11260 S1. The Political "conomy b of Agricultural Prhing Policy VOLUME4 A Synthesis of the Economics in Developing Countries Maurice Schiff Alberto Valdés A World Bank Comparative Study FILE ;DPY The Political Economy of Agricultural Pricing Policy VOLUME 4 A Synthesis of the Economics in Developing Countries A World Bank Comparative Study 77he Political Economy of Agricultural Pricing Policy VOLUME 4 A Synthesis of the Economics in Developing Countries A World Bank Comparative Study Maurice Schiff Alberto Valdes PUBLISHED FOR THE WORLD BANK The Johns Hopkins University Press Baltimore and London @ 1992 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W, Washington, D.C. 20433, U.S.A. The Johns Hopkins University Press Baltimore, Maryland 21211-2190, U.S.A. All rights reserved Manufactured in the United States of America First printing September 1992 The findings, interpretations, and conclusions expressed in this pub- lication are those of the authors and do not necessarily represent the views and policies of the World Bank or its Board of Executive Direc- tors or the countries they represent. The material in this publication is copyrighted. Requests for per- mission 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 non- commercial purposes, without asking a fee. Permission to photocopy portions for classroom use is granted through the Copyright Clear- ance Center, 27 Congress Street, Salem, Massachusetts 01970, U.S.A. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabet- ical title list and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from Distribu- tion Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W, Washington, D.C. 20433, U.S.A., or from Publications, The World Bank, 66 avenue d'I6na, 75116 Paris, France. Library of Congress Cataloging-in-Publication Data A World Bank comparative study. The Political economy of agri- cultural pricing policy. Includes bibliographical references. Contents: v. 1. Latin America-v. 2. Asia-[etc.]- v. 4. A synthesis of the economics in developing countries. 1. Agricultural prices-Government policy-Developing countries-Case studies. 2. Agricultural prices-Government policy-Latin America-Case studies. 3. Agricultural prices-Asia- Case studies. 4. Agricultural prices-Africa-Case studies. 5. Agricultural prices-Mediterranean Region-Case studies. I. Krueger, Anne 0. II. Schiff, Maurice W. III. Vald6s, Alberto, date. . IV. Title: Political economy of agricultural pricing policy. HD1417.P65 1991 338.1'8'091724 90-27232 ISBN 0-8018-4030-9 (set) ISBN 0-8018-4531-9 (vol. 4) Contents Foreword vii Preface ix Part I Historical Patterns of Price Intervention 1 Government Direct and Indirect Policies Affecting Agricultural Prices 3 2 The Effects of Intervention on Agricultural Prices 10 3 The Effects of Intervention on Price Variability 36 Part II Output, Consumption, and Trade Effects of Price Intervention 4 Supply Response and Growth in Agricultural and the Economy (in collaboration with Jorge Quiroz) 59 5 Consumption and Trade Effects of Intervention 86 Part III Consequences of Price Intervention of Income Transfer and Equity 6 The Fiscal Impact of Direct Intervention 99 7 The Effects of Intervention on Income Transfers to and from Agriculture 117 8 The Effects of Intervention on Income Distribution and Poverty 150 V vi Contents Part IV Determinants of Agricultural Taxation; Conclusions 9 An Empirical Analysis of Patterns of Intervention 185 10 Summary and Concluding Remarks 199 Appendix: Direct and Total Nominal Protection Rates, by Country, Product, and Year 233 Notes 252 References 262 Foreword This study, by Maurice Schiff and Alberto Valdds, is one of a family of research gems launched by Anne 0. Krueger, then vice president in charge of economic research at the World Bank, now professor of economics at Duke University. Maurice Schiff is senior economist in the Trade Policy Division of the World Bank, and Alberto Vald6s served as director of the International Trade and Food Security Pro- gram of the International Food Policy Research Institute from 1976 through 1990 before joining the World Bank in 1990. The core of this study is an in-depth analysis of the economic effects of agricultural economic interventions in eighteen developing countries during the period from 1960 to 1985. The theoretical approach of this study includes the economic effects of indirect inter- ventions (including industrial protection and macroeconomic poli- cies). Despite their importance, they are usually excluded. They are exceedingly difficult to quantify. In its appeal to data and to quantita- tive analysis, this study rates high. We learn the awful truth that the indirect tax on agriculture was about 22 percent. The direct tax from agricultural pricing policies was about 8 percent. Transfers of resources out of agriculture were enormous. On aver- age the net effects of direct and indirect interventions were 46 percent of the agricultural gross domestic product a year during 1960-84. The modernization of agriculture was being sacrificed at the altar of indus- trialization. The price that these developing countries were paying for their bad economic policies was very costly. In addition to quantifying the impact of direct and indirect policies on agricultural incentives, the study also measures the effect on out- put, consumption, trade, the government budget, intersectoral real income transfers, and income distribution. What is not quantified is the large increase in income that can be had from an economically efficient allocation of the resources of these developing countries. Theodore W. Schultz University of Chicago vii Preface This is the fourth volume in the five-volume series that reports the major findings of the World Bank research project A Comparative Study of the Political Economy of Agricultural Pricing Policies. The first three volumes present summary chapters of the eighteen country cases, with one volume on Asia, one on Latin America, and one on Africa and the Mediterranean region. The eighteen country studies were carried out within a common analytical framework and involved close collaboration between the authors of the country studies and the project's three codirectors. This, the fourth volume, provides a comparative analysis of the findings from the eighteen country studies and deals specifically with the economic aspects of agricultural price intervention. The fifth vol- ume, by Anne 0. Krueger, provides a comparative analysis of the political economy of agricultural pricing policies. Together, the fourth and fifth volumes provide a sythesis of the eighteen country studies undertaken in this research project. The purpose of the project was threefold: to provide systematic estimates of the degree of price discrimination against agriculture within individual countries and to explain how it changed over time; to determine how this intervention affected such key variables as foreign exchange earnings, agricultural output, and income distribu- tion; and to gain further insight into the political economy of agri- cultural pricing policy through a study of the motivations of policy- makers, the economic and political factors determining the degree of agricultural intervention, and the attempts to reform unsuccessful policies. Until recently, analysts were primarily concerned with the direct effect that agricultural pricing policies might have on agricultural product and input prices. According to international trade theory and general equilibrium analysis, however, a policy that protects one par- ix x Preface ticular sector of the economy (in this case, industry) is essentially imposing a tax on other sectors of the economy (in this case, agricul- ture). The tax is also likely to lower the real exchange rate, which will then lower the real return to exportables and unprotected import- competing sectors, which account for most of agriculture. Indeed, a country's general economic policies may have far greater indirect effect on agricultural incentives than its agriculture-specific or direct pricing policies do. Nonetheless, the former have not been system- atically incorporated into analyses of government intervention in agricultural prices. In themselves, estimates of the relative importance of direct (that is, sectoral) and indirect intervention yield useful insights into the impact of developing countries' policies. Moreover, the fact that these estimates are comparable across the eighteen countries examined enables a systematic evaluation of the political economy of agri- cultural price policy, the subject of the final volume of this set. Sys- tematic examination of the impact of these policies on output, con- sumption, government budgets, foreign trade, intersectoral transfers, and income distribution in itself is a considerable contribution to our understanding of the workings of these policies in developing countries. Several criteria were used to select the countries for the project. A foremost concern was to represent a reasonable range of country experience. Therefore, some countries in the group are exporters of food, others are exporters of agricultural (but nonfood) commodities, and still other are importers of food. An effort was also made to achieve some balance between low-income and middle-income coun- tries, as well as among geographic regions. The task would have been impossible without able researchers willing to participate in the proj- ect and to prepare the country studies (published as individual vol- umes in the World Bank Comparative Studies series). The countries included in the project, the participants, and their affiliations were as follows: Argentina Adolfo Sturzenegger, Econometrica, Buenos Aires (with assistance of Beatriz Mosquera) Brazil Jos6 Luiz Carvalho, Universidad Santa Ursula, Rio de Janeiro; Antonio Salazar P. Branddo, Fundagao Getulio Vargas, Rio de Janeiro Chile Herndn Hurtado and Eugenia Muchnik, Catholic University, Santiago; Alberto Vald6s, International Food Policy Research Institute (IFPRI), Washington, D.C. Preface xi Colombia Jorge Garcia Garcia, World Bank, Washing- ton, D.C.; Gabriel Montes, Instituto Co- lombiano Agropecuario (ICA), BogotA C6te d'Ivoire Achi Atsain and Allechi M'Bet, Centre for Economic and Social Research (CIREs), Uni- versit6 Nationale de C6te d'Ivoire, Abidjan Dominican Republic Duty Greene, Sigma One Corporation, Quito; Terry Roe, University of Minnesota, Saint Paul Egypt Jean-Jacques Dethier, World Bank, Wash- ington, D.C. Ghana Dirck Stryker, Associates for International Resources and Development (AIRD), Somerville, Massachusetts Republic of Korea Pal Yong Moon, Kon Kuk University, Seoul; Bong-Soon Kang, Seoul National University, Suwon Malaysia Glenn Jenkins and Andrew Lai, Harvard Institute for International Development, Cambridge, Massachusetts Morocco B. Lynn Salinger, AIRD, Somerville, Massa- chusetts; Hasan Tuluy, World Bank, Wash- ington, D.C. Pakistan Naved Hamid, Asian Development Bank, Manila; Ijaz Nabi, World Bank, Washing- ton, D.C.; Anjum Nasim, Lahore Univer- sity of Management Sciences, Lahore Philippines Ponciano Intal, National Economic and Development Authority, Manila, and Uni- versity of Philippines, Los Bafios; John H. Power, University of Hawaii Portugal Francisco Avillez, Instituto Superior de Agronomia, Lisbon; Timothy J. Finan, Uni- versity of Arizona, Tucson; and Timothy Josling, Food Research Institute, Stanford University, Stanford, California Sri Lanka Surjit Bhalla, World Bank, Washington, D.C. Thailand Ammar Siamwalla and Suthad Setboon- sarng, Thailand Development Research Institute, Bangkok Turkey Hasan Olgun and Haluk Kasnakoglu (with the cooperation of Arslan Gurkan), Middle East Technical University, Ankara Zambia Doris Jansen, Development Technologies Inc., Larkspur, California xii Preface In addition to recruiting country authors, an advisory board was assembled to oversee the project. Board members, knowledgeable about the issues and experienced in analyzing agricultural pricing policies, were asked to comment on all aspects of the project and to take responsibility for reading and commenting on the studies of the individual countries. Board members, their affiliations, and the coun- tries for which they took primary responsibility were as follows: Romeo Bautista, IFPRI Korea, Malaysia, Philippines Hans Binswanger, World Bank Pakistan, Sri Lanka Vinod Dubey, World Bank C6te d'Ivoire, Morocco Peter Hopcraft, World Bank Ghana, Zambia D. Gale Johnson, University of Portugal, Turkey Chicago Yair Mundlak, Hebrew Univer- Argentina, Chile, Thailand sity, University of Chicago, and IFPRI Edward Schuh, University of Brazil, Egypt Minnesota Marcelo Selowsky, World Bank Colombia, Dominican Republic Many other persons helped in bringing the project to a successful conclusion, and we thank them here. They include project adminis- trator Celina Bermudez and her predecessor, Rosario Seoane, who handled an endless flow of personnel and communications matters; Anne Muhtasib, the project secretary, who handled the voluminous correspondence and processed many edited manuscripts; and word processors Myriam Bailey and Estela Zamora. Our successive research assistants, Emmanuel Skoufias, Pierre Nadji, Lilyan Ful- giniti, Jorge Quiroz, and Claudio Montenegro, reviewed the results in the many country reports and summary chapters and assisted with data and statistical analysis for our work on the synthesis volume. Jorge Quiroz also made significant contributions to chapters 4 and 9 of this volume. The editor for the project was Phillip Sawicki, assisted by Paul Wolman, Victoria Macintyre, and Mary Ellen Buchanan. The editors for this volume were Meta de Coquereaumont and Bruce Ross-Larson. A. Vald6s is also most grateful to Isabel Vial de Vald6s and to Marcela, Cecilia, Alberto Jr., and Pilar Vald6s for their patience and support during these years. Part I Historical Patterns of Price Intervention Government Direct and Indirect Policies Affecting Agricultural Prices Industrial and developing countries treat agriculture in very different ways. Industrial countries generally intervene to raise agricultural prices; developing countries tend to intervene to reduce prices below their international level. Because agriculture is usually the largest single economic sector in many developing countries, agricultural price policies have become a major instrument for influencing the economy. For that reason, agricultural price policies often have differ- ent objectives, many of which may be conflicting and few of which may be explicitly stated. These include encouraging food production to reduce food imports and achieve self-sufficiency, stabilizing food prices for producers or reducing them for urban consumers, generat- ing revenue for the government by taxing the sector, increasing for- eign exchange reserves by encouraging exports, and making raw materials available to the industrial sector at prices lower than would otherwise prevail. Direct and Indirect Price Intervention Governments influence agricultural prices both directly, through agricultural sector policies, and indirectly, through industrial protec- tion and macroeconomic policies that tax agriculture relative to trad- ables and nontradables outside the agricultural