nlomsrwg capacity expansion plan, say in response to study 'Afsn Br.eon an. oso lower demands caused by higher prices, feed 1992) indicated that when the joint products back automatically into the asset and liability of carbon reduction were taken into account (such as reduction in environmental damages accounts in the financial statements. We thus (u aorests reductionein henvironmen damages,r achieve a completely consistent set of prices, to forestson , heal damages, reded demands, technical sector configuration, trafioc congestion, road damage), sh enft investment, and financial representations of go a long way toward compensatingthe ec- the major sector institutions that permits simulation of a wide range of policy options. A multiattribute environmental analysis is KTese methodological problems have been integrated into the model. While in this chap- recognized by GEF, which has initiated a research ter, the focus is on CO2 emissions and air program to find an operational approach for measuring quality and related health impacts, elsewhere and agreeing upon full incremental costs within the (Meier and Munasinghe 1994), using the same context of the Climate Change Convention: the so- methodology and model, we develop tradeoff called PRINCE study (Program for Measuring curves for a much wider range of environmen- Incremental Costs for the Environment). For a full tal attributes (including biodiversity and acid discussion, see, e.g., King (1993a) and King (1993b). 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 137 rain indicators). The model is described in limited to smaller run-of-river schemes, such detail in Meier (1995).9 as Broadlands, and the run-of-river variant of The main limitation of the analysis reported the Kukule project. Yet even some run-of- herein is that common to all partial equilib- river schemes will likely encounter fierce rium models, the results do not take into opposition: the proposed Talawakelle run-of- account any macroeconomic impacts and river plant, part of a proposed hydro develop- feedbacks. In a second stage of our work, ment scheme on the Upper Kotmale, would currently under way, this limitation is being cut off river flow from the St. Clair's water- addressed, in which the energy sector model is fall, a well-known scenic attraction. being linked to a general equilibrium model of Nuclear power is not likely to be a feasible the Sri Lanka macroeconomy. option for Sri Lanka for some time, even aside from the environmental issues. Present indica- The Sri Lanka electric sector tions are that even under very high rates of demand growth, it would be at least 2030 The2 preenaatinstalle capacity in Sri Legan before the system is of a sufficient size to is 1,225 megawatts, of which 1,115 mega- acomdt5060mewtsnulr watts is hydro, the balance combustion tur- bines (108 megawatts), diesel (64 megawatts), unit. ' ' ~~~~~~There are also substantial environmental and an oil-fired 44 megawatt steam plant. T Generation for 1992 was3377concerns associated with the introduction of Generation for 1992 was 3,377gigawatt-hours, colfrdpwrsaininSiLk.Th with a T&D loss rate of 18.8 percent (of environmental imat a n prepare generation). The rate of demand growth over the next decade is expected to be about 7-9 a proposed coal-burning station at percent per year. Assuming no fresh discover- Trincomalee, in the Northeast, revealed a ies of oil or natural gas, the conventional range of important issues ranging from im- wisdom is that for the next twenty years the pacts to the aquatic ecosystems of generation options for Sri Lanka are quite Trincomalee Bay caused by the discharge of generation otilons for Sri Lanka are quite thermal effluents, to air pollution concerns limited. Whldbe substantial thydropowe re caused by particulate and sulfur oxide emis- sources could be developed, the remaining sites are limited in size, and have substantially sions. In short, expansion of the power system to higher costs Than lealtcosteatvem oxptnsin (meet future electricity needs of the country Table 6-1).. The least cost system expansion will have significant environmental implica- studies conducted annually by the Ceylon Electricity Board (CEB) show that the least tions. Even if appropriate mitigation measures cost plan is to begin building coal-fired are taken at the project level, sectorwide policies must also take account of environ- coal by the end of the decade. mental implications in order to assure sustain- The larger hydro plants, such as the high able development. dam variant of the Kukule project, involve substantial resettlement and removal of forest HG reduction issues for Sri Lanka area, and are likely to be fiercely opposed by Carbon emissions in Sri Lanka, both in environmentalists. As a matter of practicality, absolute terms, as well as in per capita terms, expansion of conventional hydro will be are presently still quite low, a reflection of the dominance of hydro in the electric sector, and low energy intensity of the industrial sector.'0 9The electric sector version of ENVIROPLAN is currently being used by a number of other utilities for integrated resource planning (IRP), including BCHydro "'1993 emissions of carbon at some 6 metric tons in Canada and Andhra Pradesh in India. per capita in the United States, 2 tons per capita in 138 VOLUME II However, beyond the year 2000, CO2 emis- rapidly growing transportation sector, which sions will rise very sharply as the electric accounts for about half of the pe-troleum sector generation mix moves toward fossil product consumption. fuel, as indicated on Figure 6-2. GHG emis- The impacts of global warming that might sions are expected to triple over the next be experienced by Sri Lanka are quite uncer- fifteen years in the absence of any focused tain. Some researchers expect an intensifica- policies to reduce GHG emissions. This has tion of the monsoon in tropical latitudes, important consequences for the negotiating which, in Sri Lanka, would adversely affect posture of Sri Lanka in the event that a con- soil erosion and stability in the hill country sensus emerges for tradable CO2 emission watersheds where deforestation rates already rights. Clearly, it will be in the interests of represent a threat to the sedimentation rates small developing countries whose present experienced by hydro and irrigation reservoirs. electric systems are hydro dominated to argue Unlike many other countries in Southeast that the initial allocation of emission rights be Asia, the tectonic conditions in Sri Lanka are based on population, or income, rather than on relatively stable, with little significant seismic present fossil fuel consumption. activity of the type that has produced signifi- The future course of GHG emissions in Sri cant surface depressions of coastal areas in the Lanka itself will be largely determined by Philippines.2 Nor are the major cities pres- policy and investment choices in three main ently threatened by major subsidence prob- areas. First is the generation mix in the elec- lems caused by excessive exploitation of tric sector-clearly, the extent to which coal groundwater. Nevertheless, extensive areas of (and other imported fossil fuels) are used as a the coast, especially in the south and south- major fuel, and how much of the remaining west, are already threatened by coastal erosion hydro potential can in fact be exploited, will (that can, at least in part, be attributed to large- have a major impact on CO2 emissions. Sec- scale mining of coral reefs), and there are ond is the traditional fuels sector-biomass extensive areas of highly populated coastal still accounts for two thirds of energy con- areas that would be severely affected by sea sumption in Sri Lanka-and any shift from level rises of 1-3 meters. One of the immedi- biomass to petroleum-based fuels will have a ate consequences of sea level rise is likely to major impact on CO2 emissions." Third is the be contamination of the limestone aquifers that are important sources of groundwater in the north.'3 Europe, and between 0.1 and 0.5 tons per capita in most developing countries of Asia. We estimate per capita emissions in Sri Lanka at less than 0.1 ton per capita. Estimates of carbon emissions per unit of GNP, expressed in kilograms per million U.S. dollars, are about 1.3 for China, 0.7 for India, and 0.25 for the United States. We estimate the corresponding figure for Sri Lanka at about 0.15 (see, e.g., Banuri et al. (1995); Baron and Hills (1990)). "2An earthquake on the east coast of Luzon in "The main causes of deforestation are land clearing July 1990, for example, caused a 2.5 meter fall in large for agricultural and commercial timber operations, areas of the coast, requiring large scale evacuation of rather than fuelwood demand per se. Indeed, a coastal villages. substantial fraction of the biomass energy consumption '3For a good review of the likely impacts of is made up of coconut husks, rubber wood (old trees global warming in the South Asian countries, see, e.g., that are replaced by replanting), and other agricultural Pachauri (1991). It would seem that the impacts of sea wastes. However, if deforestation continues at its level rise, however serious for Sri Lanka, will not have historical rate, the supply of fuelwood may become the cataclysmic consequences that seem likely for the seriously constrained in the future. Maldives and the East Coast of Bangladesh. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 139 The options tion as to whether even this target can be Wind energy attained. ' We define three policy cases for analysis: in the first (indicated on the figure by Although there are a number of renewable "T&D(+)"), we bring losses to the CEB target energy technologies that may be considered by 2001; in the second (indicated by for Sri Lanka, the technology with the largest "T&D(++)"), we bring down losses further to potential impact on the operations of the CEB 10 percent overall by 2010; and in the third, a is wind energy. Utility scale solar thermal there is a delay in achieving the 12 percent plants are still extremely expensive, while the target to 2010 ("T&D(-)"). use of photovoltaics probably has greatest It might be noted that LECO, the privatized potential in remote rural areas that do not have distribution company, appears as if it will grid access (Wijeratne 1994b). The wind meet its 8 percent loss target by 1996. How- feasibility assessment recently completed for ever, since LECO buys its power from CEB at the southern lowlands of Sri Lanka estimated 33 kV, 8 percent would not be a reasonable the ultimate potential at about 300 megawatts. target for the system as a whole, which also In this study we assume that six 50 megawatt needs to include losses in the high-tension wind farms will be built in this area during the transmission system. period 1998 to 2006 (CEB 1992c). The most important finding of the monitoring program Demand-side management (DSM) conducted at the site concerns the hourly A systematic assessment of DSM options variations, with a peak during the afternoon has yet to be conducted in Sri Lanka. Indeed, hours. There are also very strong seasonal comprehensive DSM assessments have been variations. During the summer monsoon conducted for only a very few developing period, monthly outputs for a 50 megawatt countries to date. Nevertheless, there are facility are estimated at about 10 gigawatt- idcatrios th date Nthe lestsom e s an hours per month, falling to a low of about 2.5 indications that there exist some significant gigawatt-hours per month,fallig in Marf anNo- 2opportunities for the introduction of energy- vemberttConrspeque ontl, wind plhants w do- efficient, end-use technologies in Sri Lanka. Both the World Bank power system efficiency little to reduce installed capacity requirements, study of Sri Lanka (World Bank 1983) and the but will serve primarily to displace thermal electricity Masterplan (GTZ 1989) made some power and reduce expected deficits in dry preliminary estimates of the potential for load years. management and energy conservation by the T&D loss reduction systematic replacement of incandescent lights by fluorescent lights.'5 As can be seen from Figure 6-3, the prog- ress in reducing losses to the present 12 per- "In fact, actual loss rates are likely to be cent target set by CEB has been much slower somewhat higher, since CEB treats HT sales to the than anticipated. As recently as 1989, it was Lanka Electricity Company (LECO), a privatized projected that the 12 percent target would be (although still government-owned) distribution attained by 1993, yet by 1992 that target had company established in 1984, as a consumer. Since been delayed to 1996, and by 1993 to 2012! In LECO has losses of its own (and some of the municipal the 1994 study, the target year for achieve- systems that were absorbed by LECO had loss rates exceeding 30 percent prior to rehabilitation), total ment of 12 percent was advanced again to losses in the system are at least 2-3 percent higher than 2004 (which is the assumption used in our CEB's estimate based on its generation and sales. reference case). Given the lack of progress 5The World Bank study also examined the over the past few years (in 1989 and 1991 possibility of replacing self-ballasting mercury vapor losses actually increased), there is some ques- and incandescent lamps used for street lighting by high- pressure sodium vapor bulbs. 140 VOLUME II In this study, we examine two specific Mini-hydro technologies to improve the efficiency of The electricity Masterplan identified a electricity utilization: increased penetration of series of potential mini-hydro sites. In this compact fluorescent lighting (in the domestic option, we assume that the four with the and commercial sectors)," and the introduc- l tion of energy-efficient refrigerators. 17 The lowest specific generat1on cost w fll be tmple- former is largely a load management measure, mented in the 1998-2002 time frame, to ' provide a total additional hydro capacity of 30 with very large megawatt savings during the provi evening peak, but relatively modest energy megawatts. savings, while the latter is largely just an Clean coal technology energy-saving measure, with little impact on the peak. Sri Lanka has a very pronounced Conventional pulverized coal plants have evening peak that coincides with the hours been part of CEB's expansion plans since the immediately after sundown, indicative of the mid- 1 980s. Since the new clean coal technolo- importance of the lighting load. 8 gies not only provide substantial improve- ments in emissions, but also gains in effi- ciency, we have examined the impact of using "6For a review of the European experience, see, e.g., combined cycle-pressurized fluidized bed Mills (1991). The average cost of conserved energy of combustion as an alternative, under two as- these programs is reported at about 2.1 cents per sumptions about capital costs.'9 kilowatt-hour (c/kWh), including 0.3 c/kWh for indirect administrative, promotional, and evaluation costs: this No coal is significantly below the cost of new electric power plants. An analysis for Pakistan (Miller, Geller, and de A "no coal" option has been evaluated by Almeida 1992) concludes that even if the government the CEB in its annual generation planning offered a 100 percent subsidy-i.e., literally gives them away-it would save US$ 10 per bulb in avoided power studies for some time-although the rationale plant construction costs. had less to do with GHG emissions than with '7Tmproving the energy efficiency of refrigerators the extent of public opposition to large ther- appears as a top-ranked option in many assessments of mal plants located in the coastal zone. In this demand side management measures in developing countries. For example, this measure emerging as the o best among 23 options, along with improvements to would be replaced by steam cycle, residual compressors and insulation of Thai refrigerators, are oil-fueled plants at the same locations. Resid- estimated to reduce unit electrical use from 400 kWh to below 200 kWh per year (Florida Solar Energy Center 1991). Indeed, an analysis of efficiency improvements '9For a recent review of technology in major residential appliances in the United States developments, and their application to developing (Schipper and Hawk 1991) indicates that refrigerators countries, see, e.g., Tavoulareas and Charpentier and freezers have shown the best improvements in (1995). The capital cost estimates are based on EPRI efficiency since 1972, improving 70-90 percent, (1993). The optimistic estimate of pressurized fluidized whereas efficiency improvements of other major bed combustion (PFBC) is taken at the Electric Power devices have been much more modest (air conditioners Research Institute (EPRI) values: the average cost of 33 percent, space heaters 10-25 percent). PFBC plants analyzed in that report is US$1176 per 8It must, however, be stressed that there is little kW (in December 1992 dollars, excluding interest empirical data on the composition of the hour-by-hour during construction), which is only US$50 per kW loads by end-use device. However, peak air more than conventional coal plants without flue gas conditioning loads almost certainly occur during the desulfurization (FGD). The pessimistic estimate is daylight hours (since the bulk of air conditioning is taken as US$350 per kW higher. With a high installed in offices and hotels rather than private proportion of the total fuel cost accounted for by residences). Consequently, the assumption that a transportation (from Australia or South Africa), the significant share of the evening peak is attributable to efficiency gain provides an important offset to the residential lighting seems reasonable (Wijeratne 1993). higher installed capital cost. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 141 ual oil is assumed to be imported from Singa- tal concerns such as lead and particulate pore spot markets, using the 1993 World bank emissions, have begun to emerge. Urban air oil price forecast.20 quality is a priority issue in the National We examine several oil price cases. Burn- Environmental Action Plan (Ministry of ing high-sulfur fuel oil-about 3.5 percent by Environment 1991), and the Metropolitan weight is the specification on Singapore spot Environmental Improvement Program has markets-without sulfur removal may be recently issued an action plan for Colombo in problematic: when corrected for calorific which the transportation sector is a major value, this would be equivalent to about 2 target (Metropolitan Environmental Improve- percent sulfur coal (in contrast to the 0.6 ment Programme 1992). Combustion of gaso- percent sulfur coal that is planned in Sri line and auto-diesel is presently also the larg- Lanka). The differential between 0.3 percent est source of carbon emissions in Sri Lanka, low-sulfur fuel oil and high-sulfur fuel oil in and even if large-scale generation of electric- Far Eastern markets is quite varied. Over the ity from coal commences in the late 1990s, the past five years, the difference has ranged from transportation sector will continue to represent US$3.70 per barrel (bbl) in 1989 to as little as a major source of total national GHG emis- US$1 per bbl in 1994. We also examine a sions. high world oil price case (in which the price In a related study, we examined the entire rises to US$30 per bbl by 2004, as opposed to set of urban air quality improvement measures US$21.5 in the base case). being proposed by the Colombo Action Plan that range from reduction of the proportion of Maximum hydro two-stroke motorcycles (which have higher emissions than four-stroke motorcycles) and a by maximum rseductio of GHGrop ts. i usa ved olowering of the lead content in gasoline, to the by maximum use of hydro plants. Because of itouto fvhceiseto n ane the high costs, this again has to be forced into the solution. In this scenario, all the remaining nance programs (VIM), particularly for diesel major hydro plants in Sri Lanka are assumed vehicles. All these measures are justified on mo be built by 201 Si L the basis of local urban air quality concerns, Transportation sector measure-vehicle but a number of them prove to have signifi- cant GHG reduction Impacts as well. We inspection and maintenance programs: Over inc in tis n ispthe as ortation the past decade, motor vehicle traffic in Sri y . . Lanka has risen very fast. Particularly in the sector measure that has the most significant Colombo metropolitan area, problems of GHG impact, namely VIM. traffic congestion, and of related environmen- Electricity pricing reform As noted in the introduction, the rationale 20For example, in 2000, according to these for and importance of economically efficient projections, the delivered cost of residual fuel oil, c.i.f. pricing is well established and has been docu- Sri Lanka, is US$13.90 per gigacalorie (based on mented in many studies. Historically, Sri US$150 per ton marker crude price, US$135 per ton for Lanka has followed the same cyclical pattern residual oil c.i.f. Sri Lanka), as opposed to US$7.58 per gigacalorie (US$50.5 per ton) for imported Australian observed in many other developing coun- coal. tries-in which long periods during which the 2"Including the high dam version of the Kukule government has been reluctant to raise tariffs project (at 144 megawatts, rather than the 70 megawatt results in a gradual deterioration of the finan- variant that is currently under consideration), both . .. plants in the Upper Kotmale scheme (rather than just cial condition of the utlity, i turn followed the lower run-of-river project at Talawakelle), and the by sharp increases necessary to ward off a Uma Oya multipurpose scheme. crisis. Thus, for example, between 1972 and 142 VOLUME II 1978, there were no changes in the tar- of return on assets (or equity) as iff-resulting in a gradual decline in real well-although prior to 1984, when the fuel electricity price level. This was followed by a adjustment clause was invoked,24 financial series of very sharp rate increases in the period performance was subject to sharp deterioration 1978-80 (see Figure 6-4). Prices drifted during periods of drought as a consequence of downward again in the early 1980s, with a heavy use of thermal plants dependent upon correction in 1988. Yet again they drifted imported fuels. Thus, the drought year 1983 downwards until 1992, when a further series resulted in heavy financial losses. Neverthe- of significant tariff increases occurred. less, since 1983 the real price has drifted In addition to the expected and obvious downward, as has the return on equity (from a impact upon the financial condition of the post-drought peak of 4 percent in 1985 to less utility, the failure to maintain a consistent than 1 percent in 1991. These returns are pricing policy has a number of possible envi- substantially below normal target levels of ronmental impacts as well. When tariffs are 5-10 percent. substantially below long-run marginal cost (LRMC), it becomes difficult to raise the The price elasticity of demand resources necessary to expand and upgrade the Is electricity demand in Sri Lanka in fact T&D system, resulting in turn in high loss rates characteristic of overstressed systems. price elastic? Some load forecasting models used by the CEB do not in fact include a price This is compounded by wasteful consump- variable.25 And, simple econometric models tion-and hence also higher levels of genera- do not show any statistically significant evi- tion-where price levels are significantly dence of demand elasticity (Meier et al. 1993). below their economic level. The result is that Indeed, in order to derive a satisfactory econo- the environmental impacts associated with m m electricity generation are higher than they ercmdlItboesnesayofit electric g ent areh. With they examine the underlying structure of electricity would be in an efficient system. With the newdea.Ovrtepstwdcds,emn projects becoming increasingly controver- grhdemand. Over the past two decades, demand sial-hydro projects because of the impacts gP growth h exaerien rcent per year. Real associated with inundation, and fossil-fueled 197 with the liberalization of the economy projects because of impacts associated with air starte the incominayawofde admi pollution, solid waste disposal and thermal . . . effluents-eliminating those environmental tration, but civil disturbances in the late 1980s efflunts-limiatin thos envronmntal caused a decline in both economic growth and impacts associated with inefficiency becomes i'cresingl imporin the growth rate of electricity demand Elctre icity i (which was negative in 1989). Perhaps the most important underlying cally been at levels substantially below trend, however, has been the fall in kWh per LRMC.2 In 1992 a series of tariff increases customer account. For the industrial sector, raised the average price to Rs 2.6 per kWh; this is a sharp and consistent trend, indicative however, even this price level is below the long-termhsructurl tments in the latestestimate ofthe CEB (Table 6-2).23 of long-term structural adjustments in the lates estiate o theCEB (able -2).industrial sector, with many small medium- to It thus comes as no surprise that the finan- cial performance has been below target levels I24n fact, a fuel adjustment clause was introduced into the tariff in 1978, but was not implemented until 22The first rigorous estimates of LRMC for Sri October 1980 due to difficulties in calculation and Lanka, made in 1982, are by Munasinghe-see Chapter assessment. For a discussion of electricity pricing in the 9-of Munasinghe and Schramm (1983). 1970s, see World Bank (1982). 23Ceylon Electricity Board (1992a). 25See, e.g., CEB 1992b. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 143 small-sized light manufacturing enterprises In sum, it is inevitable that the combination taking the place of large state-owned facilities of special circumstances that has prevailed as the dominant consumers. since the mid- 1980s will make econometric The trend is observed also in the household analysis more difficult. Structural changes in sector, with a consistent decrease from the manufacturing, takeover of municipal sys- peak consumption of 1.2 megawatt-hours tems, civil disturbances, and severe droughts (MWh) per account per year in 1983 to 0.9 in 1983 and 1987 that resulted in significant MWh per account per year in 1991. For com- curtailments all combine to make this an mercial accounts, there is no discernible trend unusual period28 in 1987-88, but then there is a sudden fall in the period 1988 to 1991. This corresponds Pricing scenarios precisely with the onset of significant civil To examine the impacts of alternative disturbances, during which shopping hours pricing policies, we define a series of pricing were dramatically curtailed in the evening. It scenarios. A first case, obviously, is what remains to be seen (once the 1992 data are be terme case, as usl, in what compiled) whether shopping habits and store might be termed iness as u o intwic we set the tariff in such a way as to maintain hours have returned to previous patterns now the historical average of 3 percent return on that normal conditions prevail in most of the equity. This has a rather small chance of country. One might note that these declines in realization, for it generates insufficient cash to consumption per connection are not obviously also maintain satisfactory self-financing ratios. related to price: since in the period 1988-91, Consequently, we define a second "base real prices decreased. case" tariff policy based upon a postulated A further significant distortion of the aggre- covenant that requires an 8 percent rate of gate data has occurred as a result of the grad- ual ~ ~~ . taevro.oclmncpliisb. h return on equity. This reflects typical World ual takeover of locamuBank practice for power sector lending and, as CEB and by the Lanka Electricity Company is evident from Figure 6-5, implies a substan- (LECO), a private sector distribution company tially higher rate of return than has been established in 1983 to take over some of the achieved at any time over the past twenty more egregiously run municipal authorities on the perimeter of Colombo, some of which had years. An iterative adjustment algorithm iS used In loss rates of over 40 percent at time of take- over!26 The significant progress made by emodel to dentify t aa tearff level LECO n redcing ossesin it systm is required in each year to maintain the required LnotheC inareduing losseasing its cnsyptipem ibalance sheet ratio. Because of the lumpiness another factor ln decreasing c uin er of capital investment, the year-to-year level of consumer (since, all other things being equal, th taifmyso'osdrbevrain itS wholesale purchases from CEB will be less thtaifmyso cnidrbevito, its lowhesale purchases from CEBi will27 which would not only be hard to implement, the lower the T&D loss rate iS).2 "6However, the shares are still held by the rehabilitation (mostly with Asian Development Bank government. assistance), in which the distribution system in some 27In 1990, the overall loss rate in the LECO service areas was almost completely rebuilt, and an aggressive area was 14 percent. Mahara, taken over in June 1985, policy of disconnection for nonpayment of bills. with a loss rate of 48 percent, was brought down to 13 2"As soon as the model is properly specified, with percent by 1989; Welisara, taken over in December due recognition of the impact of structural changes in 1987 with a loss rate of 41 percent, had losses of 11 customer demand, statistically significant price percent by 1989; and Kolonnawa, taken over in elasticities can be identified. Estimated values lie in the November 1985 with a 50 percent loss rate, was down range of -0.07 to -0.5: only for the industrial sector was to 21 percent by 1989. This substantial improvement in the estimated value (of -0.07) not statistically performance was attained by a combination of system significant. 