sector. Indirect inter- ventions affect the prices of agricultural tradables relative to nontradables (through their impact on the real exchange rate) or to other tradables (as a result of industrial protection). These policies affect production incentives by making agriculture more or less attrac- tive than other sectors of the economy. The effects of indirect inter- ventions have until recently essentially remained outside the theoreti- cal and political debate on agricultural price interventions in developing countries. This volume presents the findings of a comparative analysis of the effects of such direct and indirect interventions in eighteen develop- 3 4 Historical Patterns of Price Intervention ing countries over a period of about twenty-five years. The study found that, on average, direct and indirect interventions in the eigh- teen countries taxed agricultural producers. The indirect effects were by far the stronger, easily outweighing the positive influence of direct interventions in the few cases where direct measures protected agri- culture. Intervention usually reduced agriculture's share of gross national product and was often related to both slower growth in agricultural production and agricultural exports and slower economic growth overall. Many of the direct measures have been equivalent to a tax on agriculture, depressing the prices received by agricultural producers below levels that would otherwise prevail. Governments have inter- vened directly through procurement measures (government market- ing boards are often the only legal buyers of agricultural outputs), quotas on exports of food crops and other agricultural commodities, and direct taxation of such exports. Some direct interventions, how- ever, have benefited agricultural producers. Governments have often subsidized the costs of farm credit and important agricultural inputs, such as fertilizer. Many developing countries, to increase their food self-sufficiency, have protected domestic producers of import- competing food products through quantitative restrictions or tariffs on imported commodities. And some countries, responding to the instability of world markets, have intervened to stabilize domestic producer prices in relation to prices on world markets (see Vald6s 1981, chapter 1). Yet other direct interventions have attempted to hold down the costs of food for urban consumers-for example, by fixing retail food prices or imposing ceilings on producer prices. Sometimes govern- ments have set up dual pricing systems to keep producer prices high and consumer prices low, making up the difference out of budgetary resources. Although the fiscal costs of food subsidies can be reduced by targeting programs carefully, such programs have sometimes proved impractical because of their complex administrative demands and the large size of the target groups. Because explicit food subsidies have often proved unsustainable (as in Sri Lanka), leading to bud- getary and balance-of-payments pressures, some governments have instead attempted to reduce consumer prices by taxing agricultural producers. Agricultural production incentives have been even more strongly influenced through the indirect effects of macroeconomic policies and industrial sector protection policies. Macroeconomic policies that cause appreciation of the real exchange rate can raise the relative cost of nontradable inputs and reduce the real purchasing power of income received from the sales of export and import-competing com- modities. And protection for domestic industry affects agriculture by Government Direct and Indirect Policies Affecting Agricultural Prices 5 raising the domestic price of importable agricultural inputs above world prices, reducing the purchasing power of farm households as consumers of manufactured goods, and causing further appreciation of the real exchange rate. Taxing agriculture has had various justifications: administrative simplicity, compensation for a thin tax base, the presumed small cost in forgone output because of agriculture's alleged low responsiveness to incentives, declining terms of trade for agricultural exports, and rising protection in industrial countries. Recently, however, econo- mists have begun to reappraise these traditional arguments for taxing agriculture. Analysis by Schultz (1964), Mundlak (1985), Mundlak, Cavallo, and Domenech (1992), Coeymans and Mundlak (1992), and others strongly suggests that agriculture is a dynamic sector that responds positively to price incentives and that "policies which tax agriculture reduce the investment in agriculture, increase outmigra- tion, and reduce the implementation of new techniques" (Mundlak 1985). Much of the earlier pessimism about the dim trade prospects for agriculture overlooked cost-reducing technological innovations (for example, in transport and storage) and the prospects for expand- ing exports of nontraditional agricultural commodities, which would give developing economies more flexibility to cope with changing world economic conditions. Whatever the original objectives of agricultural price policies, unforeseen changes in the world prices of agricultural commodities, misjudgments about both the reactions of various interest groups to price intervention and the effects of interventions on agricultural out- put, the budget, and the balance of payments have often forced poli- cymakers to change course. (These aspects of the political economy of agricultural price policies are examined in detail in a companion vol- ume to this study by A. 0. Krueger, the last volume of this five- volume series.) Purpose and Scope of the Project Research on development strategies and trade regimes in developing countries since the early 1970s has consistently found that countries that have adopted outward-oriented strategies have been more suc- cessful than countries that sought to build their industrial sector through inward-oriented strategies of import substitution (Little, Scit- ovsky, and Scott 1970; Balassa 1971; Bhagwati 1978; Krueger 1978, 1983; and Krueger, Lary, and Akrasanee 1981). But none of these comparative studies explicitly examined the agricultural sector or looked systematically at the impact of trade and balance of payments regimes or other types of indirect intervention on agricultural incentives. Table 1-1. Product Classification H S a p p S u T z a o S o n T o A B C C G e I t R o y f o m p a C o C o r l m a u S r b S I b a W p r B o f o C t a n M P t R b h g e u o a t h W 1 1 e c f p o t p u i o o o i b e h a g w T c o e i e e e a e r r o e t I i r e C e e u n a e e c e a n s y f a e a n n s s k I k s e r pa m s r r a o s t e Argentina --X---X--------- X-X -- -X - Brazil ------X ----- -M---X-----M- Chile X -M----- X -M-------------M- Colombia ---- X -- X ------ X ---------M- DominicanRep. - - - - X - --------M----------- Egypt -- - ---MX------X---------M- Ghana ---X-- M- ------M----------- C6ted'Ivoire ---XX -- -------M----------- Korea -MMM -M---M---- --M-M---M Malaysia -- ---------X--MX---------- Morocco - X - - - -------------M M - Pakistan -- -----X------X----M -M- Philippines --- --XM-------M----X------ Portugal --M---M---M-- HM-X------XMX SriLanka -- - --X--------MX-----X---- Thailand - - --- -X-------XX---X------ Turkey - M- ----X-X------X--X--X-M- Zambia -- ----MX--------------X--- a. Consists of mutton and Lamb. N Staple, O Nonstaple, M Importable, X Exportable, H Nontradable. Source: Country studies. 8 Historical Patterns of Price Intervention Studies that have examined agriculture have been narrow in scope or have relied on different methodologies, making cross-country comparisons difficult at best. Such well-known comparative studies of agricultural price policies as the Stanford Food Research Institute's examination of rice policies in Asia (Falcon and Timmer 1975) and the World Bank's case studies (unpublished) of administered agricultural prices, taxes, and subsidies of the mid-1970s deal essentially with sectoral output and input price policies. Many individual country studies have estimated the short-term effects of sectoral policies on agricultural output, food consumption, and trade flows. A few others have also explored the political factors influencing agricultural price policies in individual countries. But because the studies used different methodologies and covered differ- ent periods, they do not permit useful comparisons between coun- tries of the effects of either direct or indirect intervention. One set of studies by the International Food Policy Research Institute (IFPRI), and which is summarized in Bautista and Vald6s, (1992), did examine some of the interactions between industrial protection and agri- cultural incentives, but it did not analyze all those interactions or the indirect effects of macroeconomic policies or the underlying political economy issues. This World Bank project was designed to fill these gaps by assess- ing the effects on agriculture of both direct and indirect price inter- ventions over a long period in eighteen representative developing countries, using a common conceptual framework and methodology. The project developed measures of the impact of direct and indirect price policy interventions on relative prices within agriculture and between agriculture and the rest of the economy. These measures were then used to estimate the effects of price interventions on agri- cultural production, consumption, foreign exchange earnings, the budget, income transfers between agriculture and the rest of the economy, and income distribution. In the context of the history of agricultural price interventions in each country, these estimates also provided a basis for analyzing the evolution of the country's political economy of agricultural price policies. (The countries and commodi- ties covered by the project are detailed in table 1.1.) Other important issues for agricultural policy, including optimal investment in agriculture, optimal taxation of agriculture, and agri- culture's role in a country's overall development strategy, were not the central focus of this study. These are important issues, but they cannot be examined in an analysis that focuses on a specific sector of the economy, as this one does, and they require a broader analytical framework. (For a discussion of taxation and public finance in devel- oping countries, and the relation between public finance and eco- nomic development, see World Bank 1988, Newbery and Stern 1987, and Toye 1979.) Government Direct and Indirect Policies Affecting Agricultural Prices 9 Structure of the Volume This volume divides the analysis of agricultural price policy into four parts. Part I deals with the effects of price interventions on the level and variability of agricultural prices (chapters 2 and 3). Part II exam- ines the effects on agricultural output and economic growth (chapter 4) and on consumption and trade (chapter 5). Part III analyzes the effects on the budget (chapter 6), on intersectoral income transfers and the relation between government investment in agriculture and price policy (chapter 7), and on income distribution and poverty (chapter 8). Part IV presents an econometric model of the factors influencing direct agricultural price interventions and the relation- ships between them (chapter 9) and compares the findings on the effects of intervention with the stated objectives of those interven- tions (chapter 10). A statistical appendix presents the annual values of the direct and indirect protection rates for all the countries and prod- ucts in the project. 2Tne Eects of Intervention on Agricultural Prices: Magnitude and Patterns Government price interventions discriminated strongly-and increasingly-against the agricultural sector in developing countries during 1960-84. By far the largest impact came through indirect inter- ventions (industrial protection and macroeconomic policies), which reduced incentives in agriculture relative to other sectors of the econ- omy. On average, direct intervention taxed exportables and protected importables, which are predominantly food products. In most coun- tries, consumer prices for food products moved in the same direction as producer prices in response to direct interventions. In most cases, therefore, direct interventions raised consumer food prices by pro- tecting importables. Indirect interventions, however, lowered both producer and consumer prices. Most analyses of agricultural price policies have focused on direct price interventions; but our findings show that in almost all coun- tries, indirect interventions have reduced agricultural incentives far more than direct interventions. Direct price interventions taxed agri- culture an average of about 8 percent in the eighteen countries over the period studied, indirect interventions taxed agriculture more than 22 percent. This means that total (direct plus indirect) taxation of agriculture has been substantially higher than estimates of direct intervention alone would indicate. It also shows the importance of looking at indirect price interventions when analyzing agricultural price policy and reform. Our findings support the generalization that government price interventions in the agricultural sector tax agriculture in developing countries. For our sample of developing countries, we found that agriculture was highly taxed, on average, and that the tax increased between 1960-72 and 1976-84. In contrast, Anderson and Hayami (1986), in a study of Organization for Economic Cooperation and Development (OECD) countries plus Korea, found that protection was generally positive and that it increased between 1950 and 1980. Our findings also show, however, that this generalization hides several 10 The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 11 important distinctions. Direct intervention has taxed exportables in developing countries, but it has protected importables, thereby dis- torting relative incentives in the sector. With an average tax on exportables of 16.3 percent and average protection of importables of 18.6 percent, direct price interventions have penalized exportables relative to importables at a rate close to 30 percent, with important consequences for the output mix and growth in the sector (see chap- ter 4). Although it is frequently claimed that developing countries have subsidized the consumption of food, our study shows that direct price policies led to higher food prices in a majority of countries, raising consumer food prices an average of 6 percent. In only a few food-importing countries such as Egypt and Zambia and a few food exporters such as Argentina and Thailand did direct interventions substantially lower consumer food prices. If the wage-good argument for cheap food policies has any validity at all in most of the countries examined, it is because indirect interventions caused a fall in (relative) food prices, not because of direct interventions. For most of the countries examined, industrial protection policies reduced agricultural incentives more than did real exchange rate overvaluation, the other main indirect source of agricultural taxation. Industrial protection lowered agricultural prices relative to non- agricultural prices by close to 28 percent; overvaluation lowered prices by about 21 percent. The difference between the two effects is even greater than these numbers indicate, however, because over- valuation of the real exchange rate resulted partly from industrial protection measures.' Consequently, industrial trade policies have hurt agriculture more than the macroeconomic policies that led to distortions in the real exchange rate. Taxation of agriculture has not subsided over time. Total taxation of agriculture increased from an average of 26.4 percent in 1960-72 to 33 percent in 1976-84. In other words, by the second period, farmers were receiving prices about a third lower than they would have received had governments not intervened in prices. Price trends for the period also show that taxation of agriculture increased: the pro- ducer price trend was below the world price trend over the entire period. So despite reforms in a few countries (Chile, Egypt), in gen- eral agriculture was increasingly penalized over time-at least up to the mid-1980s. Types and Measures of Intervention and Country Groupings Some intervention measures, such as import or export taxes and subsidies or restrictions on exports, are applied at the border, and some take the form of price controls or production taxes or subsidies. 12 Historical Patterns of Price Intervention Of these measures, some are direct interventions, specifically affect- ing the prices of particular commodities and creating a wedge between the producer or consumer price and the border price. Others are economywide or indirect interventions, affecting the prices of agricultural commodities in relation to the prices of other goods and services. For each country, we selected major export- and import-competing agricultural commodities, including both food and nonfood products, on the basis of their importance and the representativeness of the policies adopted toward them relative to agriculture as a whole. For most countries, we concentrated on four to six commodities, and that coverage typically represented between 40 and 80 percent of net agri- cultural product. We used nominal protection rates to measure the impact of direct and indirect policies on agricultural incentives. First, we obtained estimates of the commodities' domestic producer, consumer, and bor- der prices, adjusted for transport costs to or from producer and con- sumer locations, storage costs, quality differences, and other ele- ments of the marketing margins. The direct nominal protection rate was measured as the proportional difference between the domestic producer (or consumer) price and the border price evaluated at the official exchange rate and adjusted for the marketing margin, quality difference, and others. We did not consider interventions outside the agricultural sector.2 Where there is a wide gap between the free on board (FOB) and the cost, insurance and freight (CIF) price due to high transport costs, an importable may become an exportable or may not be traded once interventions are removed. This affects the calculations of the effect of interventions and was taken into account. And given the volatility of world prices, some products that would have been exported in one year in the absence of interventions would have been imported in another year. Hence, we believe that measures of intervention aver- aged over time are more meaningful than those for a single year, and that is how they are presented. The indirect nominal protection rate was measured as the propor- tional difference between the domestic agricultural price in relation to a price index of the nonagricultural sector, on the one hand, and the value of that relative price when measured at the equilibrium exchange rate and in the absence of industrial trade interventions, on the other hand. There are three major elements in our calculations of the indirect nominal protection rates: first, the depreciation of the real exchange rate required for the elimination of the nonsustainable part of the current account deficit; second, the depreciation of the real exchange rate due to the removal of trade interventions; and third, the increase in the price of agricultural tradable products relative to non- The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 13 agricultural tradables due to the removal of industrial trade policy interventions. The first two are changes of the price of tradables rela- tive to nontradables; the third is a change of prices within the trad- ables category. The indirect protection rate is a weighted average of the effect of the real exchange rate overvaluation and the effect of industrial protection (see appendix 2-1). The total effect of the interventions is simply the sum of the direct and indirect effects. The methods of calculating direct and indirect protection rates, the equilibrium exchange rate, and the rate of indus- trial protection are described in detail in appendix 2-1. It is important to note that the direct nominal protection rates reported here have been adjusted in order to be consistent with the indirect and total protection rates. The adjusted direct protection rates are consistently smaller in absolute value than the direct nominal protection rates commonly used. This is explained in both note 2 and in appendix 2-1. We obtained nominal protection rates for twenty-six products in eighteen countries (see table 1-1) over a period of about twenty-five years (a total of eighty cases over the period 1960-84).3 (For a few countries, the sample period starts after 1960, and for some it extends to 1985 or 1986.) We then calculated measures of protection for three product categories-staples, importables, and exportables-and for multiyear periods, using divisia indexes.4 Next, we divided the eighteen countries into four groups according to their rate of total protection (the sum of direct and indirect nominal protection rates). Three groups taxed agriculture, on average, and one group protected agriculture. Extreme discriminators (Group I)- the three Sub-Saharan African countries of C6te d'Ivoire, Ghana, and Zambia-had an average total taxation rate of over 50 percent. The representative group of countries (Group II)-Argentina, Colombia, the Dominican Republic, Egypt, Morocco, Pakistan, the Philippines, Sri Lanka, Thailand, and Turkey-had an average total taxation rate of 36 percent (six percentage points higher than the average total tax for all eighteen countries). Mild discriminators (Group III)-Brazil, Chile, and Malaysia-had a total taxation rate of close to 16 percent. Finally, the protectors group (Group IV)-Korea and Portugal-had an average total protection rate of 11 percent. The poorest countries in our sample (Group I) taxed agriculture the most; those with the high- est per capita income (Group IV) protected agriculture. Impact of Interventions on Agricultural Prices This section examines the impact of direct, indirect, and total inter- ventions on the products considered, both overall and for specific product categories, and then looks at changes in these patterns over time. That agriculture is taxed in most developing countries has long 14 Historical Patterns of Price Intervention been recognized. What is new in our findings, however, is the magni- tude of that tax when the effects of indirect interventions are also taken into account. When all price interventions are considered, the average tax on agriculture of 30 percent is close to four times the 8 percent tax from direct price interventions (table 2.1).s That means that farmers received 30 percent less for their output than they would have received in the absence of total price interventions. In fact, total interventions taxed agriculture in all countries except Korea and Por- tugal, where protection through direct price interventions was larger than taxation through indirect price interventions. High rates of total taxation were found in countries representing all the regions examined: Argentina (total tax rate of 39 percent), Chile (22 percent), Colombia (30 percent), and the Dominican Republic (40 percent) in Latin America; C6te d'Ivoire (49 percent), Ghana (60 per- cent), and Zambia (46 percent) in Sub-Saharan Africa; Pakistan (40 percent), the Philippines (27 percent), Sri Lanka (40 percent), and Thailand (40 percent) in Asia; Egypt (44 percent) and Morocco (32 percent) in North Africa; and Turkey (32 percent) in the Near East. For extreme discriminators, the average total tax was almost 52 per- cent, reducing their terms of trade to less than half of what they would have been in the absence of interventions. Direct, Indirect, and Total Intervention Overall, direct intervention led to an average rate of agricultural taxa- tion of about 8 percent for the sample countries and products (table 2-1). The average rate was low because a few countries had a positive rate of direct protection and several others had low rates of direct taxation. Direct price interventions taxed agriculture in fourteen countries and protected it in four (Brazil-excluding coffee-Korea, Portugal, and Turkey). The average rate of direct taxation was 23 percent for the extreme discriminators (and over 32 percent with the usual definition of direct taxation-see note 2 and appendix 2-1), 12 percent for the representative group, and 0.2 percent for the mild discriminators; for the protectors group, direct interventions resulted in a protection rate of 24 percent.6 Direct taxation was highest in Ghana, where it averaged almost 27 percent (primarily because of to the 30 percent tax on cocoa). It was above 20 percent in C6te d'Ivoire (almost 26 percent, with a 28 per- cent tax on coffee and cocoa), Thailand (25 percent, with a 31 percent tax on rice), and Egypt (25 percent, with a 28 percent tax on cotton and a 30 percent tax on rice). A similar result was found in India by Gulati and Pursell, where the direct taxation rate was 21 percent over the period 1970171-87188. Indirect price interventions (exchange rate and industrial protec- The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 15 Table 2-1. Direct, Indirect, and Total Rates of Nominal Protection for the Eighteen Sample Countries, 1960-84 (period averages in percentages) Tax due to Indirect industrial Direct Total Country Period protections protectionb protectionc protection C6te d'Ivoire 1960-82 -23.3 -23.2 -25.7 -49.0 Ghanad 1958-76 -32.6 -32.4 -26.9 -59.5 Zambia 1966-84 -29.9 -21.4 -16.4 -46.3 Group I averagee 1960-84 -28.6 -25.7 -23.0 -51.6 Argentina 1960-84 -21.3 -39.5 -17.8 -39.1 Colombia 1960-83 -25.2 -37.8 -4.8 -30.0 Dominican Republic 1966-85 -21.3 -20.8 -18.6 -39.9 Egypt 1964-84 -19.6 -27.5 -24.8 -44.4 Morocco 1963-84 -17.4 -13.4 -15.0 -32.4 Pakistan 1960-86 -33.1 -44.9 -6.4 -39.5 Philippines 1960-86 -23.3 -33.0 -4.1 -27.4 Sri Lanka 1960-85 -31.1 -40.1 -9.0 -40.1 Thailand 1962-84 -15.0 -13.9 -25.1 -40.1 Turkey 1961-83 -37.1 -57.4 5.3 -31.8 Group I averagee 1960-86 -24.2 -32.8 -12.0 -36.4 Brazil 1969-83 -18.4 -21.4 10.1 -8.3 Chile 1960-83 -20.4 -37.4 -1.2 -21.6 Malaysia 1960-83 -8.2 -9.9 -9.4 -17.6 Group III averagee 1960-83 -15.7 -22.9 -0.2 -15.8 Korea, Rep. of 1960-84 -25.8 -26.7 39.0 13.2 Portugal 1960-84 -1.3 -1.0 9.0 7.7 Group IV averagee 1960-84 -13.6 -13.9 24.0 10.4 Sample averagee -22.5 -27.9 -7.9 -30.3 a. Indirect protection measures the effect of exchange rate overvaluation and indus- trial protection on agriculture's relative price, PAIPNA and is given by (Eo1E*)(PNA/PNA) - 1. It is a weighted average of the degree of overvaluation of the exchange rate and of the tax due to industrial protection. b. The tax on agriculture resulting from industrial protection, t, is given by [1/(1 + U1- 1. c. Direct protection is measured as the difference between relative producer and border prices at the official exchange rate, PA/PNA - PA/PNA, after adjusting for all relevant margins and divided by the relative price in the absence of all interventions, PA/PN. Note that this measure of direct protection is smaller in absolute value than the usual measure, which uses PAIPNA as the denominator. d. Figures for Ghana are presented only until 1976; after that, the border price at the farmgate level is negative for a number of products. The indirect protection rate for the period 1958-84 is - 47.6. e. Simple, unweighted cross-country average. Source: Authors' tabulation based on data provided in the country studies. 16 Historical Patterns of Price Intervention tion policies) taxed agriculture even more severely and in almost every country. At an average rate of 22.5 percent, indirect interven- tions reduced agriculture's domestic terms of trade at three times the rate of price policies directly affecting agriculture. The indirect tax was above 15 percent in all but two countries-Portugal and Malaysia, which benefited from stable macroeconomic policies and a relatively low degree of protection-and was very high in several countries. The average reduction in farm prices relative to nonfarm prices because of indirect interventions was 37 percent in Turkey, 33 percent in Pakistan, almost 33 percent in Ghana (but 48 percent over the entire period; see note b of table 2-1), and 31 percent in Sri Lanka. The average indirect rate of taxation was almost 29 percent for the extreme discriminators, 24 percent for the representative group, 16 percent for the mild discriminators, and 14 percent for the protectors who pro- tected agriculture only through direct price interventions. The average tax on agriculture from industrial protection policies (28 percent) was larger than the average indirect tax (22.5 percent). This implies that industrial protection policies (tariffs and quantitative restrictions) had a greater effect on the indirect tax than did over- valuation of the real exchange rate. This dominance is intensified when that part of the overvaluation of the real exchange rate that results from industrial protection policies is also taken into account. Because industrial protection policies penalize agriculture relative to other tradables while overvaluation penalizes agriculture relative to nontradables, this means that indirect intervention taxed agriculture relatively more compared to other tradables than to nontradables. The tax on agriculture due to industrial protection policies was above 15 percent in all but four countries (Malaysia, Morocco, Portu- gal, and Thailand). Taxation of agriculture through industrial protec- tion was highest in Turkey, where a 135 percent rate of industrial protection resulted in a tax on agriculture (relative to industry) of about 57 percent and contributed to the 37 percent tax on agriculture from indirect interventions. Other countries with high rates of indus- trial protection were Pakistan (81 percent, resulting in a tax on agri- culture of about 45 percent), Sri Lanka (67 percent, for a tax of about 40 percent), and Argentina (65 percent, for a tax of close to 40 per- cent). Similar results were obtained by Gulati and Pursell for India. They found that industrial protection was 59 percent, implying a 37 percent tax on agriculture. The tax on agriculture from overvaluation of the currency averaged about 21 percent and was above 10 percent in all countries except Malaysia and Portugal. Zambia had the most overvalued exchange rate over the period, with an official rate at less than