144 VOLUME 11 but likely send confusing signals to consumers With the current tariff on the order of as well. The assumption here is that an appro- Rs 2.6 per kWh, and the AIC in the range of priate smoothing process would be applied at Rs 6 per kWh, such an LRMC-based tariff the stage of policy implementation. would require a phase-in period. We assume A further option is to base the tariff on the a five-year period, such that the 1994 tariff be LRMC. When the LRMC at each voltage level at least at 20 percent of the LRMC, the 1995 is weighted by the fraction of demand deliv- at 40 percent, and so on, with the tariff at the ered at that level, an "average" LRMC of full level of the LRMC by 1998. Rs 5.29 per kWh results, implying that the Table 6-3 shows the impact of these differ- incremental marginal cost of transmission and ent pricing policy scenarios on some key distribution is Rs 3.68 per kWh.9 indicators. Even to meet all the existing cove- ENVIROPLAN calculates average incre- nants, the tariff needs to be raised to over Rs 4 mental costs (AICs), which is also a measure per kWh. In fact, of the three covenants now used by the CEB generation planning branch in place, it is the current ratio that proves to be to compare system expansion alternatives in binding for most of the next decade (see its annual generation plan.30 The AIC of gen- Figure 6-5). eration proves to be a good proxy for LRMC at the generation bus. Our base case values of Results generation AIC lie in the range of Rs 2.15-2.38 per kWh (depending upon fuel Figure 6-6 displays a plot of system cost" price and capital cost assumptions), very close vs. GHG emissions. Each point represents a to the rigorously estimated LRMC of Rs 2.21 perturbation of the base case with a technol- per kWh (all in 1992 rupees). To obtain the ogy option (open circle) or a pricing option total estimated LRMC to the consumer, used in turn as the level of the average tariff in the financial satements ad the deman forecast-32System cost is defined as the present value of all inanc 3al statements and the demand forecast- future electric system costs using a 10 percent discount ing equation," we add Rs 3.68 per kWh (i.e., rate (the rate used by the Government of Sri Lanka). that component of LRMC attributable to GHG emissions are also discounted at the same rate. T&D) to the generation AIC. The question of the choice of discount rate has of course been fiercely debated by both economists and environmentalists (see. e.g., Markandya and Pearce, 1988), and a review of this literature goes well beyond 29That is, generation Rs 2.21/kWh + T&D = Rs the scope of this study. The consensus view on this is 3.68/kWh = Rs 5.29/kWh overall. that the opportunity cost of capital, with discount rates 30The exact definition used by CEB is the present in the 6-12 percent range, is appropriate for value of total system costs (capital and operating costs environmental cost-benefit analyses (see also the of new facilities plus operating costs of existing forthcoming report of the Inter Governmental Panel on facilities) over the next 15 years divided by the present Climate Change (IPCC), Working Group III, which value of generation requirements over the same period deals with this question at some length). With respect to less the present value of generation from existing hydro. the discounting of GHG emissions, elsewhere (Meier, This is a good approximation of the AIC since in the Munasinghe and Siyambalapitiya 1993) we show on an average hydro year, the current energy demand is empirical basis how the tradeoff curves, and the almost exactly met by the existing hydro plants. resulting policy conclusions, differ as a function of the 3"Ideally, the LRMC at the consumer level would discounting of GHG emissions. The conclusions are (a) also be endogenously calculated by the model: this that the results are much more sensitive to the discount becomes important primarily for the assessment of the rate used for financial flows than for GHG emissions cost-effectiveness of T&D system rehabilitation. and (b) that because discounting of GHG emissions Obviously, if the target level of losses moves from 12 tends to give more weight to emissions in the near to 10 percent, the incremental LRMC for T&D would future than those in the distant future, the rationale for also decline. For the moment, this effect is not short-term policy interventions is more easily considered in our model. demonstrated. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 145 (solid circle). The abbreviations used for each world oil price case. This is a price effect: as option are indicated on Table 6-4. the tariff goes up, demand is depressed, in turn The base case used here-identified as lowering emissions. point "B" in the figures-is based on CEB's That DSM options appear in the "win-win" official generation expansion plan, which is quadrant may come as no surprise, and raises derived on the basis of the WASP model that the obvious question why they are not in- determines the optimal expansion sequence to cluded in the base case, i.e., in the official meet the official load forecast under a given plan. Market imperfections, high transaction reliability criterion (namely, a loss-of-load costs, vested interests, unavailability of fi- probability of about three days per year). The nance, lack of information are often cited as principal conclusion of recent CEB studies is reasons: the fact that many DSM programs that the first 150 megawatt coal-fired unit is that lower total resource costs, but result in required by 2004, with diesels and combustion increased rates, would be another explana- turbines, and two run-of-river hydro schemes tion.33 In fact, in the case of Sri Lanka, while being built in the earlier years. both DSM measures lower total resource The options fall into four types, corre- costs, energy-efficient refrigerators (EERs) sponding to the four quadrants defined by the increase rates very slightly, and CFL lowers base case. The most desirable options natu- rates significantly. In 1994, the Ceylon Elec- rally fall into quadrant El, where both costs tricity Board created a DSM unit, included an and GHG emissions decrease-options de- analysis of DSM options in its generation scribed as "win-win," or "no regrets." Com- plan, and is now attempting to get World pact fluorescent lighting (CFL), T&D loss Bank assistance to begin an active DSM reduction to the 10 percent target, and oil program. steam-electric (OS) under present oil price However, it is also evident that the impact projections are seen to fall in this quadrant. of all of these technology interventions is a On the other hand, options in quadrant I great deal less than the impact of efficient are those for which both costs and GHG pricing. Of course, it could be argued that the increase-the "lose-lose" options. Business- actual environmental impact of the pricing as-usual pricing (3 percent return on equity) effect is largely premised on the assumed and delays in achieving the 12 percent T&D price elasticity of demand, which is admittedly loss target fall into this quadrant. The options subject to some uncertainty. However, this in quadrant II lower system costs, but increase uncertainty is arguably less than the uncer- GHG emissions, while in quadrant IV, system tainty associated with the cost assumptions for costs increase, and GHG emissions decrease. many of the technology options. This empiri- In other words, these are the quadrants for cal analysis of Sri Lanka indicates that the which tradeoffs between the cost and GHG effect of efficient pricing, relative to the his- emissions objectives must be made. torical case, reduces GHG emissions (of the Options that are mutually exclusive are electric sector) by some 28 percent , which is indicated by the dotted lines. Note that the use somewhat lower than the general estimate of of oil-steam cycle plants falls in the "win-win" quadrant for the current oil price projection, 33While the U.S. literature on the appropriate criteria for but moves to the tradeoff quadrant under the selecting DSM programs is large and contentious-see, e.g., high oil price scenario. Note also that the use Joskow and Marron (1992); Hirst (1992); Hobbs (1991)-the of low-sulfur oil (LSOS(-),LSOS(+)), in- question of rate impacts receives only passing mention in creases costs over using high-sulfur oil (OS), most DSM studies for developing countries. In the creases ctPhiltippines Assessment, for example, rate impacts are simply but GHG emissions decrease. There is a fur- dismissed as "modest," and no specific calculations are ther decrease in GHG emissions in the high presented on whether proposed measures might pass a rate impact measure test (see USAID 1994, pp. 2-15). 146 VOLUME II 40 percent suggested by Anderson and Wil- prices34 and construction delays.35 Similar liams (1993) noted above. sensitivity analyses were also conducted to An optimal portfolio may be constructed by test the impact on economic growth, income combining the various options in such a way and price elasticities. Indeed, the main conclu- as to provide a progressive decrease in GHG sion can be stated quite unequivocally: while ernissions (Figure 6-7). (See Annex 6-B for a the magnitude of the environmental benefit discussion of dominance of options and trade- may be debatable because of uncertainties off curves.) We have modified the axes for concerning the price elasticity effect, at worst this plot. The x-axis now includes transporta- (i.e., in the case of no price effect), one can be tion sector emissions, and the y-axis includes certain that there is no negative effect an adjustment (dE) for the savings associated with the vehicle inspection and maintenance Conclusions program (namely, the avoided cost of fuel saved by better fuel efficiency, less the VI1M The following are the main conclusions that program cost). can be drawn from our discussion: We begin with the base case and add the cost-effective measures one by one. This (a) Setting electricity prices to reflect the could be done in a variety of different LRMC has a significant and unambigu- ways-here we select probable ease of imple- ously beneficial impact on the environ- mentation as the order. Thus, among the "win- ment (both in-country and global). The win" options, implementing further T&D loss expected benefits predicted on theoreti- reductions is seen as the easiest, and pricing reform as the most difficult. 340ur base case uses, as does the 1992 generation This can be redrawn as on Figure 6-8: we planning study of the CEB, the 1992 World Bank world show cumulative GHG emissions avoided crude oil price projection, which calls for a gradual over the planning period considered, in order increase (in 1992 dollars) from US$16 per bbl in 1992 of the cost (benefit) per ton. For Sri Lanka, it to US$21.4 per bbl (f.o.b. Singapore) by 2000, and is evident that among options requiring a cost constant thereafter. In our low case ("-wop"), we tradeoff, mini-hydros have the least impact assume that after an initial increase, the price falls, to tradeof, minihydroshave th leastimpact US$15.30 per bbl by 2010. In our "oil shock" scenario, (less than US$2 per ton of CO2 avoided), the real price of crude oil rises sharply to US$38 per while wind plants are seen to be the most bbl in 1999, and US$40 per bbl in 2000; then it expensive, at US$93 per ton. The replacement parallels the trajectory of the 1980s with a slow drift of coal by oil plants (for the current oil price downwards and a sudden collapse in 1985 (2005), followed by a gradual increase back to US$20 per bbl projection) is only US$4 per ton. by 2010. Clearly, history is not likely to repeat itself Extensive sensitivity analyses were also exactly, but such an oil shock scenario is probably a lot undertaken to test the robustness of the con- more likely than gradual changes. clusions about the effectiveness of pricing 751n the litigation delay scenario, we assume that reform, both in terms of additional environ- four years into construction, there is a two-year delay, applicable only to large hydro and coal plants. The mental attributes (such as biodiversity), as rationale is that such plants are the ones most likely to well as in terms of input assumptions. In fact, incur delays: smaller run-of-river plants are assumed the conclusions are remarkably robust with less likely to incur delays, not just because engineering respect to a range of uncertainties. For exam- geological uncertainties tend to be less, but primarily ple, sensitivity analysis results indicate the because they largely avoid the problems associated with resettlement of inundated villages. For example, the benefits of pricing policy reform over wide run-of-river variant of the 70 megawatt Kukule hydro ranges of assumptions about world energy project requires resettlement of some 27 families, while the 270 megawatt Upper Kotmale scheme requires resettlement of some 1,900 families (Japan International Cooperation Agency 1987). 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 147 cal and intuitive grounds are in fact GHG emissions reduction benefits as confirmed by the specific case exam- well. The introduction of vehicle inspec- ined. For example, the difference be- tion and maintenance programs would tween an AIC-based tariff and one based bring significant reductions in both par- merely on achievement of a financial ticulate and GHG emissions. Most of the covenant to assure a 10 percent rate of other road transport sector measures return is a 28 percent reduction of power advocated by the Clean Air 2000 pro- sector GHG emissions over the planning gram for air quality improvement in the horizon, and a 23 percent reduction in Colombo Metropolitan area have a more the health effects related to the exposure limited impact on GHG emissions. of human populations to the incremental ambient concentration of air pollutants. (e) The use of flue gas desulfurization (FGD) systems for SO2 control emerges (b) Pricing policy has a more general impact as a particularly poor option. Indeed, the than physical approaches to demand-side results emphasize the importance of management. DSM programs are diffi- making decisions about FGD systems at cult to implement, and limited in scope. the planning rather than at the project Moreover, even though the results of this level. If indeed FGD systems are to be study appear to suggest that DSM pro- required at coal plants, it may be that grams may be cost-effective, these con- coal plants ought not to be built at all. At clusions are based on very limited actual least in the case of Sri Lanka, a policy data on load shapes. that would require FGD systems results in the least cost system moving to die- (c) While some measures to reduce GHG sels and hydro plants. In any event, new emissions imply a significant increase in clean coal technologies, such as pressur- other local environmental impacts, these ized fluidized bed combustion (PFBC) measures also tend to be very expensive, reduce not only SO2 emissions, but and therefore unlikely to be particulates,andNO,andCO2emissions implemented. The maximum hydro-no (per kWh generated) as well. And fi- coal scenario, which brings very large nally, when one examines impacts (such GHG emissions reductions, would have as the value of the health index), rather a high impact on biodiversity and reset- than just emissions (such as SO2 emis- tlement requirements (because some of sions), it becomes obvious that fitting the larger hydro projects, such as upper FGD systems to remotely located coal reservoir of the Upper Kotmale project, plants with high stacks brings little re- involve reservoirs of significant size). duction in impacts in comparison to However, among those measures that are measures taken at older urban fossil more modest in cost-or even those, like plants (with typically short stacks) that certain DSM measures that w1ll reduce burn high-sulfur residual oil in close costs-there is a coincidence of global proximity to population centers. and local impact reduction. (f) Perhaps the most significant finding (d) A number of transportation sector mea- relates to the potential replacement of sures, advocated on grounds of amelio- coal plants by steam-cycle oil fired rating local air quality impacts, or gen- plants. To date, these have not been eral improvements in the fuel efficiency considered by CEB. The decision to of the sector, prove to have significant focus on coal plants to meet baseload 148 VOLUME II requirements goes back to the early However, if the increase in price is seen 1980s, some years before the collapse of purely as a GHG emissions reduction world oil prices in 1986. This analysis measure, then the relevant comparison is suggests that significant GHG emissions against other GHG emissions reduction reductions are possible by this option, options that have obtained GEF support, and at present levels of oil prices, may such as wind plants, which are certainly even provide economic benefits as well. more expensive. It is of course true-as also shown by this analysis-that if oil prices return to (g) In conclusion, the emphasis given to the US$30 per bbl level, costs would efficient pricing in both project loans increase; the coal strategy undoubtedly and adjustment lending is therefore lowers exposure to this price risk.36 justified not only on grounds of eco- nomic efficiency, but also on grounds of 36To be sure, if other coal-using utilities in the minimizing the environmental damages region were to change to residual fuel oil as a GHG of economic development. This report emissions reduction measure, there may be an upward clearly and explicitly demonstrates that pressure on fuel oil prices. However, most of these efficient pricing makes a significant other countries would be much more likely to switch to contribution to environmentally sustain- natural gas. Moreover, Sri Lanka is perhaps unique in able development. that both coal and oil are imported, and therefore the differential between them is much narrower than in countries such as Indonesia, where the economics of miiine-mouth coal-fired plants makes it very unlikely that a shift to oil for GHG emissions reduction purposes would be likely. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 149 Table 6-1: Specific Generation Costs Capacity, Specific costs, Project/plant MW U.S. cents/kWh Remaining hydro projects Upper Kotmale 150 4.69 Kukule (run-of-river) 70 5.40 Ging Ganga 49 5.30 Belihul Oya 17 6.26 Broadlands 40 7.52 Moragolla 27 7.90 Uma Oya 150 8.12 Thermal Coal (70%PF)[150 MW] 150 6.1 Coal(70%PF)[300 MW] 300 5.00 Diesel (70%PF) 10 5.51 Combustion turbine (20%PF) 22 11.00 Source: CEB Report on Long Term Generation Expansion Planning Studies, 1993-2007, October 1992. Table 6-2: CEB Estimates of the LRMC LRMC Losses, as % Fraction of total Rs/kWh of generation demand (%) Generation 2.21 1.9 HT transmission 3.9 HV 2.63 Grid substation 1.8 HV2 2.75 0.94 Transmission 4.8 MV3 4.23 11.56 Substation 1.3 MV2 5.28 48.75 Distribution 3.5 LV 5.7 38.75 150 VOLUME 11 Table 6-3: The Impact of Pricing Policy Meet all 3% return covenants AIC PV(capital investment), US$m 1,214 1,159 963 Levelized tariff, Rs/kWh 2.82 3.46 5.17 Impacts In year 2005 Fuel imports, US$m 73 64 51 Installed capacity, MW 2,170 2,130 2,050 CO2 emissions, mt/year 4.07 3.63 3.03 Demand, gwh 7,638 7,197 6,624 Tariff, Rs/kWh 3.1 4.2 6.2 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 151 Table 6-4: Summary of Options Examined Option Comments Symbol Technology options Wind energy 305 MW total wind Mini-hydro minihy DSM: energy efficient refrig- EER erators DSM: compact fluorescents CFL T&D loss reduction 12% T&D loss target by 2001 (rather than 2004 T&D(+) in reference case) 10% T&D losses by 2005 T&D(++) 12% T&D loss target delayed to 2010 T&D(-) Max hydro Builds both reservoirs in the Upper Kotmale maxHy(D) project; all hydro project listed on Table 6-1 are built by 2008. Coal plants replaced by diesels. As above, but using steam-cycle oil plants maxHy(oil) Clean coal technology Pressurized fluidized bed combustion-combined PFBC cycle units; assumed for all coal units after 2000: With pessimistic capital cost assumptions PFBC(-) (US$350/kW increase in capital cost, vs. US$175/kW in the less pessimistic case). Flue gas desulfurization systems forced onto **FGD reference case coal plants Coal plants must be fitted wth FGD; model free FGD to chose least cost expansion path No coal Model free to choose least-cost combination of OS oil-steam and hydro. Oil plants assumed to use high-sulfur oil. As no coal, but using low-sulfur oil, low price LSOS(-) differential to high-sulfur oil (US$0.5/bbl) As no coal, but using low-sulfur fuel oil, high LSOS(+) price differential to high-sulfur oil (US$3.50/bbl) As no coal, but with high world oil price OS(highWoP) (US$30/bbl by 2004) Transportation sector Vehicle inspection and maintenance programs V/M Pricing policy options Historical RoE=3% Financial R LRMC/AIC AIC 152 VOLUmEHI Figure 6-1: The Model DEMAND MACROECONOMIC ASSUMPTIONS MODEL S PRI CES TAX AND PRICE POLICY FINAN ASUPTIMS-C DENANDS . _ ~~~~~~~~ENVIRONMENTAL TECNOLOGY AK) SITN EHNCt IOE ANALYSIS ASSUMP TIONS Ih_etert qabw FUEL USE MITI&AMN71 ] 0 i IONS OPTIONS13TEANLYI (NPV, I&Mzed akwh) l households~~~~~~~~~~~0 power rs Source: Meier and Munasinghe (1994), p. 162. Figure 6-2: Expected Trends for CE 2 Emissions in Sri Lanka POWR 3CT-t kS%FrOA 2 1992 1994 1996 1998 200l0 2002 2004 2006 200)8 2010 aindustry Eroad transport households power Source: Meier and Munasinghe (1994), p. 162. Assumptions Ceylon Electricity Board (CEB) base case load forecast for electricity (about 8 percent growth per annum (p.a.)) and the Official Generation Plan (that has a first 150 megawatts coal-fired unit on line in 2001). Household, industry, and road transport emissions are based on a Petroleum Product forecast of the Ceylon Petroleum Corporation. The household sector includes only liquified petroleum gas and kerosene consumption; net contribution to GHGs from fuelwood combustion assumed to be zero. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 153 Figure 6-3: T&D Loss Predictions (technical plus nontechnical, as a percentage of generation) 30_ ACTUAL LECO 25 20 20 \ 1~~~~~~~~~~~~993 PROJECTION 1 S S \ 4 - < ~~~~~~~~(CEB Is ACTU]AL CEB (CEO) CEO TARGET T&D(-) 10 T&D( LECO TARGET 1980 1982 1954 1968 1988 1990 1962 1994 1996 1996 2000 2002 2004 2006 2006 2010 2012 Figure 6-4: The Real Price of Electricity 3 2.5- _2. inalia 21. L) D indus 15 ! Os \0.5_ 0 I.. I . I 1972 1974 1976 1978 1984 1982 1984 1986 1988 1990 Sources: Munasinghe (1990) and CEB (1992a). 154 VOLUMEII Figure 6-5: Balance Sheet Ratios 2.5 DSC 2 DCSR covenant 1.5 C CRcovenant l _g 0~~~~~~~ g RoEc~~~~~~~~ovenant x 10 x 0 0.5 , . , , , . I 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 Figure 6-6: System Cost versus GHG Emissions 3900 QUADRANT IV QUADRANT I RoE=3% 3700- F- .~~~~~~~~~ld *FGD U, *f~~~~~~~~~~d w~~~~ind u 3500 0 maxHy(D) OS(high wop w o * maxHy(OaxHy(D uz 3300- = mifmaxHy(OS) T&D(-) u LSOS(-) a OS T&D(-) X / X ~~~~~~~~~CFL -, 3100 _i/ 2900 - g~~~~~AIC QUADRANT 111 QUADRANT II 2700 40 50 60 70 80 90 100 CO, EMISSIONS, 1994-2014 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 155 Figure 6-7: The Optimum Portfoliofor GHG Reduction 3600 3400 - ; 3200 -&D(++) O VJ~~~~~~~~~~~~~~~T&D(++-F)CFL Q 3000 - +EER v) 2800 VIM i +wind/ 2600_ / 2600 +MaxHy(OS) 2400 +mrnibyK 2200 +LSOS(.j +AIC 2000 140 150 160 110 180 190 200 210 220 CO, EMISSIONS, 1994-2014 Source: Meier, Munasinghe, and Siyambalapitiya (1993). Figure 6-8: The Cost Curvefor GHG Emission Reductions 100 _nd 50 MaxHy(OS) Minih - 0 ~~~~~~-~~~ T&VD(++'.) u AIC EER -50 -100 FL -150 0 10 20 30 40 50 60 million tons CO, abated Source: Meier, Munasinghe, and Siyambalapitiya (1993). 156 VOLUME II Annex 6-A: Attributes List of attributes in the Sri Lanka energy sector model Attribute Units Impact/objective Nonenvironmental System cost present value of system economic efficiency costs Tariff levelized average rate, impact on consumers Rs per kWh Environmental Emissions of carbon dioxide [1000 tons] global warming. Population exposure person-gg/m3/yr human health impacts to air pollutants (1) Biodiversity index (2) []: (2) diminution of biodiversity; im- pact on habitat of endemic spe- cies. Increase in surface [ha] ecosystem impacts from thermal temperature > 1°C plumes. Employment discounted incremental employ- ment Emissions of acid rain precursors [1,000 t/y] potential for acid rain damages (SO2 and NO,) (1) The health impact index that reflected the population exposure to these pollutants. This was defined as Y = Ci Pi where X is the cumulative population exposure to the incremental ambient concentration attributable to the power plant. Pj is the population in the j-th grid square. Cj is the incremental average annual concentration in the j-th square, attributable to the emissions from the source in question. The Cj were calculated using a simple Gaussian plume model run for some given number of wind direction and wind speed combinations. The total concentration in squarej is then simply the sum of the individual plume contributions: Cj = Cjk where Cjk is the incremental contribution in square j attributable to the k-th plume. (2) The biodiversity index was developed to quantify the relative biodiversity value of habitat lost. The index value was based on the relative probability of finding an endemic species in each of Sri 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 157 Lanka's major terrestrial habitat types. Thus 1 hectare of lowland wet evergreen forest was assigned a value of 1; a comparable hectare of monocultures (rice paddies, tea plantations) was assigned a value of 0.001, with other habitats assigned intermediate values. While such scales do involve subjective judgment, the scale was developed by the eminent ecologist K. Arudpragasam, professor of ecology at the University of Colombo and former chairman of the Central Environment Authority of Sri Lanka. For further details, see Meier and Munasinghe 1994, Chapter 8. 158 VOLUMEII Annex 6-B: The Concept of Dominance and Tradeoff Curves On Figure 6B-1, we illustrate the concept Therefore, GHG emissions decrease relative of dominance and the tradeoff curve.37 Con- to the base case. Note also that pressurized sider the option "MaxHy(D)," in which are fluidized bed combustion (PFBC), under both built all hydro plants indicated on Table 6-1 the optimistic "PFBC(+)" and pessimistic (in the main text), and in which coal plants are "PFBC(-)" capital cost assumptions, dominate replaced by diesels. This is better (lower) in the case in which FGD is forced on conven- both GHG emissions and cost than all of the tional coal plants ("*FGD"). options in the hatched quadrant-namely, This analysis shows very clearly the desir- wind, FGD, *FGD, and RoE=3%. These ability of making decisions about whether or options are said to be "dominated" by not FGD systems are to be required on a MaxHy(D). The tradeoff curve is then defined systemwide basis, at the planning stage, rather as ihe set of nondominated options (or than on a case-by-case basis at the project "noninferior" set). level, at the stage of the project environmental Note that among a set of mutually exclu- assessment. The usual analysis of FGD sys- sive options, only one member of such a set tems that is often provided in project-level can be on each tradeoff curve. In the case of environmental impact statements, in which the the oil-steam options, the curve as drawn only options that are considered are with and assumes that the relevant scenario is under without an FGD system, is thus very likely to present oil prices, and a low differential for lead to poor decisions. low-sulfur oil (LSOS(-)); under high world oil On Figure 6B-2, we examine the tradeoff prices, the curve would shift to the north, and curve for a local environmental impact. Rather include EER, a lower T&D loss rate target, than just use local air emissions as the attrib- and none of the oil-steam cycle options would ute, we define a health impact index, in which lie on the tradeoff curve. In other words, each a Gaussian dispersion air quality model was scenario (of exogenous assumptions) will applied to prospective power plant locations, have its own tradeoff curve. and the resulting increments in ambient air We distinguish here between two flue gas quality for fine particulates and NO. was desulfurization FGD options: the one indi- overlaid with population distribution. Al- cated by "**FGD" forces FGD onto the refer- though we do not go one step further and ence case coal plants, while in the solution apply mortality or morbidity functions, the use indicated as "FGD," the model is free to build of population weighted ambient air quality whatever it likes-but if coal plants are se- changes is distinctly preferable to the use of lected, they must be fitted with FGD. Inspec- just emissions as a measure of actual impact.38 tion of the solution shows that if FGD systems This figure illustrates the importance of are required, and the model is free to build the siting assumptions. In the case "MaxHy(D)," least cost sequence (i.e., case "FGD"), then coal plants are assumed replaced by diesels, the first coal plant is delayed from 2004 in the reference case to 2007, and the Broadlands and Uma Oya hydro plants are built in 2004 3xFor example, the set of qualitative "air impact" (plants that are not built in the reference case). ratings applied in a siting study for Sri Lanka performed in the mid- 1980s suggested that the difference between the best-ranked and the worst-ranked score for this 37For a full discussion of the concepts of attribute was about 1:1.6. Yet the quantitative dominance, noninferior sets, and other ideas of calculations using changes in this health index showed multiattribute decision analysis as applied to energy- variations greater than one order of magnitude for the environmental analysis, see, e.g., Meier and same set of sites. See Annex 6-A for further details on Munasinghe (1994), Chapter 6. the health impact index calculation. 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 159 sited mainly in urban areas, whereas in the "lose-lose" quadrant for health effects. "MaxHy(OS)," the coal plants are assumed However, when the diesels are moved to replaced by steam-cycle oil plants at similar locations identical to the coal plants, their remote locations (namely Trincomalee in the health impacts also fall, indicated by the arrow northeast). Thus the MaxHy(D) case falls in on the figure. 160 VOLUMEII Figure 6B-1: Definition of Dominance and the Trade-Off Curve 3900 _ RoE=3% 3700 ' _ * ~~~~~~~~fgd . *FGD 0 3500 _ maxHy(D) F- * maxHy(OS) w ~ ~ ~ ~ ~ ~ ~ ~~~~Lch T&D(-) Oi 3300 t - TRADE-OFF CURVE CFL ; 3100 2900 - 0 AIC 2700 , I I 1 40 50 60 70 80 90 100 CO, EMISSIONS, 1994-2014 Figure 6B-2: Local Health Impacts 3900 * RoE=3% 3700 fgd D 8 3500 - * OS(high wop) a pf6c(-) - - maxHy(D) o L~~~axfly(OS) .Q 330 pfbc(+) ^ T&D(-) .5 3300 - - -- + - mniHY O * OS ~ - _ ,<&D(+) o _ CFL ;3100_ 2900 0 AIC 2700 l l l l 80 90 100 110 120 130 140 150 160 170 HEALTH IAPACT INDEX 6. Energy Sector Policy and the Environment: A Case Study of Sri Lanka 161 Bibliography Alfsen, K. H., A. Brendemoen, and S. . 1992b. "Power and Demand Forecast Glomsrod. 1992. "Benefits of Climate 1992-2012." Load Forecasting and Tariffs Policies: Some Tentative Calculations." Section, Colombo, Sri Lanka. Discussion Paper, Central Bureau of Sta- . 1992c. "Wind Energy Resources tistics, Oslo. Assessment: Southern Lowlands of Sri Anderson, D., and R. H. Williams. 