half its equilib- rium value. Ghana's official exchange rate was about 38 percent below its equilibrium value during 1958-76, and became even more The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 17 overvalued between 1976 and 1983 (95 percent below its equilibrium value) before beginning to improve after the reforms initiated in 1984. In Colombia the 19 percent tax on agriculture as a result of real exchange rate overvaluation was several times larger than the direct tax of almost 5 percent.7 Together, industrial protection policies and real exchange rate over- valuation (resulting from macroeconomic and industrial protection policies) caused a high degree of indirect taxation in most countries, reducing agriculture's terms of trade by over one-fifth (22.5 percent) on average. And for the eighteen countries, indirect taxation was almost three times higher on average over the period than direct taxation.8 Patterns of Intervention by Product Category Price interventions and their effects are not uniform across product categories-importables, exportables, and staples-in part because of different policy objectives. Taxes on agricultural exports are an impor- tant source of fiscal revenue for a number of developing countries, while a desire to reduce a country's dependence on food imports tends to limit taxation of the domestic production of importable com- modities, which are mainly food products. One would therefore expect to find that exportable products are taxed more heavily than importables. What has come as a surprise, however, is that direct interventions have actually protected the pro- duction of importables in eleven of sixteen countries (Argentina and Thailand have no important importable products), with an average protection rate of over 14 percent for the 16 countries (table 2-2). In nine of the eleven countries, the average protection rate for import- ables was above 17 percent, implying that producers received over 17 percent more for their products than they would have in the absence of direct interventions. (With the usual method of calculating nominal protection rates-see note 2 and appendix 2-1-direct protection was over 20 percent in those nine countries.) Ghana had the highest average rate of direct protection for import- ables (43 percent, with the same average rate on both corn and rice),9 followed by Korea (39 percent, with 45 percent on rice, 49 percent on barley and beef, 75 percent on soybeans, and 27 percent on pork) and Sri Lanka (39 percent on rice). Other countries with high rates were C6te d'Ivoire (26 percent on rice), Malaysia (24 percent on rice), Brazil (20 percent, with a rate of 17 percent on wheat and 20 percent on rice), the Dominican Republic (19 percent on rice), Portugal (18 per- cent, with 38 percent on wheat, 24 percent on corn, 71 percent on beef, 8 percent on milk, and 2 percent on rice), and the Philippines (17.4 percent, with 30 percent on corn and 6 percent on rice). Table 2-2. Direct and Total Nominal Rates of Protection by Product Categories, 1960-84 (period averages in percentages) Staples Importables Exportables Direct Total Direct Total Direct Total Country Period protection protection protection, protection protectiona protection C6te d'Ivoire 1960-82 26.2 2.9 26.2 2.9 -28.7 -52.0 Ghanab 1958-76 46.4 13.8 42.9 10.3 -29.8 -62.4 Zambia 1966-84 -16.4 -46.3 -16.4 -46.3 -3.1 -33.0 Group I averagec 1960-84 18.7 -9.9 17.6 -11.0 -20.5 -49.1 Argentina 1960-84 -18.5 -39.8 n.a. n.a. -17.8 -39.1 Colombia 1960-83 9.9 -15.3 14.5 -10.6 -8.5 -33.7 Dominican Republic 1966-85 -18.3 -39.6 19.0 -2.3 -24.8 -46.1 Egypt 1964-84 -24.4 -44.0 -5.1 -24.7 -32.8 -51.4 Morocco 1963-84 -15.0 -32.4 -8.2 25.6 -18.5 -35.9 Pakistan 1960-86 0.6 -32.5 -6.9 -40.0 -5.6 -38.7 Philippines 1960-86 -4.6 -27.9 17.4 -5.9 -11.2 -34.5 Sri Lanka 1960-85 39.0 7.9 39.0 7.9 -18.4 -49.5 Thailand 1962-84 -27.6 -42.6 n.a. n.a. -25.1 -40.1 Turkey 1961-83 13.8 -23.3 8.2 -28.9 3.6 -33.5 Group II averagec 1960-86 -4.5 -28.9 7.8 -13.0 -15.9 -40.2 Brazil 1969-83 20.2 1.8 20.2 1.8 5.4 -13.0 Chile 1960-83 -1.2 -21.6 -1.2 -21.6 13.5 -6.9 Malaysia 1960-83 23.6 15.4 23.6 15.4 -12.7 -20.9 Group III averagec 1960-83 14.2 -1.5 14.2 -1.5 2.1 -13.6 Korea, Rep. of 1960-84 38.4 12.6 39.0 13.2 n.a. n.a. Portugal 1960-84 15.4 14.1 17.8 16.5 0.5 -0.8 Group IV averagec 1960-84 26.9 13.3 28.4 14.8 0.5 -0.8 Sample averagec 1960-86 6.0 -16.5 14.4 -8.6 -12.6 -34.8 n.a. Not applicable. a. Direct protection is measured as the difference between relative producer and border prices at the official exchange rate, PA/PNA - PAWINA, after adjusting for all relevant margins and divided by the relative price in the absence of all interventions, PA/PA. Note that this measure of direct protection is smaller in absolute value than the usual measure, which uses P,/PNA as the denominator. b. Figures for Ghana are presented only until 1976; after that, the border price at the farmgate level is negative for a number of products. The indirect protection rate for the period 1958-84 is -47.6. c. Simple, unweighted cross-country average. Source: Authors' calculations based on data provided in the country studies. 20 Historical Patterns of Price Intervention In four countries, direct price interventions resulted in taxation of producers of importables. Zambia had the highest average rate of taxation (16 percent on white maize), followed by Morocco (8 percent) and Pakistan (7 percent)-which both protected sugar and taxed wheat-and Egypt (5 percent, with a tax on corn and wheat and high protection of sugar). The direct protection rate on importables was close to zero in Chile, which protected wheat and milk and taxed beef. Since most importables, with the exceptions of barley and corn, which are used mainly as feed, are staples, staples tend to follow the direct protection pattern of importables. A few staples are export- ables, however, (see table 1-1), so direct protection rates for staples and importables differ in some countries. Argentina has no important agricultural importables and exports a number of staples that are taxed at an average rate of nearly 19 percent (beef at 19 percent and wheat at 18 percent). Egypt's 31 percent tax on rice, an exportable, boosted the tax on staples to 24 percent, compared with 5 percent on importables. In the Philippines, the 14 percent tax on sugar, an exportable, pushed staples into the taxed category (almost 5 percent), whereas importables enjoyed a 17 percent rate of direct protection. Thailand has no important agricultural importables and exports sugar, which is protected (about 17 percent), and rice, which is taxed (about 31 percent). Since rice is by far the more important export, staples are taxed on average at over 27 percent. Consequently, the cross-country average direct nominal protection rate for staples is lower than for importables, but it is still positive at 6 percent. About half the countries directly protect the production of staples, possibly in order to increase the degree of self-sufficiency. Rice is the most highly protected staple-at 46 percent in Ghana, 39 percent in Sri Lanka, 38 percent in Korea, 26 percent in C6te d'Ivoire, 24 percent in Malaysia, and 20 percent in Brazil. Rice appears to be highly protected when it is imported and taxed when it is exported- providing revenue to the government in both cases. Colombia is an interesting example of this pattern, protecting rice when it was imported and taxing it starting in 1971 as it began to be exported. Although direct interventions protected importables and staples in most countries, they taxed exportables in thirteen of seventeen coun- tries (Korea has no important agricultural exportables). The average direct tax on exportables was 12.6 percent, and adding indirect inter- ventions brings the total tax to 35 percent. That means that farmers would have received over 50 percent (35/65) more for their exportable products relative to prices in the nonagricultural sector had the gov- ernment not intervened in agricultural prices. Egypt had the highest average rate of direct taxation of exportables (almost 33 percent, with a 31 percent tax on rice and 28 percent on The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 21 cotton), followed by Ghana (30 percent on cocoa), C6te d'Ivoire (29 percent average for coffee and cocoa), Thailand (25 percent, with a 31 percent tax on rice, the principal exportable, 14 percent on rubber, 3 percent on corn, and 17 percent protection on sugar), and the Domin- ican Republic (25 percent, with a 27 percent tax on sugar and 11 percent on coffee), Morocco (19 percent on barley), Sri Lanka (18 percent, with a 20 percent tax on tea, the principal exportable, 17 percent on rubber, and 16 percent on copra), and Argentina (18 per- cent, with an 18 percent tax on wheat and beef, 16 percent on sor- ghum, 14 percent on corn, and 10 percent on soybeans and sun- flower). Slightly lower rates prevailed in Malaysia (13 percent, with a 15 percent tax on rubber and 9 percent on palm oil), the Philippines (11 percent, with a 14 percent tax on sugar and 9 percent on copra), and Colombia, Pakistan, and Zambia (less than 10 percent). Three countries protected exportables through direct price inter- ventions. Brazil, with net protection on average (but excluding cof- fee), protected corn but taxed cotton and soybeans, Chile protected grapes and apples, and Turkey, also with net protection on average, protected cotton and taxed hazelnuts and tobacco. Portugal had a direct protection rate close to zero on exportables. Since importables were generally protected while exportables were taxed, relative prices within agriculture were distorted, sometimes severely. In C6te d'Ivoire, the direct protection rate of rice relative to coffee and cocoa was 77 percent on average over the period. In Malay- sia, corn had a 51 percent protection rate relative to sugar; in Sri Lanka, rice's relative protection over tea was 84 percent. On average for all the countries examined, the protection rate on importables relative to exportables was over 30 percent. Clearly, direct interven- tions must have led to substantial distortions in resource allocation within agriculture in a number of countries. Gulati and Pursell obtained similar results for India, with strong direct taxation of the major exportables and strong direct protection of most importables. Changing Patterns of Intervention Protection varies as external circumstances, policies, and policy objec- tives change. Both external factors-the commodity price boom, the oil price shocks, and the debt crisis-and internal factors-changes in trade and other policies, changes in government ideology-might be expected to influence price intervention policies and their effects over time. EFFECTS OF EXTERNAL ECONOMIC FACTORS. The mid-1970s were a period of substantial change in international economic conditions, and so constitute a plausible demarcation point for an examination of 22 Historical Patterns of Price Intervention changes in the pattern of price intervention. The commodity price boom of 1972-75 inflated the food import bill of most developing countries, giving rise to increased concerns about food security and self-sufficiency. In response, one would expect an increase in the rate of direct protection on importables in 1976-84 compared to 1960-72. At the same time, the two oil price shocks of 1973 and 1979 and the interest rate shock of the early 1980s caused increases in the current account deficits of developing countries that could be expected to increase the rate of indirect taxation. Although we expected an increase in the rate of direct protection for importables in the second period because of food security con- cerns, we found that the cross-country average (excluding Ghana; see notes a and b in table 2-3) actually fell from about 23 percent in 1960- 72 to 19 percent in 1976-84. If we also exclude Korea, where direct protection increased dramatically from 35 to 91 percent, the average falls from 22 to 15 percent, or by close to one-third. For the represen- tative group of countries (Group II), direct protection of importables fell by about 60 percent, from about 25 percent to 10 percent. Among individual countries, direci protection of importables fell significantly in five countries (Brazil, Colombia, Pakistan, Portugal, and Sri Lanka) and increased significantly in six (Chile, C6te d'Ivoire, the Dominican Republic, Ghana, Korea, and Malaysia). By contrast, we did not expect any significant change in direct protection rates for exportables as a result of the economic changes of the mid-1970s. The average direct tax did increase somewhat, from almost 14 to 16 percent between 1960-72 and 1976-84, with the largest increase coming in the group of mild discriminators (from 3.6 to 6.2 percent). In most countries, direct protection rates fell overall from 1973 to 1976 as countries tried to stabilize at least partially domestic prices during the world commodity price boom. Direct protection rates also fell during the boom of 1979-81 in many countries, but not by as much. (Price stabilization is examined in more detail in chapter 3.) On average, the direct tax rose from 4.9 to 6.2 percent between 1960-72 and 1976-84. As expected, the rate of indirect taxation increased in the second period: from 21.5 percent on average in 1960-72 to 26.8 percent in 1976-84, or by 25 percent. The rate increased in twelve of the eighteen countries, with the largest increases in the three Sub-Saharan coun- tries of the extreme discriminators group. The average for that group tripled between the two periods-from 18.5 to 55 percent-giving those nations not only the highest increase between periods but also the highest indirect tax of all groups in the second period. Of the countries whose indirect tax fell between 1960-72 and 1976-84, the largest drops came in Chile and Korea, which had introduced wide- ranging liberalization programs. Table 2-3. Evolution of Levels of Intervention (NPRs) of Producer Prices Direct Indirect All products Importables Exportables Country Period 1960-72 1976-84 1960-72 1976-84 1960-72 1976-84 1960-72 1976-84 C6te d'Ivoire 1960-72/1976-82 -17.5 -34.5 -32.2 -39.6 27.3 41.4 35.7 43.5 Ghana 1958-72/1976-84 -32.9 -79.6 -37.7 a 37.6 a -40.0 a Zambia 1966-72/1976-84 -5.1 -51.0 -29.6 -20.7 -29.6 -20.7 1.4 -8.1 Group I average -18.5 -55.0 -33.2 30.2 11.8 10.4 -24.8 -25.8 Argentina 1960-72/1976-84 -20.3 -26.2 -21.5 -18.5 n.a. n.a. 21.5 -18.5 Colombia 1960-72/1967-83 -24.6 -28.1 -2.0 -4.2 34.0 10.6 -10.4 -4.6 Dominican Republic 1966-72/1976-85 -26.0 -19.1 -24.1 -16.6 22.7 38.1 -32.5 -25.6 Egypt 1964-72/1967-84 -28.6 -16.6 -28.1 -23.6 -1.1 -5.7 -36.4 -32.1 Morocco 1963-72/1976-84 -12.9 -24.6 -20.5 -5.4 -3.4 -2.6 -34.0 -7.8 Pakistan 1960-72/1976-86 -38.4 -27.7 16.1 -30.0 11.1 -26.3 20.0 -31.4 Philippines 1960-72/1976-84 -21.3 -26.7 5.8 -10.9 28.1 30.4 -9.1 -16.3 0 Sri Lanka 1960-72/1976-85 -27.8 -32.8 -9.2 -19.3 85.4 17.6 -22.6 -31.4 Thailand 1962-72/1976-84 -13.9 -17.7 -33.4 -17.7 n.a. n.a. 33.4 -17.7 Turkey 1961-72/1976-83 -37.7 -39.0 16.3 7.3 20.5 18.4 9.4 -5.9 Group II average -25.2 -25.9 -10.1 -14.0 24.7 10.1 -17.1 -19.1 Brazil 1969-72/1976-83 -17.1 -18.7 45.5 0.0 83.0 3.2 -26.7 -1.3 Chile 1960-72/1976-83 -23.9 -10.7 -5.4 6.7 -5.7 7.1 26.7 -0.0 Malaysia 1960-72/1976-83 -9.3 -6.4 -8.1 -11.2 11.1 64.4 -10.8 -17.4 Group III average -16.8 -11.9 10.7 -1.5 39.5 24.9 -3.6 -6.2 Korea, Rep. of 1960-72/1976-84 -34.6 -13.3 35.4 90.8 35.4 90.8 n.a. n.a. Portugal 1960-72/1976-84 4.9 -10.0 12.3 7.8 25.2 14.9 0.2 0.3 Group IV average -14.9 -11.7 23.9 49.3 30.3 52.9 0.2 0.3 Total averageb -21.5 -26.8 -4.9 -6.2 22.9 18.8 -13.5 16.3 a. The adjusted border prices in Ghana were negative in a number of years, and so were the calculated NPRs even though protection is higher than in the case where the adjusted border price is zero (which implies an NPR equal to infinity). b. The average does not include the NPRs of Ghana. Source: Authors' calculations based on data provided in the country studies. 