1993. "The Lanka." Colombo, Sri Lanka. Cost-Effectiveness of GEF Projects." Coppel, J. 1991. An Analysis of Energy Policy Global Environment Facility, Working Measures and Their Impact on CO2 Emis- Paper No. 6, Washington, D.C. sions. Proceedings, ESCAP Conference on Banuri, T., et al. 1995. "Equity and Social Greenhouse Research, Bangkok, pp. Considerations. Intergovernmental Panel 263-86. on Climate Change." Working Group 11, Crousillat, E. 1989. Incorporating Risk and Chapter 3. Uncertainty in Power System Planning. Baron, W., and P. Hills. 1990. "Greenhouse Energy series paper no. 17, Industry and Gas Emissions; Issues and Possible Re- Energy Department, World Bank, Wash- sponses in Six Large Lower Income Asian ington, D.C. Nations." Working Paper No. 45, Center Electric Power Research Institute (EPRI). for Urban Planning and Environmental 1993. Proceedings of the SO2 Control Management, University of Hong Kong. Symposium (Section SA). Boston, Mass., Bernstein, M. 1992). "Costs and Greenhouse August 24-27. Gas emissions of Energy Supply and Energy Modeling Forum (EMF). 1993. Re- Use." Environment Department, Working ducing Global Carbon Emissions-Costs Paper 1993-40, World Bank, Washington, and Policy Options. EMF- 12, Stanford D.C. University, Stanford. Blitzer, C., R. Eckhaus, S. Lahiri, and A. Florida Solar Energy Center. 1991. Residen- Meeraus. 1993. "Growth and Welfare tial Demand Side Management for Thai- Losses from Carbon Emissions Restric- land. Report to DEC. tions: A General Equilibrium Analysis of Fuglestvedt, J., et al. 1994. A Review of Egypt." The Energy Journal 14(1):57-81. Country Case Studies on Climate Change. Buhring, W., et al. 1991. ENPEP: An Inte- Global Environment Facility, Working grated Approach for Modeling National Paper No. 7, Washington, D.C. Energy Systems. Argonne National Labo- Gesellschaft fur Technische Zusammenarbeit ratory, Environmental Assessment and (German Technical Assistance Agency, Information Sciences Division. GTZ). 1989. Electricity Masterplan. Co- Burgess, J. 1990. "The contribution of effi- lombo, Sri Lanka. cient energy pricing to reducing carbon Global Environment Facility. 1992. "Eco- dioxide emissions." Energy Policy (June): nomic Costs of Carbon Dioxide Reduction 449-55. Strategies." Working Paper series Number Busch, B., P. DuPont, and S. Chirarattananon. IH, Washington, D.C. 1993. "Energy Efficient Lighting in Thai Goto, N., and T. Sawa. 1993. "An Analysis of Commercial Buildings." Energy the Macro-Economic Costs of Various 18(2): 197-210. CO2 Emission Control Policies in Japan." Ceylon Electricity Board (CEB). 1992a. Energy Journal 14(1):83-1 10. "Long Run Marginal Costing, 1992." Hirst, E. 1992. "Price and Cost Impacts of Load Forecasting and Tariffs Section, Utility DSM Programs." Energy Journal Colombo, Sri Lanka. 13(4):75-90. 162 VOLUME II Hobbs, B. F. 1991. "The 'Most Value' Test: for the USA." Energy Journal 11(2): Economic Evaluation of Electricity 51-74. Demand-Side Management Considering Markandya, A. 1990. "Environmental Costs Customer Value." Energy Journal 12(2): and Power System Planning." Utilities 67-91. Policy 1: 13-27. Hogan, W. 1990. "Comments on Manne and Markandya, A., and Pearce, D. 1988. Environ- Richels: CO2 Emission Limits: An Eco- mental Considerations and the Choice of nomic Cost Analysis for the USA." En- the Discount Rate in Developing Coun- ergy Journal 11(2):75-85. tries. Environment Department Working Hourcade, J. C. et al. 1995. "A Review of paper No. 3. World bank, Washington, Mitigation Cost Studies. Intergovernmen- D.C. tal Panel on Climate Change." Working Meier, P. 1990. "Power Sector Innovation in Group 11 Report, Chapter 9. Developing Countries: Implementing International Institute for Energy Conserva- Investment Planning under Capital and tion. 1993. "Thailand: Promotion of En- Environmental Constraints." Annual Re- ergy Efficiency: Pre-investment Ap- view of Energy 15:277-306. praisal." Report to the USEPA, Global Meier, P., (1995). ENVIROPLAN: A Multi- Climate Change Division, Bangkok, Thai- Attribute Decision-Analysis Model for land. Power Sector Planning. IDEA, Inc., Japan International Cooperation Agency. Washington, D.C. 1987. "Feasibility Study on Upper Meier, P., and M. Munasinghe. 1994. Incor- Kotmale Hydro-electric Power Develop- porating Environmental Concerns into ment Project." August. Power Sector Decision-Making: A Case Jones, B., and E. Wheeler (eds.). 1992. Green- Study of Sri Lanka. Environment Depart- house Research Initiatives in the ESCAP ment, World Bank, Washington D.C. Region. Proceedings of a Conference, Meier, P., M. Munasinghe, and T. Bangkok, August 1991, ESCAP, Bang- Siyambalapitiya. 1993. Energy Sector kok. Policy and the Environment: A Case Study Joskow, P. L., and D. Marron. 1992. "What of Sri Lanka. Environment Department, Does a Negawatt Really Cost? Evidence World Bank, Washington, D.C. from Utility Conservation Programs," Metropolitan Environmental Improvement Energy Journal 13(4):41-74. Programme. 1992. Clean Air 2000: An King, K. 1993a. "The Incremental Costs of Action Plan for Air Quality Management Global Environmental Benefits, Global in the Colombo Metropolitan Area. Co- Environment Facility, Working Paper No. lombo. 5, Washington, D.C. Miller, P., H. Geller, and A. deAlmeida. 1992. . 1993b. "Issues to be Addressed by "An Example of Energy Savings in LDCs: the Program for Measuring Incremental Improving Electrical Equipment in Paki- Costs for the Environment." Global Envi- stan." Energy 17(10):969-82. ronment Facility, Working Paper No. 8, Mills, E. 1991. "Evaluation of European Washington, D.C. Lighting Programs: Utilities Finance Larsen, B. 1993. World Fossil Fuel Subsidies Energy Efficiency." Energy Policy (April): and Global Carbon Emissions in a Model 266-78. with Inter-Fuel Substitution. World Bank, Ministry of Environment. 1991. National Washington, D.C. Environmental Action Plan. Colombo. Manne, A., and R. Richels. 1990. "CO2 Emis- sion Limits: An Economic Cost Analysis 6. Energy Sector Policy and the Environment. A Case Study of Sri Lanka 163 Ministry of Environment and Parliamentary Proost, S., and D. Van Regemorter. 1992. Affairs. 1990. National Environmental "Economic Effects of a Carbon Tax: With Action Plan. Colombo, Sri Lanka. a General Equilibrium Illustration for Mintzer, I., and D. Von Hippel. 1993. Green- Belgium." Energy Policy 136-49. house Gas Assessment Methodology. Schipper, L., and D. Hawk. 1991. "More Report to the Global Environment Facil- Efficient Household Electricity Use: An ity, World Bank, Washington, D.C. International Perspective." Energy Policy Moreira, J. R., et al. 1992. "UNEP Green- (April):244-55. house Gas Abatement Costing Studies." Tata Energy Research Institute (TERI). t992. Brazil Country Study. UNEP Greenhouse Gas Abatement Cost- Munasinghe, M. 1990. Electric Power Eco- ing Studies: India Country Study, Phase 1. nomics. London, U.K.: Butterworths. New Delhi, India. Munasinghe, M., and G. Schramm. 1983. Tavoulareas, E. S., and J.-P. Charpentier. Energy Economics, Demand Management, 1995. "Clean Coal Technologies for De- and Conservation Policy. New York: Van veloping Countries." World Bank Techni- Nostrand Reinhold. cal Paper No. 286, Energy Series, World Nordhaus, W. D. 1991. AThe Cost of Slowing Bank, Washington, D.C. Climate Change: A Survey." Energy Jour- United Nations Environment Programme nal 12(1):37-65. (UNEP). 1992. UNEP Greenhouse Gas Ogawa, Y, and H. Awata. 1991. Simulation Abatement Costing Studies, Phase I Re- Study of Greenhouse Gas Emissions due port, Riso National Laboratory, Denmark. to Future Energy Consumption. Proceed- U.S. Agency for International Development ings, ESCAP Conference of Greenhouse (USAID). 1994. "Demand-Side Manage- Research, Bangkok, August 1991, p. ment Action Plan for the Philippines." 234-62. Office of Energy, Environment and Tech- OKO Institute. 1989. Umweltanalyse von nology, Washington, D.C. Energie Systemen: Gesamt Emissions Van Dang, G. 1993. "UNEP Greenhouse Gas Modell Integrierter Systeme (GEMIS). Abatement Costing Studies: The Case of Ministry of Economy and Technology, Thailand." Asian Institute of Technology, Hessen, Darmstadt, Germany. Bangkok, Thailand. Organisation for Economic Co-operation and Warford, J., A. Schwab, W. Cruz, and S. Development (OECD). 1993. The Costs of Hansen. 1992. The Evolution of Environ- Cutting Carbon Emissions: Results from mental Concerns in Adjustment Lending: Global Models. Paris. A Review. Washington, D.C.: World Pachauri, R. K. 1991. Global Warming: Im- Bank, November. pacts and Implications for South Asia. Wijeratne, D. C. 1993. Potential for Efficient Tata Energy Research Institute, New Lighting in Sri Lanka: An Economic Delhi. Analysis. Transactions of the Institution of Pearce, D. 1990. The Role of Carbon Taxes in Engineers, Sri Lanka, Colombo. Adjusting to Global Warming. Economic Wijeratne, D. C. 1994a. Domestic refrigera- Journal. tors: Energy Consumption and the Poten- Pearce, D. W., et al. 1995. The Social Costs of tial for Conservation. Paper presented at Climate Change: Greenhouse Damage and the Annual Meeting, Sri Lanka Energy the Benefits of Control. Intergovernmental Managers Association, Colombo. Panel on Climate Change, Working Group Wijeratne, D. C. 1994b. Sri Lanka: Renew- Im Report, Chapter 6. able Projects and Programs. Ceylon Elec- tricity Board, Colombo. 164 VOLUMEII World Bank. 1982. "Sri Lanka, Issues and Energy Sector Management Program, Options in the Energy Sector." Report Report 007/83, Washington, D.C. 3794-CE. World Bank, Washington, D.C, Yamagi, Y et al. 1993. "A Study on Eco- . 1983. Sri Lanka Power System Loss nomic Measures for CO2 Reduction in Reduction Study. UNDP-World Bank Japan." Energy Policy 21(2):123-3 1. 7 Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment Zeinab Partow and Stephen D. Mink TUNISIA' S CONCERN WITH INCREASING the country's self- sufficiency in livestock products and with the affordability of these products for its citizens has resulted in a web of pricing and subsidy interventions in the livestock sector. The environ- mental consequences of these measures, namely the degradation of Tunisia's rangelands, however, have rarely been a central consideration. 165 166 VOLUMEII While a variety of subsidies has promoted and a greater role for the private sector in the intensification of livestock production in accomplishing this. They did not want to stop parts of Tunisia, in other regions government at short-term, macroeconomic stabilization, subsidies and policies have encouraged the but rather aimed to get rid of the old dirigiste maintenance of the national herd at levels policies, to orient the economy outward, and beyond the carrying capacity of the country's to increase the role of market mechanisms. rangelands. Particularly during dry years, These reformers' objectives took a major step subsidized feed imports have provided a forward in 1987 when the regime of President substitute for reduced grazing supplies, and Bourguiba was removed in a managed transi- have succeeded in averting the large declines tion to a new government, under the leader- in animal numbers often associated with ship of President Ben Ali. droughts. This failure of livestock numbers to Despite the best intentions of the new respond to diminished feed availability in government, adjustment policies were imple- natural pastures, however, has contributed to mented with halting progress over the next significant environmental degradation of the five years. Severe droughts in 1988 and 1989 Tunisian range. caused delays in initiating reforms that had The continuing deterioration of Tunisia's high social costs initially. Just emerging from rangelands has direct effects on livestock these economic blows, Tunisia was then hit by production, and longer-term, indirect implica- the consequences of the Gulf War during tions for the entire agriculture sector. While 1990-91, most immediately by a sharp drop in important efforts at pasture improvement and tourism, one of the principal foreign exchange reforestation are under way, the positive earners of the economy. impacts of these and other measures are often Nonetheless, important elements of struc- undermined by subsidy and pricing policies tural adjustment have been accomplished, and that fail to consider or respond to environmen- there have been improvements in macroeco- tal signals. nomic indicators. Reform efforts have focused This chapter investigates the effect of on (a) privatization and restructuring of state- subsidies and pricing policies on the livestock owned enterprises; (b) liberalization of do- sector, evaluates the impact of these policies mestic price and imports; (c) rationalization of on the condition of Tunisia's rangelands, and the structure of import tariffs, followed by a considers the likely consequences of recent reduction in quantitative import restrictions; changes in these policies as part of the eco- (d) removal of most credit and interest rate nomic adjustment program that began in 1986. controls; and (e) tax reforms, including intro- duction of a value added tax in 1988 and The Context of Adjustment Policies revised direct tax in 1990. The macroeconomic context The agricultural context In the early 1980s, Tunisia began to face The agricultural sector is important in the growing macroeconomic imbalances. By overall economy, and has undergone substan- 1985, with oil reserves and revenues declin- tial changes since 1987 because of adjustment ing, slow growth of other exports, and wors- policy reforms. Tunisian agriculture remains ening indicators of the external debt burden, it predominantly rainfed and prone to production was becoming clear that current account risks linked to climate. Only 5 percent of deficits could no longer be financed. cropped land is irrigated, although this land Stabilization policies were initiated in accounts for a much higher share of sector 1986. Reformers in government recognized value added. The sector, excluding agro- the need for greater efficiency in investment industry, employs nearly one out of every 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 167 three workers in the labor force, generates marketing of cereals, exports of olive oil, and 13-15 percent of gross domestic product imports of beef and powdered milk were (GDP) and contributes about 11 percent to relaxed. Despite considerable progress, the export earnings, although these indicators are agricultural sector still faces significant con- undergoing structural decline. straints. Several objectives have motivated the These constraints are evident from the adjustment process in the agricultural sector. recent performance of the sector. Growth First, adjustment has been concerned with the averaged 4.2 percent annually in real terms in agricultural trade gap, particularly for cereals, the first half of the 1980s, then slowed to only meat, milk, and sugar, and is trying to increase 1.5 percent in the latter half because of three the country's production of these staples. drought years, before recovering with a record Second, it is oriented toward improving rural cereals crop in 1991. Domestic production has incomes, as a means of reducing migration to not kept pace with rapidly growing demand, towns and cities, where unemployment is a resulting from population and income growth, problem at 15 percent. Buffering domestic and reinforced by high consumer price subsi- production and incomes from the short-run dies on many staple foods. While exports of consequences of drought is of particular con- agricultural commodities have rapidly ex- cern, and various responses have been tried, panded (12.6 percent average for 1987-90), including subsidies for irrigation infrastructure they have been outstripped by increasing and water. Finally, maintaining affordable imports (18.8 percent). prices for a variety of food staples for consum- Livestock has traditionally played an im- ers has been an objective, and has been pur- portant role within the agricultural sector, and sued through food subsidies financed from the continues to account for almost 30 percent of Caisse de Compensation General (CGC). sectoral value added. Decades ago, nomads Pursuing these objectives, while undertak- would move their herds over long distances ing fundamental structural adjustments in the from Central and Southern Tunisia to the agricultural sector, has required a delicate North in a transhumant pattern that took balancing act. Elements of the reform policy advantage of seasonal pasture in the South and include (a) producer price realignment (mostly post-harvest crop stubble in the North. But upward), followed by adjustments to interna- intensification of agriculture has broken this tional prices; (b) reduction in credit subsidies; cycle, and herds are less mobile today and (c) reforms of parastatals involved in cereals, more dependent on locally available feed olive oils, fisheries, and livestock; and (d) resources. Sheep and goats continue to be an reduction of subsidies for inputs, including important part of the national herd, particu- fertilizer, animal feed, and irrigation water. larly where conditions are more difficult, The latter reforms were meant to encourage while intensively run dairy and poultry opera- private participation in marketing fertilizer tions have gained in importance over the past and feed inputs, and to reduce the burden of twenty years. their subsidization on the government budget through the CGC (see Table 7-1). Environmental Trends Significant liberalization of the sector has already occurred, with additional reforms Livestock raising and its related input and being implemented with strong government marketing activities have a number of poten- commitment. Cereal producer prices have tial impacts on the environment. This chapter been raised and linked to world prices, input focuses primarily on one of these-range and subsidies have been lowered, and restrictions grazing land degradation-rather than on on private sector involvement in internal potential problems such as pollution from intensive production systems and from slaugh- 168 VOLUME II terhouses. Range and grazing land degradation 100-300 millimeters per year, respectively. is influenced by a broader array of the reforms Average vegetation falls dramatically, and under reconsideration by policymakers; it also there is evidence of heavy overgrazing of poses a larger environmental problem because rangelands. Vegetative cover is estimated to of its links to erosion and siltation of hydro- average 15 percent but drops to less than 8 logical infrastructure. percent over large areas. This zone becomes increasingly arid as one moves to the south Bioclimatic zones and east and merges at about the 150 millime- Nearly twenty peceters isohyet into the Sahara, a region charac- landarea oft6.2mnillionhectares isirangeland, terized by very low rainfall (less than 100 much of it severely overgrazed. Less than a millimeters per year) and extremes of temper- third of the country's area is cultivable. An ature and wind (Kennedy 1989, p. 120). additional 6 percent of the land mass is classi- Land degradation fied as forests and woodland, primarily in the North and Northwest, and is commonly used Only piecemeal data exist on the severity for livestock grazing. The remaining land area and evolution of land degradation in Tunisia.' is desert or other noncultivable land. Nevertheless, various studies on the condition Tunisia's landmass can also be described of Tunisia's rangelands, in addition to existing according to five geographic regions defined environmental data, indicate that Tunisian by climatic characteristics. These are the rangelands suffer from significant environ- humid, subhumid, semiarid, arid, and desert mental deterioration. Unfortunately, this zones. Alternatively, the first three of these are information is inadequate to determine the often grouped as the North; the arid zone immediate environmental impacts of recent coincides roughly with the Center; and the policy changes, but speculative conclusions desert zone with the South (see Figure 7-1). will be suggested below. Future analysis will The humid zone in the northwest tip of benefit by a rangeland inventory and analytic Tunisia enjoys an average annual rainfall often capacity currently being developed by the in excess of 1,000 millimeters per year. This Ministry of Agriculture. area supports intensive cereal cultivation and Estimates made in 1980 indicated that diminishing areas of natural cork oak wood- approximately 1.2 million hectares were land. The continuous felling of the cork oak seriously affected by erosion in North and and Aleppo pine woodland, now covering less Central Tunisia-representing 25 percent of than 250,000 hectares, parallels an increase in land resources in those regions.2 Another degraded area of maquis or underbrush. Bor- study estimates 23,000 hectares of productive dering the humid zone to the east is the subhu- land are removed from agricultural use each mid zone, with average rainfall between 600 and 800 millimeters per year. It is a region dominated by cereal cultivation, with only 'The term land degradation is used here to refer relatively small pockets of rangeland remain- to a process of vegetation depletion which ultimately ing on steeply sloping land prone to erosion. results in soil erosion, loss of nutrients, and decreased Many unpalatable species of thistle and gorse yields. In parts of Tunisia where desertification is a are often present and indicate derelict or threat, degradation is also linked to wind erosion resulting from the reduction of the perennial plant cover overgrazed land. To the south are the large below a critical threshold (Le Houierou 1990, pp. semiarid and arid regions of the Central 100-104). steppes where the majority of Tunisia's range resources are located, and where rainfall is 2World Bank 1989b, p. 12. Unfortunately, the limited to 300-600 millimeters per year and study did not distinguish between arable and nonarable lands. 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 169 year due to a number of factors, primarily tional grazing (World Bank 1989a, p. 40). water and wind erosion, which affect 10,000 In Northern Tunisia, land degradation is and 8,000 hectares, respectively.3 largely the consequence of water erosion. Most natural rangelands are in a state of Gully erosion is common in the Tell, where continuous and advanced degradation as a the rate of soil loss has been measured at result of overexploitation and poor manage- 1,020 tons per square kilometer per year ment, often encouraged by inappropriate (United States National Committee for Man policies, and exacerbated by periodic and the Biosphere 1981, p. 23). Both physical droughts. Tunisia's rangelands are largely and human factors account for the high levels located in the arid and desert bioclimatic of erosion. On the one hand, the land is steep- zones corresponding with the Center and est where the rainfall is highest; on the other, South (Table 7-2). Here, wind erosion is a forest clearing, overgrazing, fuelwood collec- major problem. While reliable data on losses tion, and cultivation of even the steepest in soil fertility are difficult to find, it is esti- slopes expose the soil to wind and rain. The mated that 5.5 of the 10.6 million hectares limited rainfall often arrives in heavy showers south of Sfax are moderately or severely between the months of October and April-50 affected by desertification. Permanent loss is percent of the rainfall occurs between the estimated to average about 8,000 hectares per months of December and February-resulting year (World Bank 1989a, p. 40; Hamza and in significant water erosion. Bennour 1990, p. 79). Soil removal at a rate of over 1 millimeter per month has been mea- Carrying capacity sured on cultivated sandy steppes of Southern Rangelands in Central and Southern Tuni- Tunisia (Le Houerou 1986, p. 112). sia have thus been subjected to multiple pres- Tlhe direct causes of ran geland degradation The direct causes ofrangelasures that have tended to reduce their carrying in semiarid and arid zones of Tunisia are tree capacity. Land conversion has shifted what are and shrub cutting for fuelwood, the overgraz- ing of rangelands, and the extension of rainfed e st ferti ranglands, iedhig cultivation. ont trdtoa.agead.Tai est rainfall zones, into crop lands, reducing cultivationontotraditional grazingelandsninste Sboth the quality and the quantity of the grazing tional grazing lands In the Center and South 'al oee-nraignm resources available to ever-increasing num- are also being lost to mechanized cultivation bers l of cereals (particularly barley) during the wet bers of livestock. The extension of farnibng years as ell a to ree cops olive and into nmarginal lands has in turn contributed to yarns, awlly to tee crpu oie andu soil erosion as permanent cover is lost and as almonds), only to be left denuded and unpro- thusofmcieyowrkhelni- tected from erosive winds during drier years. creases. Carring py ha been the r Over he pat twodecaes, a estiated creases. Carrying capacity has been further Over~ ~~~~~~' th pattodcds,a.siae reduced by the gradual transformation of 500,000 hectares of land in semiarid and hilly tae to taryaliock areashavebeenbrouht ito poducion, transhumant systems to sedentary livestock mareas have berenls bWoroug ianto198 product raising over the past decades, in effect reduc- mainly for cereals (World Bank 1982, p. 27). Land policy has also encouraged the conver- ing the range's restorative capacity. sion of grazing land to crops through a policy Carrying capacity estimates are notoriously to extend landownrshirihtstcontroversial, and difficult to specify for to extend land ownership rights to those who Tuii. Idal,cryn.cpct siae invest in the land-i.e. clear and cultivate . ' it-rather than to those who use it for tradi- for different livestock management systems in the Tunisian context could be compared over time. Unfortunately, such estimates do not exist. Some analysts have extended rules of 3World Bank 1989a, pp. 38-39. An additional thumb applied in North American range man- 1,000 hectares is lost to flooding, and 4,000 hectares of agement to the Tunisian rangelands, suggest- agricultural land is diverted to urban use each year. 170 VOLUME 11 ing a stocking rate of 1 livestock unit4 per inch concentrates, forage crops, and agricultural of rainfall per square mile per year. Converted by-products. The use of both irrigation and into metric units and assurning average annual fertilizers on forage crops has increased, and precipitation in the extensive pasture areas of a more intensive agriculture, supplying more Southern Tunisia to be 100 millimeters per by-products, is being practiced, aided by year, this rule would allocate four livestock growing mechanization. Thus, two competing units to 2.63 square kilometers per year. trends characterize the livestock sector: (a) the Stocking rates in Tunisia were estimated in reduction of the carrying capacity of the 1981 to be three to eight times greater than rangelands of Central and Southern Tunisia on this ideal rate (United States National Com- the one hand, and (b) the intensification of mittee for Man and the Biosphere 1981, p. livestock production in the North, based on 20). There are, of course, serious problems cultivated crops and concentrates on the other. with extending this standard to Tunisia, partic- ularly given the low levels of milk production The Livestock-Feed Balance and liveweight characteristic of Tunisian livestock. Nevertheless the orders of magni- In this section, we turn to a more detailed tude indicate that stocking rates did exceed discussion of the livestock sector, and to the carrying capacities over a decade ago. Given role of government policies within it. The the growth of the national herd over the past balance between feed availability and live- ten years, the situation is likely to have wors- stock numbers is a central factor affecting ened. environmental conditions in Tunisia's range- As land degradation progresses, the reduc- lands. This section will outline the trends in tion in the consumable production of range- the number of animals making up the national lands is quite dramatic. Rangelands in fair herd, followed by a discussion of feed re- condition, with a plant cover of 25 percent, are sources. capable of producing 820 kilograms of con- sumable plant material annually; overgrazed Growth of the national herd rangelands only produce about 490 kilograms; Eighty years ago the national herd consisted and lands where wind erosion has blown away of the equivalent of 360,600 livestock units, most of the top soil, with plant cover only nearly half of which were cattle, a third were about 4 percent, produce under 300 kilograms. sheep, and under a fifth were goats. By 1991 (Le Houerou 1986, p. 123). Such a process has the national herd had grown nearly sixfold to resulted in more livestock grazing on poorer over 2 million livestock units: sheep increased resources-both in terms of quality and to 59 percent of the total livestock units; cattle quantity-in Central and Southern Tunisia, affecting not only the regenerative capacity of and gasdecrease 2n d the range, but also the productivity of live- percent, respectivelyf stock. Growth of the national herd has remained stock. steady over the past twenty years, albeit fluc- Alongside the reduction in carrying capac- tuating from year to year with changes in ity of rangelands has been a trend of increas- rainfall and feed availability (see Figure 7-2). ingly intensive livestock production in the North of the country. This intensification has been largely based on the consumption of feed 51n terms of heads of livestock, the 1911 national herd composition was 171,000 cattle, 123,200 sheep and 66,400 goats, while the corresponding 1991 figures were 631,000, 1258,000 and 262,600. Camels are also 4A livestock unit is here defined in bovine units and important in the South, although their numbers have not is equal to one head of cattle or five adult sheep or fluctuated dramatically over the past twenty years, goats, with or without young. remaining steady at about 180,000 units. 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 171 These generalizations, however, mask the since throughout the 1960s and 1970s, num- striking differences between the growth trends bers had not shown any fluctuation with of sheep, goats, and cattle. rainfall. Another explanation is the liberaliza- The ovine (sheep) herd has exhibited a tion of mutton and goat meat prices in 1979 more cyclical growth pattern than the bovine and the subsequent decline in the profitability herd, due both to a more rapid turnover of of beef production, a topic which will be dealt animals and the herd's heavier dependence on with later in this chapter. natural pastures and thus on rainfall condi- tions. Particularly large declines in animal Livestock distribution by geographic zone numbers take place when two or more dry and holding size years occur in succession, such as what hap- Nearly one-half of the national herd (in pened in the late 1960s when numbers fell by livestock units), is located in the Central and nearly 25 percent, and in the late 1970s when Southern regions of Tunisia. Livestock pro- numbers fell by nearly a third. This at least duction in these regions is more heavily de- was true until the 1988-89 drought. During pendent on natural pastures, and therefore on this latest drought, the ovine herd suffered less rainfall conditions, than in the North. than 3 percent decline in total population, due Sheep are concentrated (46.9 percent) in the in some measure to the subsidized provision Center of the country, although there is a of barley feed to small herders in drought- significant number in the North (39 percent). stricken areas. stricken areas. Sheep numbers in both areas are increasing, in In the 1920s to 1950s, the caprine (goat) contrast to the South where only 14.1 percent herd was much larger than it is today. At what of the nearly 6.3 million sheep are located (see may have been a peak in 1944, the herd num- bered 2.25 million. This fell to less than Figure 7-3). 