24 Historical Patterns of Price Intervention Overall, total taxation from direct and indirect intervention rose between 1960-72 and 1976-84 from a positive rate of protection of 1.4 percent to a tax of 8 percent for importables and from a tax of 35 percent to a tax of 42.8 percent for exportables. The total tax for all products increased from 26.4 to 33 percent. RESPONSE TO CHANGES IN INDIRECT TAXATION. In some countries the relationship between changes in indirect taxation and the changes in direct taxation was important. Countries that undertook major eco- nomic reform through trade liberalization and macroeconomic stabi- lization measures that reduced indirect taxation often made changes in direct price intervention policies as well. In Chile general economic reform included reduced intervention in the agricultural sector. As indirect taxation fell between 1960-72 and 1976-83, Chile eliminated the direct tax on cattle (turning a 5.7 percent direct tax on importables to zero protection in most years) and all direct protection on grapes and apples, causing direct protection on exportables to fall from 27 percent to zero in the second period (table 2-3). Other countries accompanied trade liberalization measures and lower indirect taxation with increased direct price intervention in the agricultural sector. In Korea, as indirect taxation fell from 35 percent in 1960-72 to 13 percent in 1976-84, direct protection rose from almost 36 to 91 percent and total protection increased dramatically from 0.8 to nearly 76 percent. Korea is an exception in our sample of develop- ing countries because of its high rate of direct protection and its rapid economic growth. Korea's policies followed the pattern displayed earlier by a large number of OECD countries (1987): as agricultural incomes fell further behind urban incomes, the demand for protec- tion increased. And as the share of agriculture in gross domestic product (GDP) and in consumption fell, the budgetary cost and the real income loss to urban consumers of protecting agriculture fell, lessening resistance to protection (the cost of obtaining protection fell or its supply increased). In some countries, indirect taxation increased and was partly com- pensated by higher direct protection. To provide some relief from the extreme rates of indirect taxation of agriculture, especially in the early 1980s (rates soared from 33 percent in 1958-72 to almost 80 percent in 1976-84), Ghana increased direct protection of importables and began to protect exportables (cocoa), which had previously been taxed. In the second half of 1983, Ghana introduced major reforms and began a long process of structural adjustment, leading to a fall in indirect taxation and improvements in its direct price intervention policies in agriculture (Stryker 1991). In C6te d'Ivoire the opposite occurred, as indirect taxation increased from 18 to 35 percent between 1960-72 and 1976-84. Direct The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 25 taxation rose overall from 32 to almost 40 percent and for exportables from 27 to 41 percent, although importables (rice) received higher direct protection. Consequently, total taxation rose from 44 to 60 per- cent, although in the late 1980s the tax on cocoa and coffee production became a subsidy as world prices fell. For some countries, direct agricultural pricing policies seem to have had little relation to economywide policies or to external events affect- ing indirect taxes. Zambia's rate of direct taxation changed little over time, although its indirect tax rate rose almost continuously, from low protection in 1966-67 to a tax of over 50 percent in 1980-84. And some countries exhibited considerable stability in all their economic poli- cies. Both direct and indirect agricultural pricing policies remained generally unchanged in Malaysia. Direct taxation there fluctuated between 0 and 20 percent, averaging 10 percent from 1960 to 1983; the peak came during the commodity boom in 1974. Indirect taxation fluctuated between 1.6 and 12 percent, averaging 8 percent. RESPONSE TO CHANGES IN POLITICAL IDEOLOGY. In some countries, patterns of intervention changed as the political ideology of the gov- ernment in power changed. When Bhutto's Peoples party came into power in Pakistan in 1971, direct controls proliferated, in keeping with that government's populist philosophy, changing the complex- ion of direct intervention from overall protection to overall taxation. Between 1960-71 and 1972-86, direct price interventions in wheat, rice, cotton, and sugar went from protecting these products at a rate of almost 20 percent to taxing them at 33 percent. At the same time, however, the indirect tax rate fell, from an average of 41 percent in 1960-71 to 27 percent in 1972-86. During the Zia government that followed (1977-88), the system of government interventions was largely dismantled and the rate of direct protection began to rise slowly, although it remained negative (Hamid, Nabi, and Nasim 1991). In Chile, which also experienced a radical change in political ideol- ogy when the military government took power in 1973, both direct and indirect interventions were reduced. This led to a substantial fall in the rate of indirect taxation and to elimination of both the direct tax on cattle and the direct protection on apples and grapes. RESPONSE TO FISCAL CHANGES. In some countries fiscal pressures or a decline in agriculture's share in fiscal revenues brought about a reduction in price interventions. Sri Lanka introduced major reforms in 1978 to reduce the fiscal burden of its large rice subsidies to pro- ducers and consumers. Between 1960-77 and 1978-85, direct protec- tion of rice fell from 80 to 3.7 percent. In Thailand the direct tax on exportables began to fall in the mid-1970s, from an average of about 26 Historical Patterns of Price Intervention 34 percent in the 1960s to about 13 percent in 1980-84. Siamwalla and Setboonsarng (1989) attribute the drop to the gradual reduction in the rice export tax as a share of fiscal revenue. Producer Price Trends On average actual relative producer prices for the eighteen countries during 1960-84 exhibited little trend. Producer prices fell by 0.2 per- cent a year for exportables and exhibited no change for importables. Over the same period world prices increased at an annual rate of almost 1 percent for exportables and 0.25 percent for importables. The trend in actual producer prices was lower than the trend in world prices (defined as producer prices in the absence of total intervention) principally because of the increase in indirect taxation (industrial pro- tection policies and overvaluation of the real exchange rate) over the period examined (table 2-3). Although actual producer prices exhibited little trend changes in producer prices were significant for some countries. For importables, the largest increases occurred in C6te d'Ivoire, Egypt, Korea, and Zambia, and the largest decreases occurred in Colombia, the Domini- can Republic, Morocco and Turkey. For exportables the largest increases occurred in Colombia, C6te d'Ivoire, and Egypt, and the largest decreases occurred in Argentina, Malaysia and Zambia. Producer price trends derived partly from changes in world prices and partly from changes in intervention, with relative influences varying by product type and country. In the case of importables, the annual price rise in Korea resulted from an increase in the world price and a dramatic increase in direct protection. In Zambia, producer prices for importables increased more slowly than world prices because of large increases in indirect taxation. In Morocco, the fall in producer prices resulted from a fall in world prices and a large increase in indirect taxation. For exportables, the trend in producer prices in several countries, including Argentina, Colombia, and Malaysia, depended almost entirely on world prices. In Egypt pro- ducer prices increased faster than world prices because of a fall in indirect taxation, in Zambia producer prices fell faster than world prices because of the large increase in indirect taxation. Impact of Interventions on Consumer Food Prices It is commonly believed that developing countries tax agriculture by keeping food prices below their border-price equivalent in order to appease the more vocal and influential urban consumers and lower urban wages for the benefit of private and public employers. Our results show, however, that direct price interventions slightly protect The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 27 food production on average, implying a small tax on food consump- tion. And with few exceptions, such as Egypt and Zambia, consump- tion is taxed when food is imported. Direct protection of food prices in the thirteen countries for which data on the effects of direct and total interventions on consumer food prices were available was on average positive and small (6 percent), indicating a small tax on consumers (table 2-4). Direct interventions taxed consumers in Brazil, Ghana, Korea, Portugal, and Turkey. The consumption tax is high in Brazil mainly because our analysis did not include beef consumption, which is subsidized, and because of the high share of rice, whose production is heavily protected, in the two products selected (rice and wheat; although wheat consumption was subsidized in most periods, rice was protected-at 19 percent on aver- age, but at over 300 percent in 1971 and over 100 percent in 1970, 1972, and 1979). Consumers were also taxed slightly in Morocco, Table 2-4. Direct, Indirect, and Total Nominal Protection Rates, Consumer Prices, Selected Countries (average percentage) Country Period Indirect Direct Total Ghana 1958-84 -46.1 42.5 -3.6 Zambia 1967-85 -31.9 -23.8 -55.7 Average -39.0 9.4 -29.6 Colombia 1960-83 -25.2 2.7 -22.5 Dominican Republic 1966-85 -21.3 -4.4 -25.7 Egypt 1965-84 -19.2 -31.3 -50.5 Morocco 1960-84 -16.8 4.7 -12.1 Philippines 1960-82 -23.3 -9.7 -33.0 Sri Lanka 1960-84 -32.0 -10.3 -42.3 Thailand 1962-84 -15.0 -17.9 -32.9 Turkey 1971-83 -36.3 40.7 4.4 Average -23.6 -3.2 -26.8 Brazil 1970-83 -18.9 -57.9 39.0 Korea, Rep. of 1960-84 -25.8 16.2 -9.6 Portugal 1960-84 -1.3 10.1 8.8 Average -13.6 13.1 -0.4 Total average -24.1 6.0 -18.1 Note: The products included in each country are: Brazil-rice and wheat; Colombia- rice and wheat; Dominican Republic-rice and sugar; Egypt-rice and wheat; Ghana- rice; Korea-beef, pork, rice, and barley; Morocco-sugar and wheat; Philippines-rice and sugar; Portugal-wheat and milk; Sri Lanka-rice; Thailand-sugar and rice; Turkey-wheat and mutton; Zambia-maize. Source: Authors' calculations based on data provided in country studies. 28 Historical Patterns of Price Intervention where direct protection was positive (mainly because of sugar) but low, and in Colombia. Consumers were subsidized in Egypt, the Philippines, Sri Lanka, Thailand, and Zambia through direct intervention. The result for Egypt is as expected because of its large consumer subsidies. As was shown in the discussion of producer prices, direct price interventions on rice tend to protect production (tax consumers) when rice is im- ported (Ghana, and C6te d'Ivoire, which is not shown in table 2-4) and to tax production (subsidize consumers) when it is exported (Thailand). Sri Lanka is an exception since the consumer price of rice, an import, was subsidized until the late 1970s. Finally, direct protec- tion was negative but low in the Dominican Republic as well, indicat- ing a slight subsidy. Although direct interventions taxed food consumption on average, indirect interventions lowered relative food prices in all countries, implying a subsidy to food consumption relative to nonagricultural prices. Total (direct and indirect) protection was also negative in all countries except Brazil,10 Portugal, and Turkey, resulting in an aver- age consumer subsidy of 18 percent. The highest total consumer sub- sidies were found in Zambia (56 percent) and Egypt (51 percent). In all countries except Sri Lanka with a rate of direct protection of consumer food prices larger than 5 percent in absolute value, the rate of direct protection of consumer prices and the rate of direct protec- tion of producer prices for staples have the same sign (compare tables 2-4 and 2-2). That means that when consumption has been taxed (a positive sign in table 2-4), production of staples has been protected (a positive sign in table 2-2), and when consumption has been subsi- dized, production has been taxed-both prices move in the same direction. In Korea, the direct tax on consumption covered only a part of the direct protection of rice and beef production, and the rest was financed through the government budget, leading to fiscal problems in a number of years (see chapter 6). Only in Sri Lanka did direct interventions both raise the producer price for rice and subsidize rice consumption. As might be expected, the policy led to budget prob- lems, and the subsidies were gradually eliminated beginning in 1978. (The implications of producer and consumer price policies for the budget and for intersectoral real income transfers are examined in chapters 6-8.) Conclusion Total taxation of agriculture has been substantial, averaging more than 30 percent over two and a half decades in the eighteen countries examined. Direct sectoral policies have received most of the attention, but we found that the dominant influence on agricultural prices has come through indirect, economywide policies. Indirect measures The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 29 have substantially reduced agricultural prices relative to prices in other sectors, with an effect on prices nearly triple that of direct sectoral measures. Direct price interventions have had a relatively small effect, protect- ing the production of importables (about 14 percent) and taxing the production of exportables (about 13 percent), resulting in an average tax on agriculture of about 8 percent. Of the indirect interventions, industrial protection has penalized agriculture more than overvaluation of the real exchange rate in two- thirds of the countries examined. The effect of indirect taxation has dominated both direct taxation of exportables and direct protection of importables. Direct price intervention did not lower the cost of food to urban consumers on average-it even raised it somewhat. In most countries that import food, direct intervention has raised prices for both con- sumers and producers. And in most cases