400 00 heas in 969 bt hassincebeenThe largest portion of the Tunisian goat 400'000 headsin199bthassinceherd is located in the and Southern region of steadily increasing to the current level of over the country (41.6 percent), although the num- 1.3 million. As is the case with the ovine herd, bers in that region have not increased as sig- goat numbers have in the past fluctuated with nificantly over the past ten years as they have the rains and the availability of pasture. Two in the North or the Center (see Figure 7-4). A or more years of consecutive dry weather have third of the caprine herd is located in the resulted in declines of 25-30 percent. In the North, and in this region numbers have in- most recent droughts of 1988-89, however, creased steadily over the past decade. The goat numbers actually increased by 30,000 Center accounts for 23 percent of the coun- head (2.5 percent), reflecting the impact of try's caprine herd, and here also, numbers subsidized barley provided as part of a have increased in recent years. drought-relief program. drought-relief program. ~~It iS noteworthy that sheep and goat num- Cattle numbers grew rapidly and continu- bt is teworth hat shepn oat nm- ously from the mid-1960s until 1978 due to berse Sot hve shown only sm , . . ....... . m~~~icreases and even declines over the past dairy promotion policies, but the herd size decade. The dependence of livestock on the then fell precipitously during the drought of natural range is greatest in the South, and the 1978-79. The cattle herd has not rebounded lack of increase in numbers suggests increas- from that tremendous decline in numbers and ing degradation in this region, as well as to has shown only modest growth over the growing intensification of livestock produc- 1980s. While the decline in numbers coin- tion in other parts of the country. cided with the droughts that struck the country The vast majority of the country's cattle in those two years, this is not a convincing herd is located in Northern Tunisia (81.5 explanation for the fall in cattle numbers, percent). The Center is second with 17.5 172 VOLUME II percent, and last is the arid South with only 1 tion with forest regeneration projects on state percent of the bovine herd. Total cattle num- land. The latter requires closing the areas to bers have been stable over the past decade and the traditional grazing use by neighboring show no variation with rainfall, a reflection of communities. the dairy herd's heavy dependence on concen- In the drier regions of the South, socially trate feeds and cultivated forage crops. oriented interventions have made available Cattle ownership is concentrated in the subsidized supplements of barley feed during smallest and largest farm size categories. drought. The policy goal was to protect the Nearly half of all cattle are held in farms incomes of small herders. As an indirect smaller than 10 hectares in size, while a quar- consequence of agricultural intensification, ter are held in farms of over 50 hectares. fallow is declining, but crop residues have Small farms typically have fewer options in become more available. Also, in some areas of providing feed to their animals from on-farm Central and Southern Tunisia, a policy is resources, and are more dependent on pastur- being pursued to privatize collective grazing ing their herd, partly on common property land. Some of this land have shifted to grain resources. Fewer sheep and goats are held in production, thus further reducing the area the smallest farms (25 percent and 35 percent available for pasturing herds. in farms under 10 hectares, respectively). The The role of the various feed resources sheep and goat herds are concentrated in the differs from zone to zone: (a) feed concen- middle and larger farms over 20 hectares in trates are relatively more important in the size. humid zone where much of the dairy cattle are located; (b) forage is also more important in Feed sources the humid and subhumid zones; (c) pasture Five broad categories of feed resources are provides a major part of livestock feed in the used in Tunisia: (a) pastures, both forested and arid and desert zones; and (d) hay, straw, nonforested; (b) agricultural by-products, such stubble, and other agricultural by-products as straw, stubble, hay, and silage; (c) fallow; appear to be important in all but the ard and (d) forages and forage shrubs; and (e) feed desert zones (FAO 1988, Vol. I, p. v). concentrates. With the obvious exception of Overall coverage offeed requirements feed concentrates, the composition of the various feed sources, their biomass, and pro- Two distinct trends characterize feed avail- ductivity vary enormously in response to ability in Tunisia. Feed resources destined for varying degrees of social and environmental intensive livestock production, forage and stress (Kennedy 1989, p. 120). The contribu- barley, appear to be increasing, with the tions to livestock feed requirements of the growth in barley offsetting declines in concen- different resources for 1986, a dry year, were trate feed production. However, for extensive 30 percent for agricultural by-products, 25 livestock production, feed resources in the percent for feed concentrates, 23 percent for Center and South have not been able to keep forage, 12 percent for pasture and 10 percent pace with the requirements of the growing of fallow lands (FAO 1988, Vol. I). animal population. The best pasture land is In broad terms, the strategy for improving increasingly being converted to marginal availability and quality of feed resources has cereal cultivation, with the remaining pastures followed several paths. In the North, intensifi- being degraded and overstocked, and weedy cation of fodder production is possible. This fallows previously used for grazing are being has involved greater use of fertilizer and reduced by agricultural intensification. irrigation. Also, permanent pasture is being These two divergent trends are illustrated in established in some forested areas in connec- Figure 7-5. Central and Southern Tunisia (the 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 173 arid and desert zones, respectively) cannot ruminants, has increased to 30 percent in normally meet the feed requirements of their normal years and to over 50 percent in drought livestock, with the shortfall being on the order years (World Bank 199 1c, p. 12). As an illus- of 25-30 percent. Feed requirements are tration of the growing importance of feed covered only in years with above-normal concentrates for the ovine herd, a study (Min- rainfall. The situation in the North is quite istry of Agriculture 1991, pp. 92-93) carried different, with only the humid zone falling out in 1990 in Menzel Habib (governorate of below its feed requirements, while the subhu- Gabes in the arid region) reports that farmers mid and semiarid zones manage to cover their owning small ruminants relied to a great requirements even in dry years. extent on hay and concentrates to feed their In the humid zone much of the shortfall can herds, which averaged between twenty and be met by the use of feed concentrates, partic- fifty animals. Barley and agricultural by- ularly in the case of cattle. The same is not products were also used, although pasture land true for sheep and goats in the desert and arid did not seem to play an important role in zones, resulting in overgrazing and reduced animal rations. livestock productivity (FAO 1988, Vol. 11, p. The production of concentrate feeds grew 97). dramatically between 1977 and 1981, from 100,000 tons to over 300,000 tons, but de- Feed concentrates clined dramatically back to the initial level in Over the past three decades cattle, espe- 1982, subsequent to removal of subsidies. Over th pas the decde cattl, espe Production in more recent years has remained cially dairy cattle, have relied quite heavily on stable (apart from growth of the poultry feed feed concentrates. Yet during the last ten years, small ruminant production has also sector). Of total production, concentrated feed become more dependent on concentrate feeds for sheep increased rapidly in the 1980s rela- tive to bovine use, whose production declined produced from pake.T imcrpoed brey,ndc maiz, by over 50 percent.6 This reflected the greater and soya cake. This increased dependence iS prftblt of she.n otpouto due to the expansion of cultivation in the profltabilvty of sheep and goat production South and Center which has reduced natural relative to beef following liberalizateon of range resources, and to crop intensification in sheep and goat markets. (Dairy and beef prices the North, which reduced the weedy fallow that had been an important source of grazing. In addition, as sheep have been increasingly Forage integrated in the farming systems in the North, Total area devoted to forage crops has seasonal migrations from south to north for increased by only 8 percent during 1977-91, fallow and stubble grazing have declined in but production has increased more due to importance. The reduction of natural sources increased use of irrigation and fertilizers. For of grazing has resulted in recourse to transport example, fertilizer use grew by about 40 of concentrates and hay from north to south. percent over the period. However, rainfed Consequently, sheep production has become production remains predominant. On rainfed more costly, but as a result of the liberaliza- land, cereals and annual legumes sown in tion of mutton and lamb prices in 1979, the mixtures are common, or alternatively, barley increased use of concentrates, hay, and straw to supplement range grazing is still quite profitable (World Bank 1991 a, p. 30). 'Bovine feed concentrate production grew from Over the last fifteen years the use of con- 48 to 168 tons between 1976 and 1980, but subse- centrated feed, previously limited to less than quently fell to 71 tons in 1986. Ovine feed production 10 percent of the total requirements of small in the same three years was 2, 25, and 89 tons (Chouchen and Alaya 1989, p. 130). 174 VOLUMEH! is grown for grazing early in its life cycle prior the barley harvest extremely dependent on to maturation to a grain harvest. The composi- rainfall conditions, with the magnitude of tion of forage output has shifted some from fluctuations in production between wet and perennial to annual species over this period. dry years increasing since 1984. The environ- Perennial forages such as sulla and lucerne mental implications of this expansion have declined in area from about 108,000 hectares been noted above.7 Barley price incentives do to 95,000 hectares, while annuals increased not seem to account for the large variability of from 167,000 hectares to 203,000 hectares, or year-to-year production, although this topic 18 percent. Production of annual forage crops will be addressed more fully below. increased by about 30 percent to 1.4 million tons over the period between 1985 and 1991, Fallows although the dependence of forage production Fallows in Tunisia are typically un- on rainfall was evident in 1988 when produc- managed, with the land simply used for graz- tion fell by nearly 50 percent due to severe ing of any natural growth. Total fallow area drought conditions. available for grazing has declined with the Only about 12-15 percent forage area is jump in farm mechanization through the mid- currently irrigated, but because of higher 1980s that resulted in an increase in the prac- yields, this area accounts for about 35 percent tice of clean cultivated fallow, and the conse- of total forage production. The area in irri- gftotate forage cropsctincrThea from 8, quent decline in stubble and fallow grazing. It gatedforgecropsincreased from 8.2is uncertain whether this reduction in feed hectares in 1973 to 26,000 hectares in 1989 availability from fallow has been offset by the (World Bank 1988b; Ministry of Agriculture increase in the area under cultivated forages. 1989, p. 16). Irrigated forage is largely fed Such may be case for cattle in the North, but green on a cut-and-carry system to stall-fed is less likely in the case of sheep and goats in dairy cattle, although silage is becoming more Central and Southern Tunisia for which forage common. The overall increase in forage pro- feed sources are rarely used. Thus agricultural duction, in fact, has probably benefited the boin hedrte'hnsepadgas n intensification and the consequent loss of boviin ress r ind stubble and fallow grazing land is likely to hias contributed to the significant increases in affect most seriously the small ruminants in milk production despite a stable cattle popula- the degraded pastoral areas ("Tunisia, Review tion. of Prospects for Agricultural Diversification," Barley p. 12). Barley in its various forms-green barley, Pastures grain, straw, and in feed concentrates-is Grazing land provides the majority of perhaps the single most important feed source livestock feed in the arid and desert zones. In for Tunisian cattle. Production of barley has addition to their increasing degradation, the become increasingly variable with rainfall, productivity of the rangelands is heavily Since the mid-1980s, during humid years influenced by the amount of rainfall they production almost doubled the previous output levels of good years. Yet production in recent dry years (1986, 1988, and 1989) differed 'Barley imports often partially fill the gaps little from past dry years. This performance resulting from reduced harvests during the dry years. reflects that much of the increase in produc- The contribution of these imports to the fulfillment of tion has resulted from an expansion into feed requirements is largely reflected in the figures for marginal zones previously used for livestock feed concentrate production, since it is likely that much pasture. This makes a significant portion of of the imported barley is used as input into these concentrates. 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 175 receive. The degree of variation in yields Livestock Policy and Natural between a humid year (1985) and a dry one Resource Use (1986) from pasture is on the order of 642 million forage units versus 293 million forage Since the 1970s, Tunisia has used occasion- units, respectively, a variation of more than ally inconsistent price controls and subsidies 220 percent (FAO 1988, Vol. II, p. 95). directed both at the final product, meat, as well as at inputs such as feed and fertilizers, to Summary achieve economic and social objectives. These Livestock numbers have been increasing in have included the subsidies for importing red Tunisia over the past twenty or thirty years. meat and milk products, production subsidies The increases, however, reflect expansion of to encourage continued investment; mainte- sheep and goat herds to a far greater extent nance of consumer prices at socially and than cattle stocks. In fact, cattle numbers are politically acceptable levels, and income significantly lower today than they were in the protection for livestock owners through subsi- mid- 1970s. This suggests that the small rumi- dies for animal losses during drought years. nants have greater significance in terms of This section analyzes the impact that govern- environmental pressures on degraded range- ment policies have had on the environment lands. through their impact on livestock numbers and The impact of sheep and goat numbers on on animal feed production. the environment is intensified by their heavy Meat prices and subsidies concentration in the Central and Southern regions of Tunisia, respectively, where most Most price and subsidy policies that di- of the country's rangelands are located. Cattle, rectly affect the livestock sector are managed on the other hand, are almost exclusively through the Caisse Generale de Compensation found in the humid and subhumid North, (CGC), including consumer price ceilings and where only a small portion of Tunisia's pas- subsidies on inputs for feed production (in- ture resources remains. Moreover, the bovine cluding fertilizer, feed concentrate ingredients, herd is largely dependent on feed concentrates and barley). and forage crops, and thus puts relatively little Beef was subject to production and con- pressure on rangelands. Although the depend- sumption ceiling prices from 1970 until 1982, ence of sheep on feed concentrates is on the when producer prices were liberalized. Con- rise, particularly in Central Tunisia, grazing sumer prices continued to be fixed for meat still provides the major portion of ovine and sold through butchers subcontracting with the caprine feed requirements, especially in the state-run Ellouhoum, which retained a monop- South. oly on beef imports until 1990. While a signif- It can therefore be concluded that environ- icant portion of beef is processed by private mental degradation caused by the excessive slaughterhouses and butchers, the importance growth of the livestock herd through overgraz- of Ellouhoum in beef marketing and distribu- ing is primarily due to increasing numbers of tion has meant that its policies have exerted small ruminants, particularly sheep, in Central significant downward pressure on beef prices regions of Tunisia, and only secondarily due in the market as a whole (AIRD 1989, pp. to cattle production systems. 119-20). Mutton and goat meat were also subject to price controls until 1979, when prices were liberalized at all stages (AIRD 1989, p. 126). Even under the price controls, producer prices for mutton increased in real terms (see Figure 176 VOLUME n1 7-6), and this trend has continued since price reports indicate that in 1988 some 50,000 liberalization at 1 percent annually in real Tunisian sheep were exported to Libya (AIRD terms (AIRD, pp. 131-32). Marketing of 1989, p. 119). However, it is difficult to draw sheep and goat meat is free of price or market any firm conclusions regarding the effect of controls and generally is handled by the pri- Libyan demand on the size of the Tunisian vate sector (Tunisia: Review of the Prospects herd. for Agricultural Diversification, pp. 88-89). The impact of imports on beef prices has Thus, sheep and goat producers faced better been perhaps more significant. To meet do- price incentives than beef producers over the mestic demand which is augmented by subsi- 1979-90 period. While mutton and goat meat dized meat prices, beef is bought on the inter- prices increased by nearly 30 percent between national market-largely from the European 1970 and 1986, beef prices actually declined Community-at artificially low prices and in real terms during that period (Figure 7-6). sold to butchers associated with Ellouhoum, at The combination of output price controls on a fixed price (2.1 dinars per kilogram in beef and the support measures provided by the 1987). By contrast, local meat is bought at a state in the form of input subsidies (to be higher price than imported beef (2.3 dinars per addressed below) has resulted in the stagna- kilogram in 1987) and sold at the same price tion of local production and the consequent as imported meat. The profits made by growth of beef imports. Allowing markets to Ellouhoum from the sale of imported meat are function with less intervention has contributed meant to cross-subsidize local meat produc- to the growth of the ovine and caprine herds tion, although the actual effect was the stagna- which experienced virtually uninterrupted tion of domestic beef production due to low growth in numbers since 1979. international prices that were passed through Tunisian trade policies have influenced to domestic prices (Ministry of Agriculture domestic price incentives for both beef and and FAO 1987, p. 16). mutton production. Tunisia is self-sufficient in Pricing policy for beef was reformed in sheep and goat meat; reported mutton and 1990 as part of the structural adjustment lamb imports and officially reported exports program. The reforms consisted of liberalizing are small. Tunisian mutton production appears retail prices for beef and allowing private to be somewhat protected from imports: while imports of refrigerated meat, while applying a international mutton prices fell by over 30 variable levy sufficient to provide domestic percent over 1980-86, Tunisian prices rose producers with price protection of 15 percent. over the same period. Such policies tend to With these reforms in place, price incentives exacerbate the degradation of range resources for domestic production have improved, and by raising financial returns to rangeland pro- there appears to be increased interest in invest- ducers and raising animal numbers on the ing in cattle fattening and dairy operations. range. These recent reforms in meat pricing have Significant cross-border exports to Libya to some extent been facilitated by changes in are thought to take place in some years, al- meat consumption patterns (see Table 7-4). though no estimates exist of this illicit activ- While total meat consumption per capita has ity, which the Tunisian authorities try to increased by about 40 percent in the past 15 regulate, and which the Libyan authorities years, this is almost entirely due to an increase periodically attempt to ban. Libya does import in consumption of poultry, which is now the large numbers of live goats and sheep annu- largest source of meat in the average diet. ally, and although the Tunisian share of this Declines in ovine meat consumption can be trade is generally small, it can be significant in linked to its increase in price relative to poul- absolute numbers for Tunisia. For example, try meat, while the stagnation of beef con- 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 177 sumption reflects the supply constraints re- and also maintained herd size during periods lated to inadequate production price incen- of reduced carrying capacity of the natural tives. range resources. In addition to feed concentrates, sheep Input pricing and subsidies producers use significant amounts of barley Government price policy for meat was grain as feed, particularly during the dry accompanied by the introduction of subsidies season. The strong demand for barley-further for inputs into livestock production. This was encouraged by strong demand for subsidized done to contain the growth of production concentrates where barley is a major in- costs, to protect producers' incomes, to en- put-has meant that producer prices for the courage continued investment, and to maintain cereal have enjoyed about a one-third increase consumer prices at acceptable levels. Live- in real terms during 1970-90. The favorable stock feed and fertilizer subsidies are the producer prices and growing demand have led major form of subsidy intervention by the to the expansion of barley cultivation to more CGC that affect the sector. marginal lands in Central and Southern Tuni- Subsidies for livestock feed date from 1965 sia previously used for grazing, resulting both when livestock producers were provided with in erosion due to cultivation and in increased subsidized feed concentrates to reduce the pressure on the remaining rangelands. impact of climatic variability on the livestock Although it has been argued that feed herd and to increase the herd's productivity. subsidies have contributed to stagnation of (AIRD 1989, p. 128) This prac dcti,nued forage crop production and thus discouraged to play a central role in Tunisia's drought the integration of livestock and farming sys- mergency assistance to livestock producers tems, this claim is contestable. While the area although as can be seen from Table 7-1, while in forage crops has indeed grown only feed subsidies are higher during drought years, slightly, forage production has increased due a significant subsidy has been in effect in to increased use of irrigation and fertilizers. nondrought years as well. For maize, subsidies Although the removal of feed subsidies could were removed in 1992, as is expected to be further encourage forage production, there is done for soya meal in 1993. Barley continues really no hard evidence to support this. Fur- to receive a subsidy. thermore, concentrate feed is increasingly Feed concentrate subsidies have had a used for the ovine herd, while forage is largely considerable effect on the numbers of small directed to dairy cattle; therefore the effect of ruminants. Subsidized concentrates constitute concentrate feed prices on forage production a growing share-30 percent-of total feed may not be substantial. A more important requirements of the ovine herd during normal disincentive to more rapid increases in forage rainfall years, a share comparable to that for production may be the low prices of beef and the bovine herd. These subsidies, in conjunc- dairy products. tion with the liberalization of mutton and goat The growth in forage production has been meat prices have contributed significantly to encouraged by government subsidies on irri- the profitability of mutton production and thus gation water and fertilizer that are available the rise in ovine numbers. In addition, feed for forage and other crop production. The subsidies and additional, free emergency feed quantity of fertilizer use on forage crops supplies during the recent 1988-89 drought increased by approximately 60 percent be- have maintained the size of the sheep flock tween 1981 and 1986, and the use of fertilizer when in previous droughts, a decline was on forage crops as a proportion of total fertil- evident. Thus the subsidies have encouraged izer use has also been on the rise. Subsidies on both a trend increase in livestock numbers, fertilizer use for all crops increased signifi- 178 VOLUME I cantly between the early 1970s and the mid- tended into fragile marginal lands; production 1980s, although they have declined as a per- takes place largely in the humid and subhumid centage of total CGC expenditures. Use of North. Nevertheless, in an arid country like ammonium nitrate on forage crops doubled Tunisia where water is extremely scarce, a between 1980 and 1986, while use of super- more thorough analysis of the true costs of phosphates rose by nearly 50 percent. Subsi- water subsidies is urgently needed. dies to fertilizer used on forage as a percent- age of total subsidies to fertilizers increased Policy interventions and the 1988-89 from 11.6 percent in 1981 to 17.7 percent in droughts 1986. However, reduction of fertilizer subsidies Severe droughts were experienced by Howeer, educion f frtilzer ubsiies Tunisia in 1988 and again in 1989. Barely half was an objective of the structural adjustment Tunsual aveage and rain occurred in program, and by 1992 subsidies had been the usual average annual rainfall occurred in removed on superphosphates and ammonium 1988, and 1989 was only a marginal improve- nitrate, with remaining subsidies on other ment, creating the largest deficit in rainfall fertilizers expected to be removed in 1993. experienced in the previous thirty years. In fertilizers rexpctedons be irrigatiovaed su- 1previous years, two consecutive years of Similar reductions in irrigation water subsi- drought typically induced a decline in sheep dies have occurred over the past several years, drght typically over a pecene ittle as te auhoriies ave argeed a9 pecent and goat numbers of over 30 percent, but little as the authorities have targeted a 9 percent decline was experienced in 1988-89. Herd real increase in water charges annually, until size only fell by 3 percent for sheep, while costs of operations and maintenance are cov- goat numbers by inreased. ered. goat numbers actually increased. Fertilizer and irrigation subsidies have been This remarkable shift in herders' response successful and encouraginga ttion subsidies havto drought was largely the result of the govern- successful in encouraging the production of ment's drought emergency program and its forage crops by making their production more price and subsidy policies. Barley production profitable. It has been estimated that these in 1988 fell to less than one fifth of average subsidies reduced production costs of green producto ler the ous five yearand forage, silage, and straw by 40 percent, 26 production over the previous five years, and percent, and 34 percent, respectively (Khaldi only recovered slightly in 1989 to 200,000 and Chaffai 1987, p.5). Increases in cultivated tons, remaining well below the country's feed fand Chgal ai 1987p. growing incsensinicutivated requirements. To ensure the availability of forage signal a growing intensification of feed resources to sheep and goats, the govern- livestock production. However, since the bulk ment imported to tons oarle invern- of frageprodctin isusedas eed or diry ment imported 680,000 tons of barley in 1988 of forage production is used as feed for dairy alone. Barley and bran were then provided to cattle, the potential positive impacts of inten- a t Barly subsidized provided to sification are probably not realized. As has herders at sharply subsidized prices, and, as an been argued above, dairy cattle do not cur- emergency grant, small livestock holders with beently argued above, dajo iryonmentaldothreatufewer than 10 head of sheep or goats received rently pose the major environmental threat to afreatoof20gmsfbrlyprhd Tunisia's grazing lands; the crucial problem a free ration of 200 grams of barley per head Tunisia's grazing lands; availae rucial problof cattle during the drought period (AIRD concerns sheep, and available data do not 1989, p. 130). In addition, fodder shrub re- indicate any significant increase in forage serves, established in previous years on public Fertilizer and irrigation subsidies have land, were opened for grazing, thereby buffer- other environmental impacts. On the positive ing the drought's impact. side, the inclusion of forages in the cereal As a result of these measures, mutton and production cycle is often recommended as it lamb production remained fairly stable be- prodctin cyle s oten ecomendd a it tween 1987 and 1991 at about 35,000 tons favors reconstitution of soil quality. Unlike (World Bank 1991a, p. 30). Mutton produc- barley, forage cultivation is not being ex- (ioals cont tob redu- tion also continued to be favored by its rela- 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 179 tive profitability over beef due to government where sheep herds have increased without price controls on the latter. Paradoxically, it commensurate increase in feed production. appears that not only were losses minimized, Rangeland appears to be degrading, exacer- but that some livestock owners even reaped a bated by stabilization of herd sizes in drought net profit from the barley subsidies (AIRD years through the subsidized distribution of 1989, p. 130). barley feed imports, and more generally The combined cost of the emergency inter- through the mild protection of domestic mut- ventions to the government was evaluated at ton production. Following liberalization of 45.3 million Tunisian dinars in 1988 and 41 mutton prices in the late 1970s it became million in 1989. Nearly 90 percent of this was profitable to increase the use of feed concen- for feed subsidies. Explicit agricultural subsi- trates for sheep production, but this has not dies during the two years increased to around been sufficient to alleviate the pressure on double their 1987 level in real terms. In partic- Tunisia's Central rangelands. In addition, ular, explicit production subsidies for feed government policy encouraged the conversion increased by 117 and 180 percent in 1988 and of marginal lands from pastures into cereals, 1989, respectively (World Bank 199 la, p. 6). mainly barley. Not only did this encourage While the sheep and goat population was land degradation through the removal of spared significant losses, Tunisia's grazing permanent cover, it also shifted some of the lands were affected not only by the severe best pasture lands to marginal cereal produc- drought, but also the lack of adjustment of the tion, further shrinking the rangeland resources herd size to reduced carrying capacity. Moni- available to a growing livestock herd. toring is not yet providing continuous infor- Thus, while subsidies may have a beneficial mation on range productivity, but selective effect when implemented in the North and studies have shown that one dry year can perhaps in certain parts of Central Tunisia, reduce the productivity of the range by 60 their impact on rangelands in much of Central percent (FAO 1988, Vol. II), and two consecu- and Southern Tunisia has been negative. The tive dry years result in an even more severe inability of policies to distinguish between decline. Furthermore, rangeland vegetation Tunisia's bioclimatic zones has thus contrib- does not recover immediately, but rather uted to the severe degradation of the country's requires several years of average to good range resources. precipitation to attain normal productivity The government's intent in Central and (Le Houerou 1986). The combination of the Southern Tunisia has been to protect herders reduction in rangeland productivity and the from major losses and from wide income continued high numbers of sheep and goats fluctuations, and it has accomplished this in during drought years has exacerbated pres- the short term. However, no durable solution sures on an already degraded and shrinking has been found to alleviating herd pressure range. that is slowly degrading the capacity of range- land to sustain herds and livestock income. In Conclusions addition, subsidies on feed concentrates have reduced the incentives for forage production. Environmentally, livestock sector policies Sweeping policy changes introduced gradu- have had different impacts on the North, ally since the launching of structural adjust- Center, and South regions of Tunisia. Subsi- ment reforms in 1986 will ultimately have an dies for feed, irrigation, and fertilizers have important impact on how livestock activities promoted intensification of livestock produc- affect rangeland. These policy changes were tion in the North and its integration with primarily driven by budget constraints and cropping activities. This has occurred to a market-oriented strategies, with environmental lesser extent in the Central part of the country, 180 VOLUME II consequences rarely taken into account. The operations in the North. interaction of these policies makes it difficult, While policy interventions may have had a in any case, to predict the magnitude of their small measure of success in social terms, these impact on the various feed sources, and their benefits threaten to be short-lived if the envi- introduction is still too recent for clear trends ronmental consequences of policies are not to be identifiable. It is probable that strength- adequately considered. Income stabilization of ening producer prices for beef will encourage small-scale herders will only be temporary if growth in the cattle herd, after years of stagna- it is achieved through means that result in the tion. This growth will be concentrated initially degradation of the rangeland upon which the in the North where feed resources are more incomes ultimately depend. The long term abundant, but a critical issue will be whether effects of environmental degradation will the derived demand generated by cattle herd ultimately result in the permanent reduction of growth will maintain fodder and barley pro- rangeland productivity and the consequent duction in the face of reduced subsidies on reduction of herd size and incomes from fodder crop inputs, such as irrigation water livestock. In addition, the impacts of increased and fertilizer. In Central Tunisia, the financial desertification and soil erosion will find 'heir returns to sheep herding are likely to decline way into the rest of the economy, affecting with subsidies being eliminated on feed con- agricultural productivity and infrastructure centrates, and some shift to beef production (siltation of reservoirs through increased can be expected. The impact of such a shift erosion). The challenge is to introduce less will depend on whether permanent cattle herds environmentally destructive means to achieve are maintained, or whether incentives encour- the social objectives. age a focus on using seasonally available range feed resources for sale to fattening 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 181 Table 7-1: Budget Costs of Animal Feed and Fertilizer Subsidies, 1975-91 Animal Total feed Fertilizer subsidies Feed andfertilizer Year subsidies subsidies CGC ----------- subsidies ------------ Share of government ---(million current dinars) Share of GDP expense 1975 1 5 n.a. 0.4 1.2 1976 3 1 n.a. 0.2 0.7 1977 6 1 n.a. 0.3 1.0 1978 6 3 n.a. 0.4 1.1 1979 10 4 n.a. 0.5 1.4 1980 17 7 n.a. 0.7 2.1 1981 34 10 138.8 1.1 2.8 1982 32 14 161.4 1.0 2.3 1983 9 15 174.2 0.5 1.0 1984 26 16 257.7 0.7 1.5 1985 21 17 262.2 0.6 1.3 1986 11 17 218.0 0.4 0.9 1987 5 14 196.4 0.3 0.6 1988 37 12 274.4 0.6 1.4 1989 49 20 362.2 0.7 1.8 1990 21 14 317.4 0.3 0.8 1991 18 13 289.2 0.3 0.7 Source: World Bank 1993. Table 7-2: Geographic Distribution of Grazing Lands, 1986 Of which Forage Zone Grazing area Percent improved bushes Percent Humid 89,000 2.4 2,600 n.a. n.a. Subhurnid 296,500 8.1 1,800 19,500 0.5 Semiarid 303,000 8.2 17,900 15,600 0.4 Arid 1,923,000 52.3 15,400 108,000 2.9 Desert 920,000 25.0 21,000 4,300 0.1 TOTAL 3,531,500 96.0 58,700 147,400 4.0 Source: Chouchen and Alaya 1989. 182 VOLUME I Table 7-3: Changes in the Composition of Meat Consumption 1975 1985 1990 Total meat: Consumption (kg per capita) 12.9 16.4 18.3 Percent share: Bovine 25.6 24.1 17.4 Ovine 52.7 38.7 39.1 Poultry 17.1 33.8 40.4 Other 4.6 3.4 31 Sour-ce: Chouchen and Alaya 1989. 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 183 Figure 7-1: Tunisia's Bioclimatic Zones (SRO 27964 TUNISIA C0 M e d i t e r r a n e a n BIOCLIMATIC ZONES ~t > e 37. > . 36N - 30? > / y .- -31~ ~~1 34 ~ i tt3b22 N -. .. : ,l-ARID ;- "! '~ )b (N>' t - - - S 5 J : f a~- Mdetn ! / >* L ~~~: . -1 - .. \; ,7 r SEMIARIL 1 SUBHEM 0tE : HUMIDE 7jA,SDg;2 | | ~~~~~~~~~~32. ' RIVERS N 32' S EASONAL RIVERS SALT LAKES -3 - ISOHYETS IN MILLIMETERL- * GOVERNORATE CAP TAtS * NATIONAL CAPITAL - INTERNATIONAL BOUNDARIES -'.9-dir .l, . roooR,n <'d h ,--. .3 ,' 7 ~' 8-k{ tr r l.KLMTw SA LG IA I By Y sOREoEq;..;.V 31' ThoSo,,r.r,3 (TLB4A ' o.ifiGE._~~~~~~~~~~~~~~~~~~~~2 9" 750 , T,r W.r", I h IOMETERS A L[GE R I A)rf,'p*. MAY 1996 Source: FAO 1988. 184 VOLUME II Figure 7-2: Rainfall and Herd Size, 1961-91 1400 - 1200 iooo ~ ~ i C 800 ~~~~~~~~~~~~~~~~~~~~~~~~~Rainfall I ~~~~~~~~~~~~~~~~~~~~------cattle I - --~~~~~~~~~~~~~~~~~~~goats E -------- .sheep i400 200 - . 0 Source: FAQ Production Yearbooks, various volumes. Figure 7-3: Distribution of Sheep by Region 3 2.5 2 ff 1.5 :: 0.5 ---- ., 'South 0 CY) U') CZ North Year- Cen 2.5- = ---,--ea 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 185 Figure 7-4: Distribution of Goats by Region 7M 2am . = ..... ..._ 3M ------- - --- ---- ~ ~ ~ ~ ~ ~ ~ ~~ - NnT On 1se 19 1#S 119n8 Yar Figure 7-5: Coverage of Feed Requirements by Zone Variation with Rainfall, 1984-86 160 14 _ _ _ _ _ _ _ _ _ _ _ 00 a) Humid Sub- Semi- Arid Desert Humid Ad Bio-Climatic Zone 186 VOLUME II Figure 7-6: Beef and Mutton Prices 600 . ---........ ... ... ... .. 500 - 400 / 4 0 0 ......... . . ......t_~ 300 200 - 100 .. . . . - Beef - Mutton v* , , , I , , I , I I I I 0 Year 7. Tunisia: Livestock Policies and Environmental Impacts During Economic Adjustment 187 Bibliography Associates for International Resources and tems of the World: Hot Deserts and Arid Development (AIRD). 1989. "An Eco- Shrublands. Vol. 12B. New York: nomic Appraisal of the Impact of Govern- Elsevier Scientific Publishing Company. ment Policies on Rangeland Livestock ---. 1990. "Agroforestry and Systems of North Africa and the Middle Sylvopastoralism to Combat Land Degra- East." Unpublished Paper. Somerville, dation in the Mediterranean Basin: Old Mass.: AIRD. Approaches to New Problems." Agricul- Chouchen, A., and K. Alaya. 1989. ture, Ecosystems and Environment 33: "L'elevage en Tunisie." In AIRD, An 99-109. Economic Appraisal of the Impact of Ministry of Agriculture (Tunisia). 1989. "Note Government Policies on Rangeland Live- de Conjoncture No. 3." Tunis. stock Systems of North Africa and the . Direction des Sols. 1991. "Etude de Middle East. Somerville, Mass.: AIRD. l'impact des actions de developpement et Food and Agriculture Organization (FAO). de la lutte contre la desertification." Tunis. Production Yearbooks. Various Volumes. Ministry of Agriculture (Tunisia) and Invest- Rome: FAO. ment Centre, FAO. 1987. "Tunisie: . 1987. Tunisia Expenditure Review of Developpment de la Production the Agricultural Sector. The Seventh Fourragere. Le Controle des Prix et les 5-year Development Plan, the Livestock Subventions dans le Secteur de l'Elevage Subsector, Summary. Rome: FAO. et des Fourrages." Working Paper. Rome: 1988. Tunisie: Programme de FAO. Developpement des Productions "Tunisia: Review of the Prospects for Agri- Fourrageres et de L'Elevage, Rapport de cultural Diversification." Draft. Synthese, Vol. I and II. Rome: FAO. United States National Committee for Man . 1989. "Rapport du Programme de and the Biosphere. 1981. "Environmental Soutien a L'Investissement." Tunisie: Report on Tunisia." Washington, D.C.: Developpement des Productions United States National Committee for Fourrageres et de L'elevage, Programme Man and the Biosphere, Department of d'Investissement et Projets Prioritaires, State. Mission d'Identification Generale, Vol I. World Bank. 1982. Tunisia Agricultural Sec- Rome: FAO. tor Survey, Volume 1. Washington, D.C.: Hamza, A., and H. Bennour. 1990. "Essai de World Bank. cartographie de l'occupation des terres . 1988a. Commodity Trade and Price dans les hautes steppes tunisiennes a partir Trends. 1987-88 edition. Baltimore, Md.: des images landsat thematic mapper." Johns Hopkins University Press. Revue Tunisienne de Sciences Sociales . 1988b. Tunisia: Agriculture Sector 27(102). Tunis: Universite de Tunis. Adjustment Loan. The Livestock Kennedy, P. 1989. "Monitoring the Vegeta- Subsector. Washington, D.C.: World tion of Tunisian Grazing Lands Using the Bank. Normalized Difference Vegetation Index." . 1989a. Tunisia Country Environmen- Ambio 18(2). tal Study, Main Report. Washington, D.C.: Khaldi, A., and A. Chaffai. 1987. "Instru- World Bank. ments et Institutions d'Intervention dans le . 1989b. "Tunisie: Rapport sur Domaine des Prix Agricoles." L'environnement et Plan d'Action Na- Le Houerou, H. N. 1986. "The Desert and tional, Rapport d'Experts No. 3. La degra- Arid Zones of Northern Africa." Ecosys- dation des ressources foncieres et 188 VOLUME 11 hydrauliques." Washington, D.C.: World . 1991c. "Tunisia Small Farmers Bank. Study: Livestock Sub-Sector." Washing- . 1991a. 'Tunisia: Agricultural Expen- ton, D.C.: World Bank. diture Review." Washington, D.C.: World . 1993. "From Universal Food Subsi- Bank. dies to a Self-Targeted Program." Wash- 1. 991b. "Tunisia Small Farmers Po- ington, D.C.: World Bank. tential and Prospects. A Technical Study." Washington, D.C.: World Bank. 8 Economic Policy and Wildlife Management in Zimbabwe Kay Muir-Leresche and Jan Bojo with Robert Cunliffe on Environmental Impacts Wildlife Policy and Utilization in Zimbabwe The wildlife utilization industry in Zimbabwe's private sector developed rapidly once legislation had been changed to allow landowners to benefit from wildlife conservation through utilization. This development has resulted in significant in- creases in wildlife in commercial ranching areas and has effectively halted the systematic elimination of wild animals in these areas. The authors would like to thank Doris Jansen for the cost-benefit analysis; T. Chamboko, F. Murindagomo, and 1. Bond for data on CAMPFIRE and communal agriculture; John Dixon and Kent Redford for commenting on a previous draft; and Liisa Hietala for word processing. The authors, of course, retain responsibility for the expressed opinions and possible errors. 189 190 VOLUME II Wildlife sector prices had been undermined play an important role in the hunting industry. by macroeconomic distortions and the rela- Zimbabwe follows a policy of sustainable tively closed marketing system, particularly in utilization and views all its mammals as a communal areas. More competitive marketing renewable natural resource to be managed. has changed this significantly since the early Protection of species and species diversity 1990s. Better access to foreign currency and takes precedence over protection of individu- effective currency devaluation have moved als and groups. The estimated elephant popu- market prices closer to opportunity costs. lation in 1960 was 32,700 and by 1988 it was However, market prices do not reflect the true up to 52,000, despite the fact that 44,500 social value of the wildlife resources because elephants were culled during those 28 years environmental impacts, genetic resources, and (Martin, Craig, and Booth 1989). At the end existence or bequest values are not incorpo- of 1992 it was estimated that there were some rated (see Annex 8-A). Where the externalities 80,000 elephants in Zimbabwe, and there are are significant, attempts could be made to increasingly serious conflicts as elephant internalize them through a system of taxes and populations exceed carrying capacity in na- subsidies. tional parks. Elephants are very important in The most lucrative outputs from the wild- wildlife enterprises, particularly in the peasant life industry in Zimbabwe are photographic farm sector where they account for some 70 tourism and safari hunting. Hunting is impor- percent of the revenue. tant for developing the wildlife option in areas with lower concentrations of wildlife and poor Development of a wildlife industry on infrastructure. Some specialized wildlife privately owned land production units are also offering lucrative From 1960 farmers were given permits to options-e.g., crocodile and ostrich farming. . . Meat, hides, and horns are outputs of minor ul and trandwildlife. The Department .'. ' . ~~~~~~~of National Parks and Wildlife Management slgmficance, and ivory sales are negligible significa , n o (DNPWLM) noted the positive response and because of the Convention on the International how quickly wildlife numbers recovered in Trade in Endangered Species of Flora and certain areas. The Parks and Wildlife Act Fauna (CITES) regulations. (14/1975) was implemented in 1975. It allo- Zimbabwe is well endowed with large cated full custodial rights over wild animals mammals, including elephants. The central (except for "specially protected" species) to highveld is intensively farmed with limited appropriate authorities (landholders) while the wilderness and wildlife, but with some oppor- animals are on their land. Specially protected tunities for recreational tourism. The periph- animals may only be killed in defense of life, ery, particularly along the north, south, and and their remains may not be traded. Re- western borders has large concentrations of stricted species may be utilized upon receipt game. The large-scale ranching areas are not of special permits. Local committees may also allowed to run buffalo with cattle. Buffalo declare a species specially protected within its were totally eradicated in these areas to com- ply with European Community (EC) veteri- ow ara'oee,i.ms ntne h nary regulations. The lack of buffalo signifi- animals do not benefit from this protection, nantly regucesaretions. the wlackliferanchn buf- since it is loss of habitat rather than hunting cantly reduces retumns to wildlife ranching, but that is the major danger. there is a movement to reintroduce buffalo in No person can hunt, capture, or remove any some ranching areas. Buffalo were not eradi- animal or plant from privately or communally- cated in those communal areas where the owned land without the appropriate author- tsetse fly precluded cattle, and the communal ity's or landholder's permission. The land- areas, which still have buffalo and elephant, holder may hunt or market any plant or animal 8. Economic Policy and Wildlife Management in Zimbabwe 191 on his land (except those species listed as extensive game or mixed game and cattle "specially protected" or "restricted"). How- ranching. The wildlife generally is not used ever, to market a live animal or a trophy, the for meat production, and any meat produced is landholder must first obtain a specific permit. sold locally at prices significantly lower than Under the 1975 Act, appropriate authority was beef prices. Hunting in ranches is also often immediately granted to all private landholders, sold as a complement to hunting safaris on and in the communal areas authority remained state hunting concessions and in communal with the state. areas. Hunting requires less infrastructural The introduction of wildlife legislation development and lower animal densities than allowing landholders access to their resource are necessary for general tourism. Once the has been successful in the large-scale farming infrastructure has been developed and areas areas of Zimbabwe. Photo-safaris and become less remote-and provided wildlife wildlife-related tourism have grown signifi- density is relatively high-the greater volume cantly and rely heavily on the continued via- feasible with photo-safaris may make tourism bility and attractiveness of the national parks ventures more profitable than hunting. and wildlife estates. From 1972-82 there was Gross income from international safari effectively only one tour operator, but by 1990 hunting increased from US$2 million in 1984 there were more than fifty registered opera- to US$9.3 million in 1990. Jansen (1990) tors. This has resulted in widespread increases estimated that if the entire hurtting quota (2 in wildlife in all the less arable farming areas percent of animal populations in state hunting and even in the intensive cropping zones and communal areas) were sold to foreign where farmers are reintroducing game. In clients, it would earn US$13 million, which 1960 there were only three game ranches and together with an estimated US$7 million by 350 square kilometers were designated for commercial ranchers indicates that hunting wildlife outside state land. By 1990, 55,000 revenues could be doubled. square kilometers had been partially or com- Aside from safaris, other wildlife-based pletely allocated to wildlife enterprises in activities contribute to the economy: commercial and communal farming areas, * The value of live animals for breeding with an additional area of about equal size in purposes has increased dramatically. The the parks and wildlife estate and in forest DNPWLM operates a central allocation reserves. There is a Wildlife Producers Asso- system with long waiting lists for ani- ciation (WPA) with a membership of over 10 mals, and a thriving private trade percent of Zimbabwe's large-scale farmers. emerged in the late 1980s. In 1989 there Although membership has increased in recent was a waiting list for approximately years, it has been the increase from passive to 5,250 animals to restock private land active membership, which has been signifi- with wildlife from DNPWLM. Live cant. animal prices in the private sector rose Safari hunting is the most important com- some 400 percent in three years. ponent in large wildlife enterprises, and regis- * Crocodile ranching has been encouraged tered safari operators have increased signifi- since 1960, and producers are required cantly since 1980. In the predominantly agri- to return a proportion of hatchlings to cultural zones, wildlife tourism is an increas- the wild. Numbers in the wild have ing and generally supplementary enterprise, increased significantly, and earnings in with some farmers investing in visitor camps, the crocodile industiy have increased lodges, and recreational facilities. The large from US$300,000 in 1983 to US$2.6 wildlife enterprises are found in the arid areas million in 1989. of Zimbabwe where farmers concentrate on * Ostrich ranching has been very success- 192 VOLUME II ful. Where wild eggs are harvested, a was returned to district councils, was used for proportion of the chicks are returned to development throughout the district, although the wild. As with crocodiles, those re- only a few communities in the district pay the turned are a slightly greater proportion cost and actually live with the wildlife that than has been estimated would normally generated these revenues. Thus, the local survive to that stage. communities still have little sense of owner- Ivory sales are important only in the ship or control in some districts and subsis- state and communal sector where they tence hunting is still illegal. One of the major earned almost $12 million between 1981 complaints is that district council authorities and 1988 from approximately 31,000 receive all the wildlife revenue and then tusks (Martin, Craig, and Booth 1989). allocate these revenues over a much wider Current elephant ivory stocks held by population than that living with wildlife. Both DNPWLM are valued at over US$10 wildlife and tree resources are generally mnillion with a significant proportion due viewed as belonging to all Zimbabweans, but to be repatriated to peasant communities progress is being made in some districts, when sales are made. The CITES ban which have recognized that for people in these has meant that very little ivory is sold, marginal areas, "their wildlife is their cattle." with most of the sales being made to With effect from 1988, legal control over Zimbabwean carving companies for sale wildlife resources has been granted to districts to the local tourist market. that apply for it, provided that they meet specified requirements. In 1993 there were Development of a wildlife industry on twelve communal lands in Zimbabwe with communally owned land appropriate authority and more than 19,500 Since 1980 the DNPWLM has managed the square kilometers of communally owned land officially designated to include wildlife in the wildlife in these areas on behalf of the com- laduestmanmngdudrth munities. They have leased hunting rights and CAMPFIRE program. It was anticipated that collected the fees on behalf of the communi- CoMmuItE form. anaturalesource ties, but were required to surrender revenues cooperaties, mngmenttrut orrcp to the national treasury. Although in principle coesawith management trustso or compa- the treasury is supposed to reallocate those nies with membership open to all adult males revenues to the concerned communities on the and females in the community. Each commu- presentation of proposals for development nity would draw up their own constitution. projects, in practice a large proportion of the The allocation of rights would be affected by rects in raice alrgexprpotn of the the population-resource ratio. Defining the revenues was retained. For example, of the boundaries of a communal resource area in the Z$ 5.8 million earned by safari hunting in CAMPFIRE2 program would be a process of communal areas in 1987/88, only Z$ 3.3 p g million was paid out to district councils (Zim- niato bewesenthes natin, distri and babwe Government 1988) ~~village representatives. Water, forestry and babwe Government 1988). wildlife would be closely controlled with The Parks and Wildlife Act of 1975 en- spcfecomntmagmntndces abled the Minister to designate district coun- specified community management and access cilsin cmmunl lnds s apropiateauthri- strategies but with arable land being accorded cils in commnunal lands as appropriate authori- idvda epniiiy ties with full custodial rights to their wildlife, but this only began to be implemented in 1988 under the umbrella of the Communal Area 2Communal Areas Management Program for Management Programmes for Indigenous Indigenous Resources. Much has been written on Resources (CAMPFIRE). The limited wildlife CAMPFIRE; see the original document by Martin revenue generated in communal areas, which (1986) Murphree (1988); Jansen (1990); Child and Peterson (1991); and Murombedzi (1992). 8. Economic Policy and Wildlife Management in Zimbabwe 193 In fact, appropriate authority is only granted Z$ 35 million, and other game products to the district council, and local communities (skins, meat, and ivory) account for Z$ 1 have yet to be given the opportunity to form million. These contribute more than the for- effective resource management units. Most mal sheep and goats sector. As with agricul- CAMPFIRE projects are still a long way from ture, the multiplier effects of tourism and achieving the objectives of granting control wildlife are strong, and they are important over resources to local communities. CAMP- contributors to foreign exchange earnings. FIRE has, however, succeeded in generating Tourism operations qualifying for the export debate over access to resources and may retention scheme claimed Z$ 165 million in eventually fulfill some of its original goals. 1990/91. It has been crudely estimated that Some communities currently see CAMPFIRE during this period, the wildlife sector gener- as a means of increasing agricultural potential. ated Z$ 344 million in foreign exchange This development could eventually reduce earnings (Jansen 1992). National data on wildlife and undermine CAMPFIRE, thus export earnings are not available for this creating tension between villagers and imple- period, but this is probably in the region of 20 menters. percent of total exports. The industry has strong backward linkages Wildlife and the economy with the manufacturing sector, as well as some It is extremely difficult to ascertain the forward linkages with the informal sector, producing curios and clothing. Most tourist value of Zimbabwe's wildlife industry. The industry inputs are locally obtained. In the consumption of bushmeat has been virtually large hotel sector, 80 percent of fixed costs are ignored, although the consumption of rabbits, loaexndtrsbuitscoetoO p- rodents, insects, and wild flora are widespread local expenditures, but it is close to 100 per- rodets,insets,andwildfloa ae wiespead cent for the safari camps and lodges. For all throughout the country. The consumption of tor venge costs - . . ~~~~~~~~~tourism ventures, 85 percent of variable costs larger mammals is almost exclusively con- ' X fined to isolated communities with high-wild- are local-the major linkages being food and life and low-human populations. Assuming beverages. The industry iS a high foreign that there are some 25,000 households living currency converter, but relies heavily on in such areas, a rough estimate of the value to adequate transport and communications facili- the nation of large mammal bushmeat is some ties, which are foreign currency dependent. Z$ 4 million per annum based on the 60 per- Tourism has grown significantly since cent contribution to total income made by 19987 with bed occupancy up to 50 percent in wildlie to ousehod incme in ngwa, 1990 despite an increase of 86 percent in the wildlifezio houseyh(Muindagome 1988). Anumber of beds available. Employment grew Zambezi Valley (Murindagomo 1988). 20 percent in three years, compared with The contribution of the wildlife industry to percultuin e e elsewhere ith the formal sector is also difficult to ascertain. nonagricultural employment elsewhere in the Published statistics do not separate tourism or economy, which grew by 0.5 percent. wildlife based industries as a proportion of Nonconsumptive tourismpis considerably more grossdometic rodut (GD). Mnistrial labor intenSiVe than sport hunting, and more gross domestic product (GDP). Ministerial than 50 percent of those employed in hotels statements have indicated that in recent years a tourism has earned between Z$ 300 million and lodges could be drawn from unskilled and Z$ 1 billion per annum or between 2 and local populations (Bond 1992). 