in which direct interven- tion lowered prices to producers, consumer prices also fell. Rates of direct and indirect taxation increased slightly between 1960-72 and 1976-84, but for some countries the changes between periods were substantially higher than the average. Indirect taxation fell sharply in Chile and Korea and increased dramatically in the group of extreme discriminators-C6te d'Ivoire, Ghana, and Zambia (see table 2-3). For direct intervention, the largest change was the increase in protection of agriculture in Korea. Various factors influenced changes in the levels of direct and indi- rect taxation or protection. Some countries, for example, Chile after 1973 and Ghana after 1983; reduced direct price intervention in agri- culture as part of a general reform of trade and macroeconomic poli- cies. In some cases increased intervention in agriculture accompanied liberalization, as in Korea, where growth was led by an outward- looking industrial policy while agriculture was increasingly protected through direct interventions. In other cases, direct protection of agri- culture was increased to compensate for increases in indirect taxation, as in Ghana in the early 1980s. Changes in government and political ideology also influenced the extent of direct intervention and govern- ment control of agriculture in some countries, such as Pakistan and Chile. And in some countries, budgetary considerations drove the change in price policies, as in Sri Lanka, which eliminated rice sub- sidies following the 1978 fiscal reform, and in Thailand, which gradu- ally reduced the rice export tax. (These and other political-economy issues are discussed in detail in volume 5 of this study.) Appendix 2-1. Measures of Intervention This appendix describes the construction of the various measures of intervention used to estimate direct and indirect rates of protection, 30 Historical Patterns of Price Intervention or taxation. It also shows how the equilibrium real exchange rate (the real exchange rate that would have kept the current account deficit at a sustainable level, taking into account normal capital flows, in the absence of all trade interventions) was derived and the equivalence in using nominal and real exchange rates to calculate the direct and indirect rates of protection. Finally, it shows that the indirect protec- tion rate is a weighted average of the equivalent tariff and the exchange rate overvaluation. Procedures The impact of intervention was obtained by measuring producer and consumer prices relative to nonagricultural prices in each country and comparing these relative prices to relative prices without interven- tions. For all tradable commodities, the reference prices were the border prices that would have prevailed under an intervention-free regime. For direct interventions, the border price was measured at the offi- cial exchange rate, and interventions in the nonagricultural sector were not considered. For most countries, which are price takers for the products they import or export, we used the actual border price. For countries with market power in some commodities, such as cocoa in Ghana, we adjusted the border price accordingly. To estimate direct rates of protection, we gathered estimates of domestic producer, consumer, and border prices and adjusted them for transport costs, storage costs, quality differences, and other ele- ments of their marketing margins. For instance, for wheat in Chile (an importable), we adjusted prices for customs duties and fees, trans- portation costs from the main port of entry to the mills, unloading costs and losses in transit, annual average quality differences between domestic and imported wheat, and seasonality (storage). Obtaining reliable price estimates was not always straightforward. Estimating marketing margins was particularly difficult when a gov- ernment marketing board had a monopoly or a monopsony on distri- bution of a product. Where marketing board costs constituted a sig- nificant share of the border price of exportables, producer prices were often only a small fraction of the border price. In Ghana after 1976, transport costs and other margins were so high compared to the border price, because of high inflation and an overvalued currency, that adjusting the border price for inflated transport costs yielded negative estimates of producer prices in the absence of intervention. For that reason, we exclude results for those years in our analysis in most cases. Indirect taxation became so high after 1976-with a bud- get deficit of 127 percent of government revenues in 1978, inflation of 116 percent in 1977, and an increasingly overvalued currency-that The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 31 direct protection was given to some products in partial compensation (Stryker 1991). Estimating the effect of indirect interventions involved additional steps. We needed to calculate the equilibrium real exchange rate and the tax on agricultural production implicit in protection to industry. This involved estimating equivalent tariffs for import protection, for- eign exchange demand and supply elasticities, and the change in the real exchange rate needed to yield a sustainable current account deficit. First, we calculated the depreciation of the real exchange rate required to eliminate the nonsustainable part of the current account deficit and the depreciation that would result from the removal of trade interventions. This gave us the equilibrium real exchange rate.11 We then used these estimates to derive the rate of indirect protection, or taxation, of agriculture. We converted the border price in dollars for each commodity at the equilibrium exchange rate to get an esti- mate of the border price in local currency in the absence of interven- tions. Then we estimated a new price index for the nonagricultural sector in the absence of interventions by measuring prices at the equilibrium exchange rate and in the absence of trade interventions. Using these calculations, we estimated the indirect effect of interven- tions on the price (and value added) of agricultural products relative to the nonagricultural price index. This gave us an estimate of the change in agricultural prices relative to nonagricultural prices that would occur if all interventions were removed. For instance, in the case of Argentina, the average equivalent tariff over the period 1960-84 was 65 percent, resulting in a tax on agricul- ture relative to other tradables of almost 40 percent. Overvaluation of the real exchange rate taxed agriculture relative to the nontradable sector by close to 18 percent. Together, these policies resulted in a tax on agriculture relative to the nonagricultural sector of about 21 per- cent, a weighted average of the effects of industrial protection and overvaluation of the exchange rate. Measures of Intervention Our measures of intervention were calculated as follows. The direct nominal protection rate is PNA PNA Pi (2A-1) NPRV = - _ - 1, wNA where NPRD measures the effect of price controls, export taxes or 32 Historical Patterns of Price Intervention quotas, and other sectoral policies on the domestic producer price (P) of tradable agricultural product i (Pi); Pi'is the border price of product i (FiB) converted into local currency at the official nominal exchange rate Eo and adjusted for transport, storage and other costs, and qual- ity differences, so that P' = 1iB Eo; P? is the border price of product i at the equilibrium nominal exchange rate E* and adjusted as before, so that Pf = liBE * = P' E */Eo; PNA is the nonagricultural sector price index, which consists of a tradable share a with a price PNAT and of a nontradable share 1 - a with a price PNAH, so that PNA = aPNAT + (1 - a) PNAH; and Pk is the nonagricultural sector price index, where the price index of the tradable part is evaluated at E* and in the absence of trade interventions (tNA) affecting nonagricultural trad- ables, so that PkA = aPNAT [E*I(1 + tNA)Eo] + (1 - a)PNAH- The indirect nominal protection rate is Pl Pi- P; PNA PNA PNA NPRI = --- 1 P- Pfk (2A-2) Pi ENA _1 __NA P,*AEO -1 E* Pi PNAE* E0 P* It measures the effect of the difference between the nominal exchange rate EO and the equilibrium exchange rate E* and of the effect of trade policy on PNAT* NPRl is the same for all tradable products since Pi does not appear in equation 2A-2. The total nominal protection rate is Pi _Pt~ Pi PNA PNA PNA (2A-3) NPRT ----1. P* P* ~NA N Because the denominator of equation 2A-1 differs from that of equations 2A-2 and 2A-3, NPRD + NPRI = NPRT. To make the three measures comparable, we define another direct protection rate Pi P? (2A-4) nprD - NA NA PA P*l The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 33 which measures the impact of direct policies (Pi/PNA - PiIPNA) as a percentage of the relative price that would prevail in the absence of all interventions (Pr/PA), the same denominator used in equations 2A-1 and 2A-2. Then, nprD + NPRI = NPRT. An illustration is provided in note 2. Estimates using these measures are presented in tables 2-1 to 2-4. Calculating NPRI and NPRT involves adjustments in the nominal exchange rate. In the following section, we show that the results are the same when the real exchange rate is used (as was done for the country studies in calculating protection rates). The Equilibrium Exchange Rate We assume a three-goods economy with an exportable X, an import- able M, and a nontradable H, with prices Px, PM, and PH, respectively. We also assume a domestic and a foreign currency, with relative price E, the nominal exchange rate, defined as the domestic currency price of the foreign currency. We define the real exchange rate e as the ratio of the nominal exchange rate to the price of the nontradable H: E (2A-5) e == - PH The foreign prices of X and M do not enter into the definition of e because, for a small country in the world market, these prices are given and are not affected by policy changes. We are interested in the change in e that occurs with the removal of interventions and the unsustainable part of the current account defi- cit. For countries where removal of policy interventions affects world prices, that effect was taken into account. We assume that both the demand for (QD) and supply of (Qs) foreign exchange are functions of the real exchange rate, with elas- ticities -ED and es, respectively. Then the proportional increase in the real exchange rate needed to eliminate the unsustainable part of the current account deficit (AQo) is calculated as follows: (2A-6) e = Q0 + 1 eo =esfQs + C-DQD where QS and QD are the quantities of foreign exchange supplied and demanded at the prevailing real exchange rate eo, and el is the real exchange rate that balances the unsustainable part of the current account deficit. Eliminating the tariff equivalent on the importable good (tM) and 34 Historical Patterns of Price Intervention the export tax on the exportable good (tx) leads to an increase in excess demand for foreign exchange of AQj, where (2A-7) AQ1 = tM QDED 0. 1 + t 1 -txQ~ < If we define the equilibrium real exchange rate e* as the real exchange rate that meets the condition AQO = tx = tm = 0, then (2A-8) e* = AQO+AQi)e Ws, + 'EDQDIO and AQ, is defined as in equation 2A-7. Nominal versus Real Exchange Rate in Calculating Rates of Protection The solution of the model of exchange rate determination is the equi- librium real exchange rate e* rather than the nominal rate E * used to solve for NPRI and NPRT in equations 2A-2 and 2A-3, where e - EIPH. The nontradable sector H is assumed to consist almost entirely of nonagricultural goods and services NAH, and therefore e = E/PNAH- Assume that tx measures the impact of direct interventions on Pi (t. - 0). Then Pi_ PBEo(1 - tx) PNA aPNAT + (1 - COPNAH (2A-9) =P7Eo(l - t.) aPNATEO(1 + tNA) + (1 - CONAH (PNAH) UPB Eo (1 + tNA + NAH H) a) or (2A-10) _ pe(1-t.) PNA apNAT(1 + tNA)eO + (1 - CO Then (2A-11) p1 pe PNA C'PNAT(1 + tNA)eO + (1 The Effects of Intervention on Agricultural Prices: Magnitude and Patterns 35 and Pfk PBe (2A-12) = PN aPNATe + (1 - As these equations show, to derive NPRI and NPRT it is sufficient to know eo and to obtain e*; information on E* and PkAH is not needed. The Indirect Rate of Protection as a Weighted Average of the Rates of Industrial Protection and Exchange Rate Overvaluation Substituting the definitions of PNA and P* for PNA and P* in equa- tion 2A-2 for NPRI, we obtain NPRI = - 1 PNAE* (aPNATE/Eo + (1 - ) PA (2A-13) 1= N ,+( - 1)PA E -PNAT + (1 - a) PNAH APNAT + (1 a)PNAH 1+ tNA H E. 1 CePNAT + (1 -CO)NAH - or (2A-14) NPRI = (a'1 + (1 - ) 1, I+ tNAE* where (2A-15) a = aPNAT aPNAT + (1 - a)PNAH From equation 2A-14, we see that 1 + NPRI (or NPCI) is a weighted average of 1/(1 + tNA)-the tax on agriculture relative to non- agricultural tradables due to industrial protection-and of EolE*-the tax on agriculture relative to nontradables due to overvaluation of the exchange rate. The Effects of Intervention on Price Variability Because world commodity prices are so volatile, governments com- monly intervene to stabilize domestic producer and consumer prices of agricultural products. In our sample countries, such price interven- tions, particularly direct price interventions, have generally achieved this goal. Indirect price interventions, which were the main source of agricultural taxation, contributed little to price stability. On balance these findings do little to suggest that the harmful effects of indirect interventions on agricultural incentives were mitigated by any sub- stantial beneficial effects on price variability. Direct price interven- tions protected importables and staples and taxed exportables, on average, and also stabilized the prices of importables and staples more than the prices of exportables and nonstaple products. Hence, importables and staples were twice-favored by direct interventions relative to exportables. We also found, as expected, that prices were stabilized more during periods of high volatility in world prices than at other times, but we found no tendency to stabilize prices more (or less) during periods of high prices than during periods of low prices. Arguments and Issues in Price Stabilization Most developing countries lack the risk-diffusion mechanisms needed to help agricultural producers adapt to highly volatile world prices. As a consequence, many governments intervene to stabilize domestic producer and consumer prices and thereby to stabilize income, improve production efficiency, reduce macroeconomic insta- bility, and sometimes to redistribute income among the rural, urban, and government sectors. The macroeconomic argument for price stabilization is that fluctua- tions in the world price of agricultural exports will lead to fluctuations in export revenues and so to fluctuations in government revenue, the real exchange rate, wages and prices, and other macroeconomic vari- ables. Stabilizing the domestic price of important exportables (say, 36 The Effects of Intervention on Price Variability 37 through a stabilization fund) may partly insulate the economy from external shocks and so reduce macroeconomic instability. In the case of the microeconomic arguments for price stabilization, it is important