5 pecentof GP. Te mot crdibl revnueThe role of wildlife in the economy iS 5 percent of GDP. The most credible revenue significantly increased if adjustments are estimate appears to have been around Z$ 500 made to reflect the opportunity cost of the million in 1990/91. Safari hunting and wildlife foreign recy the by ctor (se management services add another Z$ 50 foreign currency earned by the sector (see million. Crocodiles and ostriches contribute Table 8-1). For most of the 1980s, various 194 VOLUME 11 estimates indicated that the average overvalu- receives more than the 700 millimeters' an- ation of the local currency was 50 percent, nual average considered necessary for semi- climbing to over 70 percent by 1990. From intensive farming, and in most parts less than mid-1991, the government started a deliberate a third of this area is actually arable. In the policy of devaluation to realign the currency. intensive farming systems followed in Zimba- Tourism in Zimbabwe is wildlife- and bwe, the natural growing season is confined to wilderness-based and has significant potential the rainy months. for expansion. The great numbers of large The indigenous vegetation is savanna mammals, particularly elephants, and the grassland along the central plateau, with relatively sophisticated infrastructure give wooded savanna throughout most of the rest Zimbabwe certain advantages over other of the country, and with some montaneforest tourist destinations. The current structural in the Eastern Boarder districts. In general, the adjustment program will encourage invest- topography, soils and climate of Zimbabwe ment in export industries with more realistic are not favorable for intensive agricultural exchange rates and easier access to imported production. More than 75 percent of the coun- inputs. The international recession will have try is subject to conditions that make dryland less effect on tourism to Zimbabwe, which crop production a risky venture. While agri- relies most on high-income tourism. culture accounts for only some 10-14 percent of GDP, it has a major impact on growth Economic Development through its backward and forward linkages, and Wildlife and through its contribution to export earnings and to both formal and informal employment. The national economy, the resources, and Since independence, there have been some land use systems significant advances in the provision of social Zimbabwe lies within the tropics. However services, leading to a decline in infant mortal- given altitudes of betwthent600andHowever ity and increases in education and access to given altitudes of between 600 and 1,200 phsca.Hoer,tseoilsrve, meters, most of the country has a subtropical theriwith High defse expenditean climate, which is perfect for year-round tour- larget e ion of tepuisector, rud ism. he riny easo coicide wit norhem Iarge expansion of the public sector, resulted ism. The rainy season coincides with northern insgfcatnreesngormntep- winters, but even during this season, the days iu t eoom is ghlymregted,end are mostly sunny, with the rain occurring in barrer To entry significatsdies sporadic thunderstorms and late afternoon to thentrialor, aglarg prprioneo showes. Zibabweis bradly ategoized to the Industrial sector, a large proportion of showers. Zimbabwe IS broadly categorized GD spnongvrmtcnupin into five natural regions, based on rainfall, expent an r opulationsgro thav Region 1 has the most rainfall, but because of esuled in veryapipoitin eono mic topography, production is dominated by exotic peromne, it GDP pesapita eclnin timber plantations, tea, coffee, and horticul- percenmsnce inDPenencBy19 Zimba- ture. Region 2 normally has more than 700 berhad a lngerextenal Bt of some mllimeters of rain and is the main crop-grow- Z$ 12 billion. ing region in the country. Regions 3, 4, and 5 Since independence in 1980 there are no are increasingly arid and, although cultivated longer de jure racial divisions, but land-size in communal areas, are predominantly only categories remain effectively unchanged since suited to extensive livestock production. The permission to subdivide is almost impossible annual variations are large, and the reliability to obtain. Ownership of the land in the former of monthly rainfall is much lower than the Tribal Trust Lands is officially vested in the seasonal total and decreases in general from President, but land is farmed along traditional north to south. Only 37 percent of the country 8. Economic Policy and Wildlife Management in Zimbabwe 195 communal tenure arrangements. This implies tions from the central government. This may good security over homestead and arable land, change with the increased awareness of the but the grazing and woodland areas remain potential role of wildlife in generating growth effectively open access. Land allocation con- through foreign currency generation and the tinues to be a source of conflict within com- strong backward linkages which exist in munal areas since local government, political Zimbabwe for the tourist industry. The wild- parties, and tribal authorities all have varying life sector, despite the advances with CAMP- degrees of control. The commercial sector is FIRE, is still seen as white dominated and made up predominantly of large-scale farms carries much resentment. Black entrepreneurs (the former white farming area) and a few need to be given exposure and opportunities to smaller-scale freehold farms (from former participate in this growth industry. African Purchase areas). The communal sector includes resettlement schemes where tenure is An economic analysis of wildlife not freehold but retained by the state. utilization The government is involved in an active land resettlement program and has declared its intention to take over 50 percent of the large- Child (1988) carried out a detailed time- scale sector for reallocation to smallholders. series analysis of cattle and wildlife on Buf- Tenure arrangements in the resettled areas are falo Range in the southeastern lowveld. The very uncertain. Insecure tenure may have analysis showed that the cattle section gross serious implications for environmental integ- profits increased during the 1960s, but fell rity. The uncertainty of the resettlement pro- rapidly from a peak in 1975 to losses in 1984. gram has also created long-term insecurity for The falling profits were a result of declining large-scale farmers, which could also affect herd productivity with calving rates dropping the environment. For most of the population, and livemass gains falling off dramatically, however, there are high expectations that old even before the major drought. This was wrongs will be redressed soon by land redistri- primarily a result of veld degradation caused bution. Given the limited capacity for govern- by overstocking. ment to implement the program expeditiously, The wildlife populations on Buffalo Range, a serious crisis of expectations may result in however, continued to grow, and the range attempts to settle people on any available land. was less degraded in that sector of the farm. In This could have serious consequences for 1973 Taylor carried out transections on the communal areas with low human-population ranch to compare environmental conditions, densities that are developing wildlife systems. but the results between the cattle and wildlife Zimbabwe's population is growing at 2-3 sections were indeterminate (Taylor in Child percent per annum and, given the extreme 1988). These transects were repeated in 1985, poverty of communities in the marginal lands, and this time environmental conditions in the population pressure could be translated into game section were significantly better than in demands for access to resources in neighbor- the cattle section. Child concluded that despite ing parks. The overgrazing and the cultivation the heavy implicit subsidies for beef which of poor land have led to widespread erosion favored the cattle section in most years and and depletion of the forest and grass cover, so the overvalued exchange rates which greatly that the neighboring parks appear to have disadvantaged wildlife, wildlife offered the resources desired to support the increasing most lucrative and sustainable option for population. Buffalo Range (Child 1988). Political support for wildlife is negligible, Jansen, Bond, and Child (1992) carried out and the wildlife sector receives limited alloca- a survey of eighty-nine cattle, wildlife, and 196 VOLUME II combined cattle-wildlife ranches in 1989/90 in avoid favoring the wildlife option. the more arid Natural Regions 4 and 5 in order The analysis showed both cattle and wild- to determine the relative profitability of cattle life to be economically profitable with the and wildlife. They estimated returns to invest- average rate of return on investment for wild- ment and the comparative advantage of ranch- life enterprises at 21.5 percent and at 13.1 ing cattle only, wildlife only, and combined percent for cattle. The returns to land, how- cattle and wildlife ranches. The survey year ever, were higher for cattle than for wildlife was an average rainfall year, and there were after adjusting for government controlled no unusual occurrences which would mean prices and exchange rate distortions, with unusually high or low profits for either com- average economic returns to land for cattle at modity. Z$ 17 per hectare versus Z$ 14 per hectare for Financial. Cattle-only enterprises had an wildlife. average 1.8 percent financial (private) return The financial results from the study are on investment, and the return to cattle on consistent with the reported shift into more ranches combining cattle and wildlife was 2.6 wildlife enterprises, since these lead to a percent. The weighted average return of cattle higher net absolute return per hectare than enterprises was Z$ 2.78 per hectare. Only four cattle enterprises, even when such are com- of the seventy-seven ranches producing beef bined with minor wildlife activities. Returns had greater than 10 percent return on invest- are generally low per area unit, but it should ment. The speculative return on holding land be emphasized that the average farm in the was excluded from all analyses. Only three sample is large, some 22,000 hectares, and cattle enterprises had returns greater than that returns are net of production costs. Fur- Z$ 25 per hectare. Thirty-nine percent of the thermore, many other factors (the possibility cattle enterprises had a negative adjusted net to offset losses in these enterprises against revenue and to continue in operation most of capital gains elsewhere, maintaining a certain the ranches were destocking or borrowing. life style, the prospect of speculative returns to Wildlife-only ranches were the most finan- land holdings and attachment to sunk invest- cially viable with average returns on invest- ment, as well as unfamiliarity with wildlife ment of 10.5 percent. More than half the enterprises) all serve to explain the inertia in wildlife enterprises had a greater than 10 leaving a financially unprofitable business. percent return on investment, and only four Role of policy. It is clear from a comparison had negative adjusted net revenue. The across financial and economic results that weighted average return of wildlife enterprises profitability is currently greatly depressed by was Z$ 5.8 per hectare. economic policy interventions. Both cattle and Economic. An economic (social) analysis of wildlife enterprises show much higher profit- these returns was carried out using estimated ability under economic prices. With respect to opportunity cost prices to incorporate the employment, wildlife enterprises are less effects of market prices not reflecting true wage-labor intensive and require more skilled social values with respect to the price of manpower than cattle enterprises, making cattle, the exchange rate, price of land, the cattle the preferred option from an employ- interest rate, and the cost of degradation from ment perspective. This results from the current overstocking. Jansen, Bond, and Child (1992) heavy concentration of the wildlife enterprises vary their assumptions in a number of sensi- on safari hunting. Tourism ventures have tivity runs, but the conclusions essentially all much greater employment potential and are remain the same. The base-run assumptions becoming an increasingly important compo- assumed a particularly conservative estimate nent of the wildlife industry. of overvaluation of the Zimbabwe dollar to 8. Economic Policy and Wildlife Management in Zimbabwe 197 The communally owned, small-scale farm plains game hunting sold on most commercial sector ranches. Fifty-two percent of all elephant and Between January 1989 and April 1991, 39 percent of all buffalo trophies came from twelve districts received "appropriate author- communal areas in 1990 (Bond 1992). ity" over their wildlife, although in many of CAMPFIRE revenue generation and distri- these districts, not all wards have a wildlife bution. Although CAMPFIRE programs are resource. All these districts are engaged in administered by district councils there are obtaining revenue from wildlife utilization usually only a few wards which have the (i.e., entering into contracts with safari hunt- wildlife resources to be actively involved in ing companies and tour operators), are at- wildlife projects. It would be desirable to devolve authority to ward or even village tempting to manage their wildlife (i.e., pro- deov uhrt owr ree ilg temping o mnagether widlif (ie., ro- levels, but this is not possible because the tecting it and setting quotas for its exploita- tion), and are distributing benefits to the district council is the lowest level of govern- participating wards and "producer communi- ment with legal authority as opposed to advi- ties." Administratively, communal areas are sory powers (Peterson 1992). In addition, the divided into district councils which have legal fugitive nature of the resource and the large status and are serviced by the Ministry of areas required for safari hunting would make Local Government and Housing. These dis- cooperation between neighboring producer trict councils are then further subdivided into units important. wards with ward development committees The total gross income from wildlife accru- (WADCOs) comprising a number of villages ing to the district councils from CAMPFIRE .wAt copisne n.mber co vmilages Z$ 3 million in 1991. Total wealth gener- (VIDCOs). The WADCOs and VIDCOs are ated, but accruing more widely (e.g., including recent political institutions and are more safari operators and airlines), is considerably effective in some areas than others, greater. The DNPWLM has issued "Campfire By 1992 CAMPFIRE had grown to encom- Guidelines" which recommend that no more passbyth 1992 CaMPFR heaad alargrpown latoienc than 15 percent of revenue be allocated to the pass both a large area and a large population dsrc oni ntefr falv,adn among whom the benefits of the wildlife are district council in the form of a levy, and no to be shared. Safari hunting dominates the more than 35 percent be spent on management revenue earned. Most districts now put their costs, so that at least 50 percent of revenue hunting quotas out to competitive tender. The can be passed on to the community in the form safari operators then market the hunting inter- of cash or project benefits. These guidelines were not met in four of the nine districts for nationally. International safari hunting ac- wihdt saalbe nNaiym n counts for more than 85 percent of the revenue which data is available. In Nyaminyami and in most CAMPFIRE areas. The bulk of the Guruve, community benefits were distributed revenue is derived from elephant hunting. Any as ward dividends, but represented only one- ban on imports of elephant trophies into the third of revenue and in Binga only 40 percent. major client regions-Europe and the United In Hurungwe none of the wildlife revenue was States-would have a large negative impact distributed to the producer communities: the States-ommunityldlhae aslargemnegative impcouncil spent all of it on a truck. In contrast, onucomutysuse r wildlife scemes. normallyrward dividends represented considerably more Quotas used for, culling are normally~ re- ta 0ecn frvnei ordsrcs stricted to those animals that have a limited than 50 percent of revenue in four districts. safari market, e.g., impalas, and no districts Not all of the earned revenue is available have yet established mechanisms to legalize, for the provision of benefits to the communi- allocate, and manage subsistence hunting. The ties, since there are costs incurred in managing communal areas play an important role in the the wildlife. The larger programs, such as safari hunting industry and complement the Nyaminyami and Guruve, have wildlife man- 198 VOLUME II agers who receive salaries and are provided and poaching to the concerned villages. This with vehicles and housing. Many districts would give them hunting access to those choose to employ game guards whose main animals where there is little conflict with the tasks are antipoaching and problem-animal more lucrative safari hunting. However, they control. should be required to pay royalties to the local Capital costs, principally for vehicles and village community to avoid overexploitation. electric fencing, have been provided by donors The management costs in wildlife protection in several districts, and thus far CAMPFIRE would be significantly reduced as the villagers revenue has generally not had to be used for would have greater incentives to police them- capital costs. There is a danger, however, that selves and report outside poachers. It is widely the recurrent, maintenance, and eventual accepted that self-policing is very much less replacement costs of the donated capital assets expensive and more effective than centralized may lead to increased management costs, and regulation in theory; the challenge is to de- thus result in a smaller percentage of the velop appropriate institutions for self-regula- revenue being available for community bene- tion. fits. The budget of the Nyaminyami Wildlife Cost-benefit analysis of wildlife utilization Management Trust (NWMT), shows a buildup in communal areas of a "wildlife management bureaucracy." Recurrent costs have increased 360 percent in Jansen (1992b) undertook a cost-benefit twoRyeaurre os, s wheeasevee increased by0 pe t ianalysis in three districts. The analysis pro- two years, whereas revenue increased by 79 jetfowrcvrigaenyrprodad percent. If the cost of living with wildlife jects ward coerin a of the CAMPFIRE remains higher than the benefits received in bgnmihtefrtya fteCMFR remains higher thantheenefitn program in each district. Scenario 1 assumes the producer communities, then the policing no change in management, dividend distibu- costs will continue to escalate as locals con- noaching, anageman imirain. Ths tinue to encourage poaching and human immi- tion, poaching, and human immigration. This gration. The NWMT has capital assets of leads to a decline in revenue, and increases in nearly Z$ 800,000. In 1991 it withheld 15 costs and in poaching and immigration in nearly~~~~~~~~~~ Z.8000 n191i ihed1 Nyaminyami and Guruve. In Mahenye, by percent of its net revenue for a depreciation NymyaindGuv.IMhee,b rerve, ofurth red revenue ava .iabl contrast, the status quo is based on continued reserve, further reducing the revenue available godmneetadcnrlofpchg for distribution. ~~good management and control of poaching Ino the districutson. wheretheCAMPFIRand immigration. Macroeconomic reforms are not implemented, and the Zimbabwe dollar program has been implemented, more recently remains overvalued. All values are presented the proportion of wildlife revenue distributed in constant 1991 Zimbabwe dollars. to the wards is considerably higher. This may In Nyaminyami there have been substantial be because the "wildlife management establish- increases in recurrent costs during the first ment" has not yet had an opportunity to be three years. This is in part due to the inexperi- created. Alternatively, perhaps districts that enced management and to the decision to are only just now beginning to implement employ sixteen game guards with conditions CAMPFIRE (and the nongovenmental orga- of service similar to that provided by nization (NGO) collaborative group that DNPWLM. It is important that the community assists them), are leaming from the mistakes find more cost-effective poaching control of the "pioneer" districts (Nyaminyami and mechanisms. Scenario 1 assumes that this Guruve) and will try to keep wildlife manage- situation is not rectified: efforts to date to do ment costs within or below the 35 percent . . . guideine.Mechaismscoul be etablshed so have failed. As a result, it is assumed that guideline. Me m crecurrent costs increase at a rate of 11 percent that would give full control over utilization in real terms. 8. Economic Policy and Wildlife Management in Zimbabwe 199 Capital costs, which have been substantial continue to be spread evenly among all twelve and almost totally financed by donor aid wards. As a result, benefits per household (Zimbabwe Trust), are assumed to represent never exceed Z$ 100 per annum and in fact replacement of existing assets plus the addi- become negative after year 7. Thus, even tional buildings planned for NWMT. No though this project has a positive NPV for the distinction is made between the source of district, it should not be implemented, unless finance for the capital costs, since they repre- some alternative mechanism of allocating the sent use of resources that would presumably net benefits is instituted. In other words, a be available elsewhere in the economy if they positive NPV is a necessary, but not a suffi- were not used for NWMT. Despite the prob- cient, criterion for going ahead with the pro- lems of continuing with the status quo, the net ject. If the project is to be implemented as an present value (NPV) for the district is positive, alternative to some other land use option, then Z$ 1.5 million using a discount rate of 10 its NPV will need to be higher than the com- percent, and Z$ 333,000 using a discount rate peting options, and since wildlife is a less of 40 percent. well-recognized option, it will need to be Scenario 2 is a more optimistic scenario, considerably higher before it is chosen in including 25 percent adjustments for over- place of conventional options. valuation of the Zimbabwe dollar. It assumes Scenario 2 shows that with a more optimis- that the Economic Standard Adjustment Pro- tic revenue generation, cost control, lower tax gram (ESAP) succeeds, local management levy, and lower population increase, the bene- improves, and poaching and immigration are fit per household increases to Z$ 470 per controlled. As a result, the net present value to annum (in real terms) in year 10 (Figure 8-1). the district, based on a 10 percent discount This assumes that the dividends are paid in the rate, increases to Z$ 3.6 million. The optimis- form of household cash dividends, with social tic scenario assumes that present constraints service or infrastructure projects paid directly on performance, particularly in Nyaminyami by farmers. and Guruve, are eased, and ESAP succeeds. Similar cost-benefit analysis for the Revenue increases by 10 percent annually as CAMPFIRE project in Guruve District shows a result of devaluation of the Zimbabwe dollar that like Nyarninyarni, the project has received and good wildlife populations due to declines substantial capital investment donated by in poaching activity and immigration. This Zimbabwe Trust, and has a good revenue from scenario assumes that recurrent costs do not hunting and tourism. Also, like Nyaminyami, increase in real terms after 1993 because of it has sizeable and increasing recurrent wild- improved management, capital costs are life management costs. If the ward dividends restricted to replacement, immigration is were to be split in the future among all eight controlled, and population increases by 3.5 wards, as is the case in Nyaniinyami, Scenario percent. Ward benefits are distributed only to 1 shows that the benefits per household dimin- producer wards. ish and in fact become negative by year 10. This type of analysis fails to take into Under the optimistic scenario 2, with benefits consideration the distribution of the net bene- split between only three producer wards, the fit stream. If the net benefits do not accrue to distribution increases to over Z$ 500 per the producer community, the link between the household and could result in a viable wildlife benefits and costs of wildlife will not be utilization option. This assumes that the link made, and the current trend of encouraging between the benefits and the wildlife resource wildlife eradication, immigration, and in- is successfully established. This is not yet the creased cattle stocking will continue. case, but the situation is improving steadily as Scenario 1 assumes that the ward benefits institution building at the ward and village 200 VOLUME I level is taking place. It should be noted that munities. Even where it is possible to define under both scenarios, the NPVs are positive, boundaries, it is difficult to ensure that those but the distribution per household is signifi- paying the highest costs receive the greatest cantly different. benefits. Even within wards, there are differ- Cost-benefit analyses of the CAMPFIRE ences in animal densities and in crop and project in the Mahenye ward of Gazaland livestock damage due to wildlife. Defining the District (Chipinge) indicates that the hunting producer community is difficult and even revenue is considerably smaller there, but so where the district has decided that the resi- are the recurrent and capital costs. As a result, dents "living with wildlife" should be the the NPVs are not much lower than in principal beneficiaries, this is difficult to Nyaminyami and Guruve. Scenario I shows determine. A number of criteria can be the effect of poor macroeconomic policies, but used-according to animal densities, based on with wildlife revenues distributed solely to the the exact location of where the trophies were producer community, household dividends taken or based on relative crop and livestock peak at Z$ 813 in 1993 and decline thereafter. damage. This cannot necessarily be done at Under the optimistic scenario 2, the household the ward level, and as in Beitbridge district, it dividends steadily increase, and are signifi- may be one or more village (VIDCO). It is cantly larger than in Nyaminyami and Guruve, only when wildlife is accepted as "belonging" exceeding Z$ 1,000 per household from 1993. to those who live with it and pay its costs that Although the revenue potential for the district the various models that enable the producer is lower, there are considerably fewer house- communities to receive larger shares will be holds involved even if Nyaminyami and accepted without creating community strife. Guruve focus the benefit distribution more It is important for communities to play an carefully. This indicates the importance of active role in deciding how the wildlife reve- limiting the population benefiting from wild- nue is to be distributed, but even if they are life if revenues generated are to be meaningful closely involved, there is often a problem with to households and change their decision- the implementation of community projects. making to favor wildlife. This is largely due to a lack of implementation skills at the district and ward levels. In many Summary of necessary conditionsfor suc- wards, the ward dividends remain in the bank, cess of CAMPFIRE projects with their purchasing power eroding due to the There are a number of principles that can be high inflation rate in Zimbabwe. Also eroding is the link between the wildlife costs and universally applied. The most important of wildlife benefits: the costs are felt immedi- these is that the benefits must outweigh the costsfor the residents if they are to encourage ately in terms of crop damage and threat to wildlife by reducing poaching and habitat human life, while the benefits are delayed. destruction. A unifor distributin mThese delays unfortunately parallel the delays destuctin. Aunifrm dstriutio modl of these communities experienced in the past in the benefits shared among all the wards in a district, or even all the wards with some wild- getting wildlife revenue back from Treasury. life, is unlikely to result in benefits outweigh- It is also essential that the communities ing costs in the important producing areas receive an adequate proportion of the value of Where wildlife revenue is used to fund social the wildlife sold ftom their land. Safari hunt- services and infrastructure projects in a dis- ing involves very high overheads in marketing trict, the links between costs and benefits and operations, but it is only with the intro- become weak. duction of more open and competitive market- With the fugitive nature of wildlife, it is ing that prices paid by operators increase. The difficult to closely define the producer com- local communities need to establish openly 8. Economic Policy and Wildlife Management in Zimbabwe 201 competitive systems and need to be trained This may be partially due to the fact that the and assisted in marketing their resources to irrigation schemes do not necessarily deplete operators. much of the surrounding woodland, even though the increase in population pressure will Comparative analysis of wildlife and subsis- do so in the long term. These resources play tence agriculture in communal areas an essential role in coping with drought and in In the communal areas in Zimbabwe, ani- improving the quality of nutrition, and the mal production regimes largely consist of impacts of reducing access to these resources must be Incorporated in all development cattle that are used primarily for draft power, straegies, rporarey in areaslopmedt as a store of wealth, and for providing milk 1992). and manure. The real and perceived benefits ) of cattle are high, and the costs are low since In a recent attempt to compare household ' ~~~~~~~~returns from agriculture and wildlife, a survey for peasant farmers the costs of food and care . . . T was carried out in four wards in the Sebungwe are negligible: the community bears the costs rgo fteZmeiVle Mrnaoo of depletion of the open access grazing re- frthomng) the suzVey showedtatoin sources and the government bears the cost of Gokwecover 80 percen o houehatd Gokwe over 80 percent of annual household most veterinary costs with their animal health income is from cropping and 8-10 percent and dipping programs. from off-farm employment, with 5-15 percent Small domestic animals are also part of from livestock sales. If wildlife earnings had most household economies and in some of the been paid out in cash, they would have repre- more remote areas, wildlife plays an important sented 2-4 percent. In the Omay, on the other role in subsistence income. In a few isolated h areas where tsetse flies are still around, wild- had,over 70 percent was fromo life dminats, andeven mall dmesti employment, and only 17 percent was from life dominates, and even small domestic crpigLvetksaswreron3 livestock is limited. Household surveys have cropping. Livestock sales were around 3 percent, and the household revenue from shown that in the more and areas of the coun- percent, a hold reenu from wildlife would have been 4 percent if it had try, nonfarm income is very important and that been paid out in cash and not retained by the without off-farm income, government spon- woutd droff-farmlief in govessarnment pn ward for projects. The survey also showed that soe druh rleisncsayvnin Incomes in noncattle areas were half those in average rainfall years. Thus, most agricultural c attle areas derentals wr officers and farmers prefer irrigation as a obseared Si midiZambezivalley) strategy for increasing incomes and generating observed in Gutsa (mid-Zambezi valley) revenues in the marginal areas. However, the between houses with and without access to irrigation schemes are generally uneconomic, draft power (Barrett 1992) and in most com- unless the cost of establishing water sources Is munal household surveys throughout Zimba- not incl . Ebwe, access to draft power has been the most not included. Even when the water provision imotn vaibei eeriigicm costs are excluded, the rates of return are i ta n e n n normally around 10-15 percent. The higher levels. return schemes tend generally to be small Even though the C P rE proecti schees wth lmite infastrcturl inest- active in three out of the four wards surveyed schement. with limited infrastructuralinvest- and considerable wildlife still exists in these The analysis of agricultural projects seldom areas, wildlife plays a very small role in Tncludesyany of theisubitenc benefits frldom household revenue generation, particularly where these revenues are retained at the ward wildlife, since hunting is illegal. The value of level for projects. Wildlife plays a more sig- woodland resources are also seldom incorpo- nificant role in household costs (through crop rated as an opportunity cost to clearing land and livestock damage and the time spent for cultivation (Bradley and McNamara 1993). 202 VOLUME II protecting resources and people from wild- sive self-sufficiency strategies. Wildlife pro- life), but these have not been estimated. duction systems, which include high-value As a result of tsetse fly eradication, both the production of specialty outputs, may become Omay and Nenyunka, which have a relatively more financially attractive to peasant house- abundant wildlife population, are faced with holds and considerably increase both rural and land pressure resulting from potential immi- national growth (Muir 1987). grants and their cattle. At this stage of devel- Given the difficulties of comparing cattle opment, where access to land is not the limit- and wildlife in peasant farming communities, ing factor, the existing households welcome it may be better to estimate opportunity costs immigrants because they bring with them the of cattle by comparing the with and without potential of access to cattle for draft power, scenarios. This is carried out in Table 8-2 and increased populations justify more access where the assumption is that cattle and popu- to government services (roads, schools, clin- lation densities preclude the option of wildlife, ics, etc.) but bring with them concomitant increases in It is only under very particular circum- average household incomes. The subsistence stances with very poor agronomic conditions incomes are higher as a result of direct income that wildlife can substitute for cattle at the generated by livestock, increased cropping household level, unless a considerably in- incomes, and reduced losses from wildlife. creased proportion of the revenue generated Total area cultivated also increases because by wildlife can be retained by producers. there are more households farming. It is as- Draft power is the most valuable output from sumed that the increased income would be cattle, and it is possible that where wildlife evenly distributed among households. Income and cattle cannot be compatible, other forms distribution is highly skewed in rural areas, of draft power may reduce the opportunity and less than 20 percent of the population costs of wildlife versus cattle for the house- normally own adequate cattle for draft. holds. Measures to reduce wildlife predation Under current revenue distribution in com- (e.g., fencing of cropping zones) may also munal areas, wildlife is only rarely more reduce opportunity costs. Localized control profitable to the individual householder than systems, which include allowing individual subsistence agriculture. The question therefore hunting may also result in substantial in- is whether wildlife is the more economically creases in household welfare at small losses, viable option in communal areas at the na- and possibly substantial gains, to wildlife tional level. An example is given in Table 8-2 numbers. of the opportunity costs of wildlife and subsis- Despite the importance of cropping and tence agriculture using Nyaminyami District. maize in revenue generation in the nonwildlife The analysis shows that the best option from areas, maize gross margins are mostly nega- an international economic perspective is for tive when the cost of own labor and draft wildlife with improved agricultural output, but power is included in the calculations. The with no increase in population. This is because decision to grow maize is rational when the the increase in human population reduces cost of producing the maize is compared to the wildlife revenue and, as the international cost of importing from major urban centers, community benefits by 20 percent more than particularly when the risk of poor access is the nation, it has a larger stake in the wildlife added to the landed cost. The lack of cash option. resources to purchase maize also encourages The best option from a national economic home production. Therefore, it is possible that perspective is wildlife. This option assumes if markets were more effectively integrated, that population is increasing to 7,500 house- households would not need to rely on expen- holds because of the additional income gener- 8. Economic Policy and Wildlife Management in Zimbabwe 203 ated by more people farming in the area.3 The The impact of macroeconomic policies on crude assumption was that household farm the wildlife sector income doubled with improved agriculture by Foreign exchange reorganizing the current systems into a best case scenario with wildlife revenue only For most of the 1980s, overvaluation was declining by 12.5 percent. If agricultural greater than 50 percent, but with ESAP this incomes cannot be improved while retaining had fallen to 15-20 percent by the end of wildlife, the wildlife-only option is preferred 1993. The overvalued exchange rate has had a from a national social welfare perspective. significant negative effect on the wildlife This option would also be best for the house- industry. The limited access to foreign ex- holds since they would achieve the increased change had severely constrained access to incomes from better agriculture, as well as imported inputs, particularly vehicles, spare from wildlife revenue. If agricultural incomes parts, and the luxury commodities high-value cannot be improved with wildlife, then the tourism and hunting demands. The regulations best option for the farmers is to eliminate pertaining to access to foreign exchange wildlife and encourage immigrants and cattle. significantly increased transaction costs and The calculation assumes that the same per favored established industries and firms, capita agricultural productivity is achieved by constraining wildlife sector growth and effec- reducing revenues by only 12.5 percent, but tively reducing returns to CAMPFIRE. For with only half the new immigrants so that the example, in a survey of safari operators, the population is 7,500 households. If agriculture most significant constraint to growth cited was is improved, but there is no imrnigration, then the shortage of foreign exchange. wildlife revenue remains the same. In reality Recent changes under ESAP have consider- it would probably improve if the subsistence ably improved the wildlife sector's access to agricultural system had been reorganized to foreign exchange. For example, the introduc- allow for the interface. tion of the export retention scheme (ERS) in July 1990 allowed exporters the right to utilize The policy environment and its 35 percent of export earnings to import inputs impact on the future of wildlife (except for those in a "negative list"4). Al- though it favors existing operators, ERS The Zimbabwean economy is characterized quotas are now tradeable at a premium of by its highly centralized and heavily regulated between 20 percent and 50 percent, and that structure. There are many state-owned enter- has helped new entrants. There are still con- prises and parastatals. International borrowing siderable transaction costs in accessing these subsidized the expansion of the civil service rights and delays in clearing imported inputs, and expenditures for social services. In the but the situation has improved. The increased 1980s real gross national product (GNP) per retention scheme also has reduced incentives capita stagnated, and debt service problems to engage in illegal foreign currency deals and arose. This was the context in which Zimba- underinvoicing. bwe embarked on an economic structural adjustment program (ESAP) in the 1990s. 3These estimates all assume that the immigrants are 4However, some of the items on the "negative moving from overcrowded areas and that all the slack list" (imported luxuries and vehicles above a certain created by their exodus is taken up by the remaining value) are necessary for the wildlife industry, and there population. This is not likely to be true in the short- may need to be some exceptions for wildlife enter- term. prises. 204 VOLUME II Fiscal policy white-dominated industry, unless an affirma- tive action program with special credit facili- The Zimbabwe government has greatly tie an triigi. u npae expanded its bureaucracy since independence, ties and training IS put In place. and expenditure on salaries is extremely large. By contrast, the Ministry of Environment and Employment Tourism (which includes DNPWLM) received The government is very concerned about 0.68 percent of total government expenditure increasing employment and reducing under- in 1990/91. DNPWLM received only half the employment. The industrial sector cannot expenditure on Veterinary Services and only grow fast enough to absorb the rapidly in- 60 percent of that allocated to the Ministry of creasing population, and the agricultural Political Affairs. This is an indication of the resettlement program is supposed to help. low priority accorded wildlife despite its However, limited employment generation potential for growth. means that the current trend of informal settle- The budget deficit and high government ment will continue, and this population pres- expenditure have directly contributed to the sure threatens communal lands, including need for tight monetary controls. This has many of the CAMPFIRE areas. Most wildlife affected the wildlife sector at a time when it options in communal areas at the moment are appeared ready for major expansion. Contin- land intensive and absorb very few people, ued fiscal imprudence with expenditure di- relative to subsistence agriculture. It is impor- rected toward party rather than national priori- tant to develop institutions and technologies ties is a serious threat to the entire Zimbabwe which enable wildlife systems to employ or at economy, including the wildlife sector. least accommodate more people, without reducing wildlife densities. Monetary policy One area in which the government does Investmentpolicy appear to be committed to structural adjust- The Zimbabwe Investment Centre (ZIC) ment is in monetary policy, since interest rates has been established to facilitate private sector have been forced up by requiring the banks to investment. Government policy places a remain very liquid. This has been part of the premium on investment in the mining and policy to contain inflation, but unfortunately manufacturing sectors as compared to the this tight monetary policy has not been wildlife sector. matched by austere fiscal policy. The position has been made significantly worse by the The impact of sectoral policies on the drought, which has forced government to wildlife sector incur large deficits in the importation of es- Sectoral allocation sential food and raw materials. Given the lack of government progress on The viability of the wildlife industry in containing nondrought-related expenditure, Zimbabwe rests very heavily upon an effective the prognosis for the future must include the national parks structure. Capital expenditure burden of very high national debt. Monetary for the Department of National Parks and policy alone can only control inflation at the Wildlife Management in 1991 had declined 27 expense of growth, through severely high percent since independence. Recurrent expen- interest rates. These high interest rates also diture for DNPWLM has fallen 6 percent in affect the wildlife industry. This situation real terms since 1988. This is despite the excludes new entrants with limited collateral significant increase in the demand for and will therefore exacerbate the trend of a DNPWLM services. 8. Economic Policy and Wildlife Management in Zimbabwe 205 The result of inadequate financing has been subsequent distortions will have severe im- a decline in the capacity of this department to pacts on the economic efficiency of agricul- service the rapidly expanding wildlife industry ture and may lead to disincentives for wildlife- and to sustain the national parks estate. In based activities. addition to the very low budget, DNPWLM The negative attitudes of most of the agri- has had to service an expanding mandate and cultural bureaucracy to wildlife derives from has had to face significant increases in poach- uncertainty over its actual economic value, as ing. Time, skills, financial, and other re- well as lack of technical and business exper- sources have been directed away from tradi- tise in wildlife enterprises. This has more tional concentration on park management and serious consequences in the communal areas research to service all these new demands. and resettlement schemes, since land use The decline in funding has resulted in deterio- planners hesitate to allocate land to wildlife. ration of roads, boreholes, buildings, plant, Similarly, those responsible for resettlement vehicles, and equipment. It has also affected find wildlife to be land intensive, and very research, extension and interpretive services, restrictive to their mandate to resettle the park management, and law enforcement, thus maximum number of people from over- threatening wildlife conservation, habitat crowded communal farming areas. protection, and the economic productivity of the wildlife sector as a whole. Marketing and price policies Land In the large-scale sector, cattle are the major competitor with wildlife for land. Beef The current land policy aims to reallocate production has been heavily supported by the half the land in the large-scale freehold sector government since the 1930s. This included to the small-scale sector, but it will retain the large marketing and processing investment in same basic dualistic structure. There is consid- beef by the state. Producers or consumers (and erable concern that farms with wildlife enter- at times both) have been directly and indi- prises will be targeted for resettlement. These rectly subsidized by government marketing fears arise from the belief that wildlife is and pricing policies for beef over the decades. considered a luxury and not a productive land The distortions created by the implicit and use system by some bureaucrats and politi- explicit support for beef have particularly cians. There is, therefore, an urgent need to affected the development of the goat industry disseminate infornation on the productivity of and have probably had similar effects on game wildlife. meat. Since independence the Cold Storage The country is to introduce a land tax that Commission (CSC) has received subsidies of will encourage large-scale farmers to utilize Z$ 40 million per annum. Prior to 1985 both those areas of their properties currently producers and consumers were subsidized, but underutilized. On areas that cannot be used for subsequently producer prices were reduced, agriculture, there may be increasing invest- and only consumers have been subsidized by ments in wildlife and recreational tourism, and the state, and indirectly by farmers. this would increase the growth of the wildlife sector. Regulations that discourage subdivi- Veterinarv and health controls sion will reinforce this trend, since the farmers The erection of fences and the eradication will not be able to sell the land they are not of buffalo from large-scale ranches have had using for agriculture. There is some uncer- tainy asto he ntur of he lnd ax, ith negative effects on both the economy and the tanmty as to the nature of the land tax,it viability of wildlife enterprises on ranches. It somedinferene thit bs to afect is unlikely that the value of the buffalo eradi- production systems. If this is the case, the 206 VOLUME II cated in order to allow for beef exports to still prohibited from hunting any wild animals, Europe can ever be recouped, despite the very even in districts that have been granted appro- high prices obtained from privileged access to priate authority. The Zimbabwean economy is their controlled markets. No analysis has been very highly regulated, and these national, carried out on the economics of the fences nor municipal, and industry restrictions often act on their impact on wild animal migrations. as barriers to entry and have restricted the The continued effect of keeping buffalo out of growth of wildlife and tourism ventures. cattle areas affects wildlife profitability, since buffalo add significantly to the value of hunt- Conclusions ing on ranches (Child 1988). Movement Wildlife in the commercial ranch sector is controls implemented to satisfy EC require- often more financially profitable than cattle, ments considerably increase the cost of live particularly in the more ard parts of Zimba- animal sales by requiring long quarantine bwe. An improved macroeconomic environ- periods. There are also a number of municipal ment and a less regulated economy will con- and national regulations that restrict the free tbute to its viability and continued growth. movement of game meat sales and require trI various inspectionsf This will also benefit cattle enterprises greatly various inspections, and may even tilt the balance in its favor vis- a-vis wildlife. In the remote, semiarid commu- External trade controls lnal sector, the CAMPFIRE program appears Exports of wildlife products and of live to be the most economic option from a social animals are subject to various permits, li- welfare perspective, but it is seldom the ratio- censes, and restrictions. These restrictions are nal choice for households living in these normally either to fulfill the requirements of areas. Thus, more effective mechanisms and international conventions (e.g., CITES) or are institutions need to be developed so that the an attempt to build up the local industry. The gainers compensate the losers. restrictions on the sale of live animals are Finally, the direct benefits to the world controversial and tend to result in industry community from retaining wildlife can be concentration. The export of game meat and considerable. These would include greater other wildlife products are severely restricted species diversity, greater tree coverage, and by the importing countries. the bequest and existence values of the ani- mals. If these can be appropriately valued, Sale and utilization controls then in certain areas the international commu- nity clearly benefits from land allocated to Although mayduolde or aprrate f u- wildlife. Such global benefits need to be thortiesmay tilie wid anmalsfoun on considered through institutions such as the their land, the method and sale of various g . . services and products are controlled (Booth Global Environment Facility (GEF). 1992). In the communal areas, residents are 8. Economic Policy and Wildlife Management in Zimbabwe 207 Table 8-1: Adjusted Contribution of Wildlife to the Economy, 1990/91 Total Earnings Forex Earnings Sector (Z$ million) (Z$ million) 1. Tourism 731.0 531.0 2. Hunting 81.5 81.1 3. Wildlife management services and live animals sales 6.2 negl. 4. Game products (meat, hides/skins, horns) 1.0 negl. 5. Crocodile production 26.6 26.6 6. Ostrich production 35.4 35.4 TOTAL 881.7 674.1 Note: Based on Equilibrium Exchange Rate. All the figures must be considered very rough estimates due to severe data limnitations. Source: Jansen (1992a). 208 VOLUME II Table 8-2: Comparative Analysis of Wildlife and Subsistence Agriculture: Total Revenue Estimates Using Nyaminyami District, Zimbabwe Part 1: Wildlife Revenue Estimates and Allocation (U.S. dollars per annum 1992) Households Wildlife International National District (HHs) All HHs HHs Direct Revenues Hunting 1,034,066 827,252 212,000 Tourism 781,250 625,000 7,200 70667 14 35 Total 1,815,316 1,452,252 219,200 Airfares 450,000 Outfitting 200,000 51,703 Taxidermy 51,703 10,000 6,000 Curios 10,000 20,000 -- 10,000 17 43 Meat 20,000 -- 219,200 -- Ivory -- 1,533,956 86,667 Total 2,547,019 Environ. protection Not quanti- Not quanti- Not quanti- Genetic resources fied, but signifi- fied, but fied, but Bequest cantly signifi- signifi- Existence positive antly cantly TOTAL wildlife positive positive (continued) Part 2: Alternative Land Use Scenarios for Nyaminyami Households All Wildlife International National District (HHs) HHs HHs Subsistence agriculture with wildlife 5,000 HHs agric in- come US$ 262,000 262,000 262,000 52 52 TOTAL incl. wildlife income US$ 2,809,019 1,795,956 219,200 348,667 70 96 Subsistence agriculture with no wildlife 10,000 HHs US$ 1,048,000 1,048,000 1,048,000 105 105 5,000 HHs US$ 524,000 524,000 524,000 105 105 Retain wildlife with reorganized and improved agriculture 7,500 HHs US$ 3,014,641 2,128,211 219,200 872,667 122 148 5,000 HHs US$ 3,071,019 2,057,956 219,200 610,667 122 148 International refers to revenue generated for the world population. National refers to the revenue generated which remains within Zimbabwe. District refers to the proportion paid to district council. Households refers to proportion of income accruing to producers. All households refers to per household distribution based on 5,000 households, in the two lines and as indicated for the remainder. Wildlife households is based on 2,000 households living with wildlife. Refer to the notes that follow for detailed assumptions used in Table 8-2. 8. Economic Policy and Wildlife Management in Zimbabwe 209 Notes to Tables 8-1 and 8-2 1. Hunting revenue. The estimated total (as for 1991) value from the hunting sold based on pro- 8. Ward allocation to individual house- jected international hunting prices for 1993 holds. Assumes 5000 households in with 5 percent deducted for international Nyaminyami. Assumes that the ward divi- inflation. International safari prices have not dends are distributed evenly between the 5000 in fact risen, and it is possible that the 1992 households (whether as cash or projects) value should be the same as that for 1993, but 9. Ward allocation if only to wildlife house- a conservative estimate was preferred. Precise holds. Assumes that the revenue is shared 1992 data were not easily available. only by those living in areas with substantial 2. Hunting revenue retained in Zimbabwe. wildlife resources (Omay). The number of A deduction of 20 percent to cover external households is estimated to be 2000. marketing costs, convention, wholesaler and 10. Value of airfares of incoming hunters. It agents commissions and fees. The remainder is guesstimated that there are 145 hunters and is what is paid into Zimbabwe. companions based on the number of hunts. No 3. Hunting revenue retained by the district airfare value is given for those coming in for council. The trophy and concession fees paid tourism on the assumption that they would fly to the district council (Z$ 1,060,000 converted to Zimbabwe, even if there was no to US$ at 5:1). Nyaminyami wildlife. 4. Tourist revenue generated by 11. Value of guns, etc. Assumes that each Nyaminyami wildlife. An estimate of the safari party spends $1,400 in their home revenue earned by Bumi Hills Safari Lodge country on equipment, ammunition and outfit- and Tiger Bay because of the wildlife in ting (safari outfits etc). Omay. If there was no wildlife it is estimated 12. Value of taxidermy. Each client spends 5 that Bumi would loose 50 percent of its cus- percent on mounting trophies. tom and Tiger Bay 25 percent (it is closer to 13. Value of curio sales. An estimate based the national park and caters more heavily to on Binga sales of Z$ 80,000 - estimated that fishermen). Both were estimated at 5000 bed the handicraft center at Bumi will earn nights, Bumi at US$250 per night and Tiger Z$ 50,000. Bay at US$125 per night, full board and tours. 14. Value of curio sales to households. The 5. Tourist revenue retained in Zimbabwe producers retain 60 percent. Although aver- A deduction of 20 percent for external market- aged over all Omay residents in fact it is likely ing agents fees etc - this is probably too high to be much more concentrated income and as much of the marketing costs and fees may cannot be easily represented in the individual be retained locally. household dividends. 6. Tourism revenue paid to district council. 15. Value of meat. Includes meat, hides, etc, Tourism numbers for 1992 were calculated as from cropping and problem animal control - lower than 1991 (particularly for Bumi) be- assumes less available but at higher price so cause of the effect of drought on international remains same as 1991. National social price is visitors. Assuming that the per visitor fee to doubled as meat was sold at subsidized price. be paid to the district council in 1992 is Opportunity cost value not adjusted at house- higher, the revenue to the DC, is assumed to hold level because for those individuals the be the same as in 1991 (Z$ 36,000). opportunity cost is hunting time and illegality 7. Hunting and tourism revenue allocated risk which may even be less than the money for wards. Assumes that the DC pays 33 saved by access to cheap meat. percent of its total earnings as ward dividends 16. Current value of subsistence agriculture 210 VOLUME I in Nyaminyami. Based on Z$ 262 per house- 18. Household value of subsistence agricul- hold per annum currently generated by crop, ture with cattle and no wildlife. This as- livestock and beer sales and grain retentions sumes that the increased incomes are evenly (average crop returns for Negande Ward in distributed when in fact they are usually Murindagomo 1992). Most household income highly concentrated with those who own is from off-farm work and drought relief, cattle. The current residents will benefit from Z$ 262 by 5,000 households converted to U.S. increased access to cattle, more people in the dollars. area to lobby for services and to sell beer to; 17. Value of subsistence agriculture with but it is unlikely that they will in fact double cattle and no wildlife. Using survey results of their current incomes. vidcos with and without cattle (Zhomba and 19. Value of wildlife and subsistence with Ntamo, Nenyunka Ward, Gokwe in project to reorganize agriculture. Some Murindagomo 1992) average farm income wildlife experts maintain that it is not cur- doubles. This would be as a result of increased rently desirable to incorporate cattle into access to draft power, immigrants with greater subsistence agriculture systems in wildlife access to cash or credit for input purchases areas and furthermore that any increases in and reduced wildlife depredation. It is as- population would not be sustainable. How- sumed that there are no agronomic differences ever, it seems that reorganization including between the vidcos which may have caused possibly greater separation of people and the differential. It is further assumed that with wildlife, access to alternative draft or live- the elimination of tsetsefly and increased stock holding pens etc could be used to try to demand for land, immigrants move in and the enable increased agricultural production population is doubled resulting in increased alongside continued wildlife enterprises - this habitat destruction and poaching which reduce has been achieved in the commercial ranch wildlife to insignificance. The calculation is sector. It is more difficult in the communal double Z$ 262 multiplied by 10,000 house- sector but may be possible. holds converted to U.S. dollars. 8. Economic Policy and Wildlife Management in Zimbabwe 211 Figure 8-1: Comparison of Projected Household Income 500 - - --------.! 450 . . 400 350 300 UStatus Quo 250 .UOptimistic 200 . . :0'. . . ,: : . . 1500 EttIU 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Nysminyami 600 . .. . ..... .. ..................... . ........... ............... 500 E Status Qu ' Optimistic 400 300 200 . 100 0 19S9 1990 1991 1992 1993 1994 1995 1996 1997 1998 Guruve 1200 Statu. Quo 1000 *o timistic 800 600 400 200 . 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Msbeniye Ward 212 VOLUME II Annex 8-A: Relative Environmental Impacts of Wildlife' Introduction The Evidence on Herbivore Impacts on Zimbabwean Concerns about overgrazing and rangeland Ran emand a degradation have played a central role in the Rangelands shaping of the national livestock policy, giv- National surveys ing rise to schemes of limiting stock numbers and in some cases of actually destocking. For Zimbabwe as a whole, over the ten- Attention has also been drawn to the impacts year period from 1963 to 1973, closed wood- especially of elephants in the destruction of land and open woodland declined in area by 4 woodlands. Degradation is seen as manifest- percent and 16 percent, respectively, while the ing itself where a plant community, which is area of sparse woodland and cultivated land dominated by perennial grasses changes, to a increased by 12 percent and 8 percent, respec- community that includes a greater proportion tively (Whitlow 1980). The main decreases in of woody species, annual grasses, and bare woody vegetation were recorded in areas of ground. Thus, rainfall that was used efficiently high to moderate population densities, particu- in the production of a relatively stable supply larly within the communal lands. This was of fodder is now wasted, less fodder is pro- ascribed primarily to the expansion of culti- duced, and the fodder supply is more variable vated lands, and to a lesser extent to the col- from year to year. lection of firewood and building materials. Initially utilization results in changes in Attention was also drawn to the rapid rate of the herbaceous species composition of the destruction of woodlands on some of the state veld, especially from perennial grasses to a land in the west and northwest of Zimbabwe, combination of annual grasses and less palat- which was ascribed mainly to the buildup of able perennial grasses with reduced vigor. high elephant densities within these areas. Giround cover and litter cover decrease, thus leading to increased soil compaction and Commercial livestock production systems crusting, and thus enhanced runoff and ero- Stocking rate has repeatedly been shown sion. Rainfall infiltration is reduced substan- to have a marked effect on both rangeland tially and, because the density of grass roots composition and animal productivity near the soil surface is lower, a greater propor- (Clatworthy 1989). There is a consistent trend tion of the water that does enter the soil be- for perennial grass cover to decline with an comes available to deeper-rooted woody increasing stocking rate. For animal perfor- species. This gives them a competitive advan- mance, both wet season gains and dry season tage, which they are able to maintain at the losses are affected by stocking rate, with the expense of grasses, so that the supply of former decreasing and the latter increasing as fodder is permanently reduced. As bush en- the stocking rate is increased. However, for a croachment and the above induced drought given stocking rate, the wide range of grazing conditions continue to develop, sheet and procedures examined have had almost no gully erosion become more widespread, and differential effects on botanical composition previously permanent streams become more nor on the condition of the veld. intermittent. Kennan (1969), O'Connor (1985), and Clatworthy (1989) all emphasize that the 5This annex is based exclusively on the background paper prepared by Robert Cunliffe. 