to distinguish between the impact of price stabilization on urban consumers and on farmers and rural consumers, and to distinguish between the effects of stabilization on prices, nominal income, real income, and consumption. Stabilizing the price of sta- ples with a large share in the urban consumer budget in order to prevent large fluctuations in urban consumers' real income is a politi- cal imperative in many developing countries. Stabilizing the prices of food products also tends to stabilize the real income of rural con- sumers, such as landless laborers. If the payment for their labor closely approximates their consumption patterns (in terms of food, say, or if they receive free housing and a plot to grow food), however, the stability of their real income will be less dependent on the price of food. For farmers there are two principal arguments for stabilizing prices: reducing rural income fluctuations and improving production effi- ciency. What matters to risk-averse farmers is stabilization of general consumption (or utility) over time. Traditionally, through output diversification and liquidity, farmers have been able to stabilize con- sumption partially, but at a cost of reduced average consumption.' The question is whether the government can provide stabilization at a lower social cost-that is, more efficiently-than the private sector. Many developing countries lack private insurance mechanisms against price risk (such as futures markets or forward contracts), or such mechanisms are not generally available to small farmers. The governments in those countries may be better able to pool risks, so shifting some of the price risk to the government may be socially optimal (see Vald6s and Siamwalla 1988). But the response of farmers to government intervention-or the substitutability of private and public stabilization effort-should also be considered. Rural Income Stabilization Several factors affect the first objective, reducing rural income fluctu- ations. Generally, price stabilization is more effective and less costly in stabilizing nominal producer income for tradable products than for nontradable products. If prices vary mainly because of domestic sup- ply shocks, as is the case for nontradable products (or tradable prod- ucts for which the country has market power on the world market), output and price will be negatively correlated. In that case, price fluctuations may help to stabilize income, and price stabilization poli- cies may increase income variability. If price variability arises entirely from supply shocks and if the price elasticity of demand equals 1 (in 38 Historical Patterns of Price Intervention absolute value), then nominal income is constant and price stabiliza- tion will increase income variability. In fact, as long as the elasticity of demand is larger than 0.5, price stabilization will increase income variability (see appendix 3-1). For staples, however, the price elastic- ity of demand is generally smaller than 0.5 (see Knudsen and Nash 1990), so price stabilization policies should help to stabilize income, but not as much as prices. (How well policies are implemented and who actually benefits are other issues entirely.2) If price variability is independent of domestic supply shocks, as for a tradable product with a small share on the world market, price stabilization will clearly stabilize producer income as long as there are no quantitative restric- tions on trade. The costs of price stabilization are also lower for tradables than for nontradables. The price of tradable products with a small share in the world market (such as sugar or wheat) can be stabilized through import or export taxes and subsidies, which should not be too costly if the price is stabilized at the mean. The buffer stocks needed to stabi- lize nontradable products (such as white maize in East Africa) may involve large costs if substantial price stabilization is intended. A review of price stabilization schemes for tradables (Knudsen and Nash 1990) concluded that variable tariffs and subsidies are generally more effective and less costly than buffer stocks managed by state marketing boards. Another study (Newbery and Stiglitz 1981) also found the case for using public buffer stocks to be generally weak.3 Production Efficiency The second argument for price stabilization for farmers is that exces- sive price variability, if nonsystematic, increases uncertainty and reduces production efficiency. When future returns are relatively sta- ble and predictable,4 cropping decisions (for example, what and how much to plant, how intensively to cultivate) and investment choices lead to more efficient outcomes. When uncertainty is high, farmers are reluctant to adopt more profitable but riskier new techniques, they tend to diversify their output and so to lose the potential gains from specialization. Farmers may also try to stabilize their income by working as wage earners in other occupations, and they may hold their assets in more liquid form. These adjustments to risk, although privately optimal, may imply high social costs. Hence, the benefits of price stabilization in terms of better allocation of resources will be higher when farmers have a greater ability to adjust to price uncer- tainty. When farmers are unable to diversify output or to find other occupations, the efficiency gain of price stabilization will be limited, although the private benefits may be very large. Although stabilization is expected to improve efficiency, it may also The Effects of Intervention on Price Variability 39 entail efficiency costs of its own. Stabilization policies that prevent prices from varying within a crop year or across regions within a country (the practice in a number of Sub-Saharan African countries) result in significant inefficiencies in resource allocation and prevent the development of private storage and transport activities. And poli- cies that attempt to stabilize even the predictable component of price behavior rather than only the random component from one year to the next may also entail efficiency costs, such as forgone output when the world price is high or excessive output when the world price is low. Such a policy removes or reduces producers' incentives to adjust their plans to the expected change in world prices in order to capture profit opportunities. Also, when stabilization is effected through mar- keting boards that hold buffer stocks, storage costs can be substantial. In general, then, producer price stabilization is more beneficial in the case of tradable products that are important in total net trade (because of the contribution to macroeconomic stability), that are mostly sold off the farm (the higher on-farm consumption, the lower the effect of a price change on real income), and for which the country is a price taker. On the other hand, the efficiency gains are lower when the effects of price variability are harder to avoid-where resources do not allow much diversification, and labor mobility and access to credit are limited. In a simple model, Mirrlees (1988) showed that the optimal price stabilization scheme is a tax/subsidy price insurance system and that the producer price is a positive function of the world price, optimal price stabilization does not imply complete elimination of price trans- mission from the world to the domestic market (see chapter 10). He also showed that the degree of price stabilization rises as price uncer- tainty increases. These findings are consistent with the empirical results that follow. Measures of Price Variability Setting aside the merits of the various arguments for stabilizing agri- cultural prices and of the various methods of doing so, here we address the actual effects of interventions on price variability, looking at five issues. First, were direct and indirect agricultural price inter- ventions actually successful in stabilizing domestic producer and con- sumer prices? Second, did direct and indirect interventions have dif- ferent effects on price variability? Third, did the degree of price stabilization vary by product group? Fourth, was the degree of stabi- lization different when world prices rose and when they fell? And fifth, did the degree of price stabilization differ in the 1960s and in the 1970s and 1980s, when world prices were more volatile? To answer these questions, we used three measures of price vari- 40 Historical Patterns of Price Intervention ability: the sample standard deviation, the coefficient of variation, and the Z-statistic. Each has its merits as a measure of variability. The standard deviation measures the degree to which a variable is dis- persed around its mean value. The coefficient of variation-the stan- dard deviation divided by the sample mean-standardizes the scale of the variable, so that the degree of dispersion of variables with widely differing average values can be compared. The Z-statistic, which is a measure of the average deviation of the price from its value in the preceding period (annual price change), is probably a better indicator of producers' concerns-farmers are more likely to be concerned with price deviations from previous year prices than with deviations from the sample mean. The standard deviation and the Z-statistic are directly comparable, one measuring the deviation from the sample mean and the other the deviation from the value in the last period. (See appendix 3-2 for a definition of the Z-statistic.) We calculated the sample standard deviation for the actual price of each category of agricultural products in relation to the price of non- agricultural products (the actual relative price) and did the same for the relative price in the absence of direct interventions in agricultural prices. For the group of products as a whole, we also calculated the standard deviation of the relative price in the absence of total price interventions. Then we did the same for the other two measures-the coefficient of variation and the Z-statistic. For the effect of direct interventions, we used the ratio of the value of each statistic for the actual relative producer price to the value for the relative price in the absence of direct interventions. For the effect of total interventions, we used the ratio of the value of each statistic for the actual relative producer price to the value for the relative price in the absence of total interventions. Because the three price ratios (actual relative price, relative price in the absence of direct intervention, and relative price in the absence of total interventions) exhibited significant trends in most countries, we also analyzed detrended prices. According to most measures, we found that price interventions reduced price variability in the majority of countries. For producer prices Korea is the only exception for the group of all products exam- ined. For consumer prices the main exceptions are Ghana and Korea, where most measures indicate an increase in price variability, and Portugal and Sri Lanka, where the increase results mainly from direct price interventions. The change in price volatility varied according to the measure used, the product category, and the type of price intervention. In most cases, however, the ratios of the (producer to world price) standard deviation, coefficient of variation, and annual price change statistic were smaller than 1, implying that direct and total interventions The Effects of Intervention on Price Variability 41 dampened the transmission of international price variability to domestic prices. The reduction in price variability was stronger for staples than for nonstaple products and stronger for importables than for exportables. Direct interventions contributed the bulk of the reduction in price variability, and indirect interventions had a minor effect on price variability. The results also held for the variability of prices around their trend. Producer Price Variability Effectiveness of Stabilization Efforts On average for all countries examined, direct and total interventions stabilized domestic producer prices relative to world prices in each commodity group-importables, exportables, staples, and nonstaples (table 3-1). The standard deviation showed a reduction in variability of more than 30 percent in all product categories for both direct and total interventions, except nonstaples, where the reduction through direct interventions was only 10 percent. The coefficient of variation and the annual price change (Z-statistic) ratios also showed the impact of direct and total interventions in reducing price variability. These results also hold for countries in Groups I, II, and III but, because of the impact of price interventions in Korea, not for Group IV (protectors). Direct and total interventions have tended to reduce annual price variations as measured by the Z-statistic more than they did price variations around the mean (as measured by the standard deviation). An extreme example is that of Thailand, where direct interventions reduced domestic price variability around the mean by 53 percent but reduced the annual price variation by 98 percent. The reason for this large difference is explained in Appendix 3-2. It seems likely that farmers are more concerned with annual price changes than with price variations around a twenty-five-year average. If so, then the larger reduction in annual price variations (over 40 percent for direct interventions and over 50 percent for total interventions on average) than in variations around the mean would indicate some success in reducing the price risks of special concern to producers. These results hold for Groups I, II, and IV but not for Group III (mild discrimina- tors). The latter is essentially due to opposite results obtained for Brazil. Since the coefficient of variation measure is defined as the standard deviation divided by the mean, it will show a smaller reduction in price variability than the standard deviation measure when protec- tion rates are negative and a larger reduction when protection rates are positive. Given that direct and total rates of protection were both Table 3-1. Relative Producer Price Variability: Actual and Counterfactual Direct All Total all Country Period Statistics products Import Export Staples Nonstaples products Cote d'Ivoire 1960-82 SDP/SDB 0.40 1.31 0.34 1.31 0.34 0.20 CVP/CVB 0.63 1.02 0.57 1.02 0.57 0.42 ZP/ZB 0.36 1.09 0.31 1.09 0.31 0.20 Ghana 1960-78 SDP/SDB 0.32 0.48 0.38 1.38 0.33 0.24 CVP/CVB 0.57 0.48 0.75 0.94 0.59 0.62 ZP/ZB 0.12 0.33 0.22 1.02 0.16 0.08 Zambia 1966-84 SDP/SDB 0.71 0.56 0.99 0.56 0.99 0.19 CVP/CVB 0.92 0.76 1.04 0.76 1.04 0.41 ZP/ZB 0.51 0.31 0.89 0.31 0.89 0.24 Group I averages SDP/SDB 0.48 0.78 0.57 1.08 0.55 0.21 CVP/CVB 0.71 0.75 0.79 0.90 0.73 0.48 ZP/ZB 0.33 0.57 0.47 0.80 0.45 0.17 Argentina 1960-84 SDP/SDB 0.77 n.a. 0.77 0.61 0.89 0.70 CVP/CVB 1.01 n.a. 1.01 0.82 1.10 1.17 ZP/ZB 0.81 n.a. 0.81 0.65 0.91 0.92 Colombia 1960-83 SDP/SDB 0.49 1.03 0.61 0.53 0.93 0.41 CVP/CVB 0.55 0.88 0.72 0.51 1.02 0.61 ZP/ZB 0.48 0.58 0.51 0.44 0.95 0.36 Dominican Republic 1966-85 SDPISDB 0.98 0.66 0.92 0.31 0.84 0.81 CVP/CVB 1.30 0.58 1.35 0.45 1.00 1.36 ZP/ZB 0.89 0.31 0.79 0.34 0.62 0.76 Egypt 1964-84 SDP/SDB 0.31 0.59 0.34 0.20 0.41 0.35 CVP/CVB 0.48 0.66 0.62 0.36 0.59 