8. Econon;ic Policy and Wildlife Management in Zimbabwe 213 impact of livestock on veld condition is de- the communal lands. Stocking rates in the pendent on both rainfall and soil conditions. A communal lands, frequently of the order of number of studies have demonstrated that 2-3 hectares per livestock unit, are typically long-term rainfall variability has an overriding higher than those found on other types of land. effect on herbaceous compositional trends, Livestock, particularly cattle, form an integral independent of grazing regimes, particularly in part of the agro-pastoral farming system, the drier rangelands where rainfall is more particularly through the provision of draft variable and primary production is strongly power and manure, which is widely used to limited by soil moisture availability. In maintain or improve soil fertility. Recent moister savannas, annual variability in rainfall studies, such as those by Swift et al (1989) on is lower, and the relationship between vegeta- nitrogen cycling in communal land farming tion production and rainfall is less marked. systems and Barrett (1992) on the economic The influence of grazing varies, with the most role of cattle in communal land farming sys- pronounced effects being observed in low tems, indicate that the majority of households rainfall regions. do not own sufficient livestock to satisfy their needs for draft power and manure. Wild herbivore systems Despite the widespread concern about the Despite the diversity of large herbivores, deterioration of communal land rangelands, which is a characteristic of African savanna remarkably little research has been directed at communities, only a limited number of spe this problem. This has been highlighted by a cies have been reported as having dramatic number of recent studies that question the effects on the environment. Analysis shows official view of the extent and significance of that communities tend to be dominated by a degradation in these areas (Abel and Blaikie > . . ~~~~~~~~~1989; Barrett, Brinn, and Timberlake 1991; few large species, such as elephant, hippopot- Sandford 1982; Scoones 1990; Warren and amus, buffalo, zebra, and wildebeest anew 1988). (Cumming 1982). It is these species, particu- Agnew 1988). larly the elephant, which have been reported Scoones (1990) examined ldvestock pro- as having a dramatic effect on the environ- duction records over several decades for seven ment. heavily stocked communal lands in southern Craig (1992) and Martin (1992) have Zimbabwe. Government officials, in 1944, calculated that elephant densities need to be considered the stocking rates of six of the held below about 0.5 animals per square seven communal lands studied by Scoones to kilometer, in order to maintain the existing be in excess of the recommended carrying woodland canopy cover intact. This is far capacities for these areas. Yet for each com- lower than the densities currently occurring in munal land, the 1986 cattle population density was considerably in excess of its 1944 stock- many of the national parks and safari areas, y ing rate. Despite large fluctuations in response 0.25 to 2.1 a ere kilomt r . to rainfall, cattle populations appeared to have been successfully maintained over the last Communal land agro-pastoral systems twenty-five years for all communal lands. There were obvious signs of soil erosion on Since early in the colonial period, over- the rangelands, and yet there was no evidence stocking and degradation of communal range- that cattle populations were declining, and lands have caused concern. For much of this neither was their potential productivity being century, and continuing to the present, govern- detrimentally impacted. Scoones (1990) sug- ment officials have sought ways in which to gested that the impact of erosion on secondary control or even reduce livestock numbers in production is sufficiently low as not to be felt 214 VOLUMEII on a relatively short time scale, or else that the A comparison of separate game and cattle sites where erosion is occurring might not enterprises, which have been run since 1960 contribute significantly to the overall livestock on adjoining, ecologically very similar areas, production, or both. under the same overall management was The Save study (Campbell, Du Toit, and undertaken on Buffalo Range (Child 1988). Attwell 1989), however, perceived the envi- Rangeland condition has been assessed in ronment as being degraded through deforesta- detail in 1973, in 1986 (Child 1988), and in tion, overgrazing, loss of soil, loss of soil 1990, although the results of this latest survey fertility, siltation of rivers and increased have not yet been made available. runoff. It concluded that there was a severe In 1973 the cattle section was ecologically imbalance between the human population and in better condition than the game section at the resource base, such that the environment this time. Thirteen years later there were was less able to satisfy the needs of current definite indications that range condition on the and future human populations. game section had improved, whereas it had deteriorated on the cattle section. The area Comparative studies under cattle now exhibited twice as much erosion as that under game, and although there Kelly and Walker (1976) undertook a comparative study of four sites under different was an increase in soil capping on both the land tenure, which were apparently similar in cattle and game sections, the increase on the land tenurespet, wichp werthe aypparentl itsimr iformer was threefold that of the latter. These every respect, except for the type and intensity chneinrgeodtonanbrltd of grazing and browsing by large herbivores, dirchanges in range condition can be related Fiften yars aterDe Jger 1988 repated directly to trends in secondary production. The Ffeasureenty s later veJageration (1 ragepedi levels of meat harvested fluctuated, but over a -measurements of vegetation and range condi- twn sx a ro so h w tm tion. Kelly and Walker's study concluded that ysy toy the intensively utilized communal land site yielded very similar amounts of meat. Cattle was in fact degraded in comparison to the productivity, however, was reduced by declin- other three sites, in that the perennial grasses ing calving rates. had largely been replaced by annual grasses; Child (1988) relates declining cattle pro- herbaceous production was more variable ductivity to the overstocking. The stocking between years. rate of wildlife fluctuated considerably, in- between years. De Jager's follow-up study failed to reveal creasing rapidly during the high rainfall years any firm evidence of directional trends that of the 1970s but declining massively during could be related to differences in the intensity the drought years from 1982-84. After 1984 of utilization by herbivores. The herbaceous populations increased through natural growth at a slower rate than on the cattle section, vegetation of the communal land site was still wi was rste with cattle fromelse domiatedby anualgrases,as opose to which was restocked with cattle from else- dominated by annual grasses, as opposed to whr.Anopitntmeashetckg perennials on the other three sites. Basal cover rate of wild herbivore grazers on the game of grasses was also lowest for the communal seon everbexce thatocl on the land site, as was the litter cover and the rate of section At the time of the 18 me infiltration. There were indications that the communal land soils were now of lower surements, the cattle stocking rate was some fertility than those of the other three sites, but 1.5 times that of game. none of these differences were significant. This study suggests that the meat produc- noneofy thesetation hadifences w significan. tion potential of the cattle and game enter- Woodensity vegetandti hadtchangedey cmons bly prises is similar and that rangeland impact is ..in density aendte togeneityfamongr plt largely a function of the stocking rate of twithn sites. tended to mask differences be- herbivores. Child (1988) makes the important tween sites. 8. Economic Policy and Wildlife Management in Zimbabwe 215 point that the generation of revenue from the different kind, associated with the unique cattle enterprise is directly related to the sec- effects of different herbivore species ondary production of beef, whereas for the (Cumming 1982). The comparison of Child wildlife enterprise, the major form of revenue (1988) between wildlife and a commercial generation is through safari hunting opera- cattle system suggests that the environmental tions, thus enabling more conservative stock- impacts of wildlife and cattle are determined ing rates to be maintained. largely by the stocking rate, rather than by the different types of herbivores, and this is sup- The Major Factors Influencing ported by the literature reviewed by Environmental Impacts Clatworthy (1989). The research reviewed above further suggests that, particularly in There is no single biological optimal semiarid environments, vegetation changes are carrying capacity that can be defined inde- unreliable indicators of rangeland degradation. pendently of the different management objec- Rates of soil loss and changes in soil chemis- tives associated with different forms of animal try, and physical properties may be more exploitation. Thus, for semiarid rangelands, reliable indicators. which are subjected to erratic rainfall and Given that the environmental impacts of characterized by large fluctuations in plant different herbivore communities are deter- species composition, cover, and biomass, the mined essentially by the stocking rate rather problem becomes one of distinguishing be- than the types of herbivores involved, some tween drought-induced fluctuations and per- consideration of the stocking rates required to manent changes in vegetation states. It is meet the specific objectives of ditferent ani- doubtful that our current knowledge of sa- mal production systems, becomes important. vanna systems allows these distinctions to be Data from Buffalo Range illustrate that wild- made at present with any degree of confi- life can be profitably stocked at a significantly dence. lower rate than that requiredfor cattle, due to Comparisons of herbivore community the high values associated with hunting safa- structure between wildlife and domestic live- ris. Studies of communal land farming sys- stock systems, in terms of the biomass contri- tems show that the current high livestock butions of the different classes of feeders, lead numbers consistently fall far short of the to the important conclusion that the impact of populations required to satisfy the multiple wild, as opposed to domestic, herbivores on needs of farmers. Thus, the communal lands the environment, is more likely to be a ques- are likely to continue to be heavily stocked. tion of degree rather than of fundamentally 216 VOLUMEII Bibliography Abel, N. 0. J., and P. M. Blaikie. 1989. "Land Conservation Project Task 2 The Role of Degradation, Stocking Rates, and Conser- Wildlife in the Economy. Reports pre- vation Policies in the Communal Range- pared for Dept. National Parks and Wild- lands of Botswana and Zimbabwe." Land life Management, Zimbabwe. Degradation and Rehabilitation 1: 1-23. Booth, V. 1992. "Wildlife Utilization, Man- Agritex. 1990. "Feasibility and Design Report agement and Planning Outside the Parks of Nabusenga Small-scale Irrigation and Wildlife Estate." In Price Waterhouse Scheme." Irrigation Division, MLARR, and Env. Resources, Wildlife Management Zimbabwe Government, June. and Environmental Conservation Project Anderson, G. D., and B. H. Walker. 1974. Background Papers prepared for "Vegetation Composition and Elephant DNPWLM funded by Japanese Grant Aid, Damage in the Sengwa Wildlife Research coordinated by World Bank, AF6IN. Area, Rhodesia." Journal of the Southern Bradley, P., and K. McNamara (eds.). 1993. African Wildlife Management Association, "Living with Trees, Policies for Forestry 4:1-14. Management in Zimbabwe." WB Techni- Barrett. J. C. 1992. "The Economic Role of cal Paper Number 210. Cattle in Communal Farming Systems in Campbell, B. M., R. F. Du Toit, and C. A. M. Zimbabwe." Pastoral Development Net- Attwell. 1989. "The Save Study: Relation- work Paper 32b. London: ODI. ships Between the Environment and Basic Barrett, J. C., P. Brinn, and J. Timberlake. Needs Satisfaction in the Save Catchment, 1991. Tsetse Control, Agropastoralism Zimbabwe." Supplement to Zambezia, and Land Degradation: A Case Study in University of Zimbabwe, pp 1-119. Chiswiti Communal Land. Final Report. Child, Brian. 1988. "The Role of Wildlife Unpublished report of the Tsetse and Utilization in the Sustainable Economic Trypanosomiasis Control Branch, Depart- Development of Semi-arid Rangelands in ment of Veterinary Services, Harare. Zimbabwe." Unpublished D.Phil disserta- Behnke, R. H., and I. Scoones. 1992. "Re- tion, School of Forestry, Oxford Univer- thinking Range Ecology: Implications for sity. Rangeland Management in Africa." Child, B., and J. Peterson, Jr. 1991. CAMP- Dryland Networks Programme Paper No. FIRE in Rural Development: The 33, Overseas Development Institute, Lon- Beitbridge Experience. DNPWLM and don. CASS Joint Working Paper 1/91. Behr, J., and J. A. Groenewald. 1990. "Com- Child, Graham. Forthcoming. Wildlife in mercial Game Utilization on South Afri- Zimbabwe, draft book in preparation. can Farms." Agrekon 29(1):55-58. Childes, S. L. 1984. The Population Dynamics Benson, D. E. 1986. "Game Farming Survey." of Some Woody Species in the Kalahari Farmer's Weekly, April-May. Sand Vegetation of Hwange National Berry, M. P. S. 1986. "A Comparison of Dif- Park. M.Sc. thesis, University of ferent Wildlife Production Enterprises in Witwatersrand. the Northern Cape Province, South Af- Chimedza, R. 1992. "Women, Household rica." S. Afr. Tudskr. Natuurnay 16(4): Food Security and Wildlife Resources." 124-28. Draft working paper, Dept. Agric. Bond, Ivan. 1992. "Tourism, Hunting and Econ.& Ext., University of Zimbabwe. Management Services." In Price Water- Clatworthy, J. N. 1989. A review of rangeland house and Environmental Resources Ltd. utilization trials in Zimbabwe, 1970 to Wildlife Management and Environmental 1985. In A. R. Maclaurin and B.V. 8. Economic Policy and Wildlife Management in Zimbabwe 217 Maasdorp (eds), Rangeland Potential in prepared for the International Develop- the SADCC Region, Proceedings of a ment Research Center, Nairobi. Regional Workshop, Bulawayo, 1-5 June, Dassmann, Raymond F., and Archie S. Moss- 1987, Ministry of Lands, Agriculture and man. 1961. "Commercial Use of Game Rural Resettlement, Harare. Animals on a Rhodesian Ranch." Wildlife Collinson, R. F. H. 1979. "Production Eco- 3, September/December. nomic of Impala." In M. A. Abbot (ed.), De Jager, P. 1988. Environmental Degrada- Proceedings of a Symposium on Been and tion in Communal Land. M.Sc. thesis, Game Management, pp 90-103. University of Zimbabwe. Pietermartizburg: Cedara Press. Dublin, H. T., A. R. E. Sinclair, and J. Colvin, I. S. 1984. An Enquiry into Game McGlade. 1990. Elephants and Fire as Farming in the Cape Province. M.Sc. Causes of Multiple Stable States in the Thesis, University of Cape Town. Serengeti-Mara Woodlands Journal of Conroy, A. M., and I. Gaigher. 1982. "Veni- Animal Ecology 59:1147-64. son, Aquaculture and Ostrich Meat Produc- Dunham, K. M. 1990. Biomass Dynamics of tion." South Africa of Animal Science Herbaceous Vegetation in Zambezi 12:219-22. Riverine Woodlands. African Journal of Conybaere, A. M. 1991. Elephant Occupancy Ecology 28:200-212. and Vegetation Change in Relation to Ellis, J. E., and D. M. Swift. 1988. "Stability Artificial Water Points in a Kalahari Sand of African Pastoral Ecosystems: Alternate Area of Hwange National Park. Ph.D. Paradigms and Implications for Develop- thesis, University of Zimbabwe. ment." Journal of Range Management 41: Coulson, I. C. 1992. Elephants and Vegetation 450-59. in the Sengwa Wildlife Research Area. In Elwell, H. A., and M. A. Stocking. 1974. R. B. Martin, G. C. Craig, V. R. Booth, Rainfall Parameters and a Cover Model to and A. M. G. Conybeare (eds), Elephant Predict Runoff and Soil Loss from Graz- Management in Zimbabwe, Second Edi- ing Trials in the Rhodesian Sandveld. tion, pp. 55-62, Department of National Proceedings of the Grassland Society of Parks and Wildlife Management, Harare. Southern Africa 9:157-64. Craig, G. C. 1992. A Simple Model of . 1982. "Developing a Simple yet Tree/Elephant Equilibrium. In R. B. Mar- Practical Method of Soil-Loss Estimation. tin, G.C. Craig, V. R. Booth, and A. M. G. Tropical Agriculture (Trinidad) 59:43-48. Conybeare (eds), Elephant Management in Epstein, H. 1971. The Origins of the Domestic Zimbabwe, Second Edition, Department of Animals of Africa, Volumes I and II. New National Parks and Wildlife Management, York: Africana Publishing Corporation, Harare, pp. 81-86. 573 pp. and 719 pp. Cumming, D. H. M. 1982. The Influence of Gambiza, J. 1987. Some Effects of Different Large Herbivores on Savanna Structure in Stocking Intensities on the Physical and Africa. In B. J. Huntley and B. W. Walker Chemical Properties of the Soil in a Mar- (eds.), Ecological Studies, Vol 42: Ecol- ginal Rainfall Area of Southern Zimba- ogy of Tropical Savannas. Berlin: bwe. M.Sc. thesis in Tropical Resource Springer-Verlag. Ecology, University of Zimbabwe. Cumming, D. H. M., and I. Bond. 1991. Ani- Guy, P. R. 1981. "Changes in the Biomass mal Production in Southern Africa: Pres- and Productivity of Woodlands in the ent Practice and Opportunities for Peas- Sengwa Wildlife Research Area, Zimba- ant Families in Arid Lands. A report bwe." Journal of Applied Ecology 18: 507-19. 218 VOLUME II Hawkes, R. K. 1991. Crop and Livestock Kaufman, 1. 1992. "Development of Wildlife Losses to Wild Animals in the Bulilima Tourism in Zimbabwe." In Price Water- Mangwe Natural Resources Management house and Environmental Resources. Project Area. CASS/MAT Working Paper Kelly, R. D., and B. W. Walker. 1976. "The Series 1/91. Effects of Different Forms of Land Use on Hitchcock, P., and F. Nangati. 1992. "Interim the Ecology of a Semiarid Region in Assessment of Zimbabwe Natural Re- Southeastern Rhodesia." Journal of Ecol- sources Management Project." Draft ogy 64:553-76. report, Price Waterhouse, Zimbabwe. Kennan, T. C. D. 1969. "A Review of Re- Ivy, P. 1969. "Veld Condition Assessments." search into the Cattle/Grass Relationship In Proceedings of the Veld Management in Rhodesia." Proceedings of the Veld Conference, Bulawayo, May 1969, Management Conference, Bulawayo, 105-112, Department of Conservation and 27-31 May, pp 5-26, Government Printer, Extension, Salisbury. Salisbury, Rhodesia. Jansen, D. 1990. "Sustainable Wildlife Utili- Kerr, M. A. 1972. Annual Report: Rhodesia. zation in the Zambezi Valley of Zimba- Department of National Parks and Wild- bwe: Economic, Ecological and Political life Management. Tradeoffs." WWF Multispecies Animal LaGrange, M. 1988. "Innovative Approaches Production Systems Project, Paper No. 10. in the Control of Quelea in Zimbabwe." In . 1992a. "The Role of Wildlife in the R. Marsh (ed.), Proc. 13th Vert. Pest Economy." In Price Waterhouse and Envi- Conf Univ. of California, Davis. ronmental Resources. Marks, S. A. 1973. "Prey Selection and An- .1992b. "The Economics of Campfire: nual Harvest of Game in a Rural Zambian Lessons to Date" Background report pre- Community." East African Wildlife Jour- pared for AFTEN, World Bank, Novem- nal 11:113-28. ber. Martin, R. B. 1986. "Communal Areas Man- Jansen, D., I. Bond, and B. Child. 1992. Cat- agement Programme for Indigenous Re- tle, Wildlife, Both or Neither: Results of a sources (CAMPFIRE)." Working doc. Financial and Economic Survey of Com- 1/86, Dept National Parks and Wildlife mercial Rancches in Southern Zimbabwe Management, Harare. WWF, Multispecies Animal Systems . 1989. "The Status of Projects Involv- Project Paper No. 27, Harare. ing Wildlife in Zimbabwe." DNPWM, Jansen, D., and K. Muir. 1991. "Trade and Harare. Exchange Rate Policy and Agriculture." Martin, R. B. 1992. Relationship between Paper presented at Conference on Zimba- Elephant and Canopy Tree Cover. In R. B. bwe's Agricultural Production Revolution, Martin, G. C. Craig, V. R., Booth, and Victoria Falls, Zimbabwe. A. M. G. Conybeare, (eds), Elephant Jansen, D., and A. Rukovo. 1992. Agriculture Management in Zimbabwe, Second Edi- and the Policy Environment in Zambia tion, pages 77-80, Department of National and Zimbabwe: Political Dreams and Parks and Wildlife Management, Harare. Policy Nightmares. OECD Technical Martin, R., G. Craig, and V. Booth. 1989. Paper No. 74. Elephant Management in Zimbabwe, Johnstone, P. A. 1973. "Evaluation of a Rho- DNPWM, Harare. desia Game Ranch." Journal of Southern McDowell, R. E., D. G. Sisler, E. C. African Wildlife Management Association Schermerhorn, J. D. Reed, and R. P. 5(l):43-51. Bauer. 1983. Game or Cattle for Meat 8. Economic Policy and Wildlife Management in Zimbabwe 219 Production on Kenya Rangelands? Cor- Murphree, M. 1988. "Decentralizing the Pro- nell University, 77 pp. prietorship of Wildlife Resources in Monke, E. A., and Scott Pearson. 1989. The Zimbabwe's Communal Lands." Paper Policy Analysis Matrix for Agricultural presented to African Studies Association, Development. Ithaca: Cornell University Cambridge. Press. -. 1991. Communities as Institutions ftr Mossman, S. L., and A. S. Mossman. 1976. Resource Management. Center for Ap- "Wildlife Utilization and Game Ranch- plied Social Sciences, Univ. of Zimba- ing." IUCN Occasional Paper No. 17. bwe. Muir, Kay (ed.). 1983. "Agricultural Market- O'Connor, T. G. 1985. "A Synthesis of Field ing In Zimbabwe." Working Paper 1/83, Experiments Concerning the Grass Layer Dept. Land Management, University of in the Savanna Regions of Southern Af- Zimbabwe. rica. South African National Scientific . 1987. "Marketing Wildlife Products Programs." Report No 114, Foundation and Services." In IGF, Paris, Proceedings for Research and Development, Council of International Symposium on Wildlife for Scientific and Industrial Research, Management in Sub-Saharan Africa Octo- Pretoria. ber, Harare. Pangeti, G. 1989. "Administration and Inter- . 1989. "The Potential Role of Indige- agency Co-ordination of Wildlife Manage- nous Resources in the Economic Develop- ment and the Effects of Legislation in ment of Arid Environments in Sub-Saha- Implementation." Wildlife Resource Mani- ran Africa: The Case of Wildlife Utiliza- agement workshop, Hwange. tion in Zimbabwe." Society and Natural Peterson, J. 1992. A Proto-Campfire Initiative Resources 2.4:307-18. in Mahenye Ward, Chipinge District Murindagomo, F. 1988. "Preliminary Investi- CASS Occasional Paper 3/1992, Center gations into Wildlife Utilization and Land for Applied Social Sciences, Univ. of Use in Angwa, Mid-Zambezi Valley." Zimbabwe. Unpub. M.Phil thesis, Dept. Agric. Econ. Richardson, F. D. 1983. Short and Long-term & Ext., Univ. of Zimbabwe. Influences of Under-nutrition on Range . 1992. "Wildlife Management in Cattle Production. Zimbabwe Agricul- Zimbabwe: The CAMPFIRE tural Journal 80:175-82. Programme." UNASYLVA (International) Sandford, S. 1982. "Livestock in the Commu- 43(168):20-26. nal Areas of Zimbabwe." Report prepared . Forthcoming. "A Comparative Anal- for the Ministry of Lands, Resettlement ysis of Wildlife and Agriculture in Com- and Rural Development, Harare. munal Subsistence Communities in Scoones, I. C. 1990. Livestock Populations Gokwe and Nyaminyami." Doctoral dis- and the Household Economy: a Case sertation in preparation. Study from Southern Zimbabwe. Ph.D. Murombedzi, James. 1992. Decentralization thesis, University of London. or Recentralization: Implementing Camp- Smithers, R. H. N. 1988. The Mammals of the fire in the Omay Communal Lands of the Southern African Subregion Univ. of Nyaminyami District Council of Zimba- Pretoria, Pretoria. bwe's Wildlife Management Programme. Stanning, J. 1987. "Household Grain Storage CASS Working Paper 2/1992, Center for and Marketing in Surplus and Deficit Applied Social Sciences, Univ. of Zimba- Communal Farming Areas." In Rukuni bwe. 220 VOLUME II and Eicher (eds.) Food Securityfor South- Walker, B. H., D. A. Matthews, and P. J. Dye. em Africa Dept. Agric. Econ., University 1986. "Management of Grazing Sys- of Zimbabwe, pp. 145-84. tems-Existing versus an Event-Orien- Stocking, M. 1986. The Costs of Soil Erosion tated Approach." South African Journal of in Zimbabwe in Terms of the Loss of Science 82:172. Three Major Nutrients. Consultants' Warren, A., and C. Agnew. 1988. An Assess- Working Paper No. 3, Soil Conservation ment of Desertification and Land Degra- Programme, Land and Water Division, dation in Arid and Semi-arid Areas. AGLS, FAO, Rome, pp. 164. Drylands Programme Research Paper, . 1992. "Land Degradation and Reha- No. 2, International Institute for Environ- bilitation Research in Africa 1980-1990: ment and Development, London. Retrospect and Prospect." Dryland Net- Watt, M. 1913. "The Dangers and Prevention works Paper No. 34, International Institute of Soil Erosion. Rhodesian Agriculture for Environment and Development, Lon- Journal 10:5. don. Western, D. 1979. "Size, Life History and Swanepoel, C. M. 1989. Patterns of Habitat Ecology in Mammals." African Journal of Use by African Buffalo in Mana Pools Ecology 17:185-204. National Park, Zimbabwe. D.Phil. thesis, Whitlow, J. R. 1980. "Deforestation in Zimba- University of Zimbabwe. bwe." Supplement to Zambezia, 1980, Swift, M. J., P. G. H. Frost, B. M. Campbell, University of Zimbabwe, pp. 1-35. i - J. C. Hatton, and K. B. Wilson. 1989. . 1988. "Land Degradation in Zimba- "Nitrogen Cycling in Farming Systems bwe: A Geographical Study." Natural Derived from Savanna: Perspectives and Resources Board, Harare. Challenges. In M. Clarholm and L. Whitlow, J. R., and B. Campbell. 1989. "Fac- Bergstrom (eds.), Ecology of Arable tors Influencing Erosion in Zimbabwe: A Land, Kluwer Academic Publishers, pp. Statistical Analysis." Journal of Environ- 63-76. mental Management 29:17-29. Thompson, P. J. 1975. "The Role of Ele- Zimbabwe Government. 1988. "Value of phants, Fire and Other Agents in the De- Wildlife" Parliamentary Debates, House cline of Brachystegia boehmii Woodland." of Assembly, Hansard series, Sept. pp. Journal of the Southern African Wildlife 1628-44. Management Association 5:11-18. Zimbabwe Government, DNPWLM. 1992. "Wild Life Policy." Dept. of National Parks and Wildlife Management, Harare. ABBREVIATIONS AND ACRONYMS AFC average financial cost AIC average incremental cost AIM Action Impact Matrix BAT best available technology CAC command-and-control (measures) CEB Ceylon Electricity Board CEE Central and Eastern Europe CES constant elasticity of substitution CET constant elasticity of transformation CFC chlorofluorocarbon CFL compact fluorescent lighting CGC Caisse de Compensation General CGE computable general equilibrium (model) CIDIE Committee of International Development Institutions on the Environment c.i.f. cost, insurance, and freight CITES Convention on the International Trade in Endangered Species of Flora and Fauna CMEA Council of Mutual Economic Assistance CSC Cold Storage Commission (Zimbabwe) DC district council (Zimbabwe) DHE district heating enterprises DNPWLM Department of National Parks and Wildlife Management (Zimbabwe) DSM demand-side management EA environmental assessment EAP environmental action plan EC European Community ECLAC U.N. Economic Commission for Latin America and the Caribbean EER energy-efficient refrigerator EPRI Electric Power Research Institute ERS export retention scheme ESAP Economic Standard Adjustment Program ETP Economic Transformation Program FGD flue gas desulfurization f.o.b. free on board FY fiscal year GATT General Agreement on Tariffs and Trade GDP gross domestic product GEF Global Environment Facility GHG greenhouse gas GJ gigajoule GNP gross national product GTZ German Technical Assistance Agency (Gesellschaftfiir Technische Zusammenarbeit) IAST Institute of Agrarian Studies 221 222 VOLUMEI IPCC Intergovernmental Panel on Climate Change LP linear programming LRMC long-run marginal cost LSS Living Standards Survey MADIA Managing Agricultural Development in Africa MEC marginal external cost MIT Polish Ministry of Industry and Trade MOC marginal opportunity cost MPC marginal private cost MUC marginal user cost NAFTA North American Free Trade Agreement NEAP national environmental action plan NEPP National Environmental Policy of Poland NGO nongovernmental organization NPR nominal protection rate NPV net present value NWMT Nyaminyami Wildlife Management Trust (Zimbabwe) OECD Organisation for Economic Co-operation and Development OED Operations Evaluation Department of the World Bank OLS ordinary least squares OS oil steam-electric p.a. per annum PFBC pressurized fluidized bed combustion PM particulate matter PPGC Polish Power Grid Company PPP Purchasing Power Parity SAM social accounting matrix SNA System of National Accounts SOCB state-owned commercial bank SOE state-owned enterprise SRMC short-run marginal cost T&D transmission and distribution TJ terajoule UN United Nations UNSO United Nations Statistical Office VIDCO village development committee VIM vehicle inspection and maintenance (program) WADCO ward development committee WCED World Commission on Environment and Development WHO World Health Organization WPA Wildlife Producers Association ZIC Zimbabwe Investment Centre I THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Telephone: (202) 477-1234 Facsimile: (202) 477-6391 Telex MCI 64145 WORLDBANK MCI 248423 WORLDBANK Cable Address: INTBAFRAD WASHINGTONDC World Wide Web: http.//www.worldbank.org/ E-mail: books@worldbank.org 1379 ':~~~~~~~~~~~~~~~~~I" 'x~~~ ~~~~~~~ ~~9 2IJL1I Il Cover design bySoleil Associates ~~~ISBN 0-8213-3797-1 :: Mt