0.68 ZP/ZB 0.23 0.53 0.22 0.11 0.32 0.24 Morocco 1963-84 SDP/SDB 0.75 0.48 1.14 0.75 n.a. 0.74 CVP/CVB 0.94 0.64 1.50 0.94 n.a. 1.12 ZPIZB 0.82 0.52 1.21 0.82 n.a. 0.73 Pakistan 1968-86 SDP/SDB 0.25 0.19 0.37 0.21 0.57 0.20 CVPICVB 0.35 0.25 0.52 0.29 0.75 0.40 ZP/ZB 0.31 0.29 0.41 0.25 0.50 0.18 Philippines 1960-86 SDP/SDB 0.47 0.77 0.57 0.36 0.92 0.37 CVP/CVB 0.52 0.34 0.69 0.41 0.95 0.54 ZP/ZB 0.61 0.33 0.75 0.31 1.05 0.48 Sri Lanka 1960-85 SDP/SDB 0.56 0.67 0.60 0.67 0.38 0.34 CVP/CVB 0.66 0.48 0.85 0.48 0.55 0.60 ZP/ZB 0.36 0.59 0.60 0.59 0.35 0.29 Thailand 1962-83 SDP/SDB 0.47 n.a. 0.47 0.43 0.80 0.45 CVP/CVB 0.70 n.a. 0.70 0.67 0.90 0.78 ZP/ZB 0.02 n.a. 0.02 0.02 0.34 0.02 Turkey, 1961-83 SDP/SDB 0.58 0.57 0.58 0.56 0.62 0.37 CVP/CVB 0.58 0.55 0.58 0.53 0.64 0.59 ZP/ZB 0.63 0.25 0.61 0.23 0.67 0.39 Group II averages SDP/SDB 0.56 0.62 0.64 0.46 0.71 0.48 CVP/CVB 0.71 0.55 0.85 0.55 0.83 0.79 ZP/ZB 0.52 0.42 0.59 0.38 0.63 0.44 Source: Authors' calculations based on data provided in country studies. (Table continues on the following page.) Table 3-1. (continued) Direct All Total all Country Period Statistics products Import Export Staples Nonstaples products Brazil 1969-83 SDP/SDB 0.56 0.52 0.71 0.52 0.71 0.42 CVP/CVB 0.52 0.45 0.68 0.45 0.68 0.49 ZP/ZB 0.83 0.69 1.03 0.69 1.08 0.64 Chile 1960-83 SDP/SDB 0.71 0.70 1.32 0.70 1.32 0.40 CVP/CVB 0.73 0.72 1.13 0.72 1.13 0.53 ZPIZB 0.72 0.71 1.14 0.71 1.14 0.71 Malaysia 1960-83 SDP/SDB 0.89 0.52 0.90 0.52 0.90 0.83 CVP/CVB 1.00 0.44 1.04 0.44 1.04 1.01 ZP/ZB 0.78 0.39 0.79 0.39 0.79 0.77 Group III averages SDP/SDB 0.72 0.58 0.97 0.58 0.97 0.55 CVPlCVB 0.75 0.54 0.95 0.54 0.95 0.68 ZP/ZB 0.78 0.60 0.99 0.60 1.01 0.71 Korea, Rep. of 1960-84 SDP/SDB 2.05 2.05 n.a. 2.00 3.30 2.29 CVP/CVB 1.34 1.34 n.a. 1.32 1.24 2.02 ZP/ZB 0.95 0.95 n.a. 0.89 2.52 0.69 Portugalb 1960-83 SDP/SDB 0.94 0.53 1.00 0.68 1.04 0.67 CVP/CVB 0.87 0.46 0.99 0.59 1.00 0.68 ZP/ZB 0.99 0.72 0.99 0.66 1.05 0.72 Group IV Averages SDP/SDB 1.49 1.29 1.00 1.34 2.17 1.48 CVP/CVB 1.10 0.90 0.99 0.95 1.12 1.35 ZP/ZB 0.97 0.83 0.99 0.78 1.79 0.71 Total Average SDP/SDB 0.68 0.64 0.68 0.69 0.90 0.56 CVP/CVB 0.76 0.56 0.83 0.66 0.87 0.78 ZP/ZB 0.58 0.48 0.66 0.54 0.80 0.47 a. Mutton is included in nonstaples. b. Tomatoes are included in staples. c. SD = Standard deviation, CV = coefficient of variation, Z Z-statistic (explained in appendix 3-2), P = producer price, B = border price. Source: Authors' calculation based on data provided in country studies. 46 Historical Patterns of Price Intervention negative on average, the reduction in price variability was less in both cases as measured by the coefficient of variation (on average for all countries examined, 24 percent for direct and 22 percent for total) than by the standard deviation (32 percent and 44 percent). For importables, which were protected by direct interventions, the reduc- tion in variability is larger with the coefficient of variation measure (44 percent) then with the standard deviation (36 percent). Another way to examine the relation between domestic and border price variability is by estimating a behavioral model (see chapter 9). The model's results are consistent with those presented here. For instance, in the case of exportables, the model indicates an average short-run transmission to domestic prices of about 68 percent of the shocks in border prices. Table 3-1 shows that the ratio of domestic to border price variability when measured by the standard deviation is also 68 percent; it is 66 percent when measured by the Z-statistic and 83 percent when measured by the coefficient of variation. For individual countries and product categories, we judged inter- vention to have stabilized prices whenever at least two of the three ratios for our (producer to world price) variability measures are smaller than 1. This occurs for all product categories and types of intervention in two-thirds of the countries: Argentina, Brazil, Col- ombia, the Dominican Republic, Egypt, Malaysia, Pakistan, the Phi- lippines, Sri Lanka, Thailand, Turkey, and Zambia (table 3-1). For Argentina the ratios for the standard deviation and annual price change statistic are smaller than 1, but the coefficient of variation ratio is larger than 1 for some product categories. This occurs because the high rate of agricultural taxation significantly lowers the producer price compared to the world price, so the coefficient of variation is larger for the producer price than for the world price. Other countries had at least one product category for which prices were not stabilized, generally through the effect of direct interven- tions. In Chile that category was exports (and nonstaples, which were the same product group, consisting of apples and grapes), as it was in Morocco (barley). In Portugal it was nonstaples, which had standard deviation and annual price change (Z-statistic) ratios indicating a slight increase in price variability. In C6te d'Ivoire, price variability was reduced by 37 to 80 percent, except for rice (an import and a staple), which had a standard devia- tion ratio of 1.31 for direct price interventions; the coefficient of varia- tion ratio was close to 1, however, because rice was highly protected. Direct intervention had a similar impact on the price variability of rice (a staple) in Ghana. In Korea the standard deviation ratios are all 2 or greater. The coefficient of variation ratios are lower because of high protection rates, but they are still larger than 1, indicating increased price vari- ability. When prices are detrended, however, the coefficient of varia- The Effects of Intervention on Price Variability 47 tion ratio shows a slight reduction in variability from direct price interventions (from 1.34 to 0.94). The ratios of the annual price change statistic are smaller than 1 except for the nonstaple products, barley and soybeans. The standard deviation and the annual price change ratios differ because producer prices experienced a few large changes but fewer changes overall than world prices, resulting in a relatively larger standard deviation for the producer price, but a rela- tively lower value for the annual price change statistic (see appendix 3-2). Effects of Direct and Indirect Interventions Total interventions reduced price variability more than did direct interventions when measured by the standard deviation and the annual price change statistic but not when measured by the coeffi- cient of variation (table 3-1). Direct interventions had a far larger impact on price variability then did indirect interventions, however. Adding the effects of indirect interventions to those of direct inter- ventions somewhat reduces price variability as measured by the stan- dard deviation (from a 32 to a 44 percent reduction in variability) and by the annual price change statistic (from a 42 to a 53 percent reduc- tion) but increases it slightly as measured by the coefficient of varia- tion (from a 24 to a 22 percent reduction). The standard deviation shows an increase in price stability from indirect interventions while the coefficient of variation does not because indirect interventions significantly taxed agriculture. Indirect interventions led to a lower standard deviation but also to a lower level of agricultural prices, so that little change occurred in the coeffi- cient of variation. In other words, indirect interventions reduced incentives and price variability equiproportionally (the latter being measured by the standard deviation). Direct interventions reduced price variability (as measured by the standard deviation) more than incentives, however, so the coefficient of variation fell. That indirect intervention did not substantially improve price sta- bility should not be surprising. Real exchange rate overvaluation resulting from expansive macroeconomic policies (such as large bud- get deficits) is usually accompanied by high and variable inflation rates and high relative price variability. Moreover, the main compo- nent of indirect intervention is industrial protection (see chapter 2), and there is no reason to expect industrial protection policies to reduce relative price variability. Also, the world prices of industrial goods are much less volatile than those of agricultural products. In fact, we found that in most countries indirect interventions resulted in an increase in price variability as measured by the coefficient of variation. In Group I, indirect interventions did significantly reduce price 48 Historical Patterns of Price Intervention variability. This essentially results from the case of Zambia, the only country where indirect interventions had a stronger impact in stabi- lizing prices than did direct interventions. In several cases indirect interventions led to an increase in price variability. These include Egypt, Group II (representative discriminators) on average (regarding the coefficient of variation), Korea (regarding the standard deviation and the coefficient of variation), and Group IV (protectors) on aver- age (regarding the coefficient of variation). Another interesting result emerging from the model in chapter 9 is that the degree of transmission of shocks from the border price to the domestic price is inversely related to the degree of real exchange rate overvaluation. That is, the degree of price stabilization resulting from direct interventions increases as the real exchange rate overvaluation increases. Recall that according to our definition of equilibrium real exchange rate, overvaluation may result from either macroeconomic policies or industrial protection policies. The relationship is highly significant statistically (at the 1 percent level), although it is some- what weaker empirically. For instance, in the case of exportables, when the real exchange rate (RER) is in equilibrium, the coefficient of transmission is 0.72, and the degree of stabilization resulting from direct intervention is 28 percent. When the RER overvaluation is 30 percent (60 percent) the coefficient of transmission falls to 0.68 (0.64), and the degree of stabilization rises to 32 percent (36 percent). The reasons for this relationship are examined in chapter 9. Price Stabilization by Product Category Direct interventions reduced price variability considerably more for staples than for nonstaples-by 31 percent compared with 10 percent for the standard deviation measure, 34 percent compared with 13 percent for the coefficient of variation, and 46 percent compared with 20 percent for the annual price change statistic-and more for import- ables than for exportables-by 36 percent compared with 32 percent for the standard deviation measure, 44 percent compared with 17 percent for the coefficient of variation, and 51 percent compared with 33 percent for the annual price change statistic. Since both consumers and producers have an interest in price sta- bility for staples, but price stability for nonstaples is of more concern to producers, it is not surprising that direct price interventions stabi- lized staples more than nonstaples. The same reasoning applies to importables compared to exportables, since importables consist mostly of staples (twenty-seven of thirty-four in our sample) and exportables consist mostly of nonstaples (thirty-two of forty-five in our sample). Overall, then, producers of importables (and staples) benefited twice compared with producers of exportables, since pro- The Effects of Intervention on Price Variability 49 ducer prices for importables were both more stable and more highly protected by direct intervention (see chapter 2). These results remain fundamentally the same when detrended prices are used because the trend of the actual producer price was close to the trend of the price in the absence of direct and total inter- ventions in many countries. For instance, with detrended prices, direct interventions reduced price variability by 45 percent for import- ables and 19 percent for exportables compared to 44 and 17 percent, respectively, for nondetrended prices (using the coefficient of varia- tion measure). The results were also similar for individual countries in most cases. Effects of Commodity Booms and Busts Did the authorities try to stabilize prices more (or less) during com- modity price booms than during busts? For both producer and con- sumer prices, one might expect different responses during booms and busts. When world prices rise, governments may prefer to cap- ture the increase instead of passing it on to producers. And when world prices fall, producers may oppose efforts to lower prices, or governments, expecting world prices to rise again, may try to main- tain prices until that happens. In C8te d'Ivoire, the government kept the producer price of coffee steady for several seasons after the col- lapse of the world price in the mid-1980s before reducing it. A sym- metrical analysis applies for consumer prices. When prices fall, gov- ernments might be tempted to boost their revenues by capturing the price difference instead of passing the lower prices on to consumers. And when world prices rise, the authorities may be reluctant to pass on all or even part of the increase to consumers because of strong opposition; attempts to allow prices to rise have led to riots in several countries, including Egypt, Morocco, and Zambia. And, indeed, our analysis in chapter 2 showed that direct protection rates fell during the world boom in commodity prices in the mid-1970s, suggesting that governments were trying to avoid passing on large price increases. To test whether producer price stabilization differed during booms and busts,we ran a regression of the annual change (the first differ- ence) in the producer price on the annual change in the world price and on a dummy variable for price rises and declines. The coefficient of the dummy variable was found not to be significantly different from 0, whether empirically or statistically. These results held both for direct and total interventions. Hence, our findings imply that the stabilizing effect of price interventions were not significantly different when world prices rose than when they fell. Table 3-2. Coefficients of Variation of Domestic Relative Producer Prices and Border Prices, 1960-70 and 1971-84 1960-70 1971-84 Domestic Border Ratio Domestic Border Ratio Countries Period price (1) price (2) (1)/(2) price (4) price (5) (4)/(5) C6te d'Ivoire 1960-70/1971-82 0.10 0.16 0.63 0.18 0.34 0.54 Ghana 1958-70/1971-76 0.17 0.32 0.55 0.03 0.22 0.13 Zambia 1966-70/1971-84 0.15 0.25 0.59 0.20 0.21 0.97 Group I average 0.14 0.24 0.59 0.14 0.25 0.55 Argentina 1960-70/1971-84 0.29 0.28 1.04 0.24 0.31 0.77 Colombia 1960-70/1971-83 0.06 0.08 0.73 0.24 0.40 0.60 Dominican Rep. 1966-70/1971-85 0.13 0.10 1.25 0.51 0.39 1.31 Egypt 1964-70/1971-84 0.11 0.14 0.76 0.16 0.36 0.43 Morocco 1960-70/1971-83 0.27 0.15 1.82 0.16 0.26 0.62 Pakistan 1968-70/1971-86 0.04 0.12 0.30 0.11 0.19 0.56 Philippines 1960-70/1971-86 0.07 0.16 0.41 0.24 0.36 0.68 Sri Lanka 1960-70/1971-85 0.13 0.13 1.00 0.24 0.35 0.67 Thailand 1962-70/1971-83 0.09 0.09 1.09 0.15 0.25 0.58 Turkey 1961-70/1971-83 0.07 0.16 0.45 0.37 0.58 0.63 Group II average 0.13 0.14 0.89 0.24 0.35 0.69 Brazil 1966-70/1971-83 0.18 0.31 0.59 0.13 0.23 0.53 Chile 1960-70/1971-83 0.05 0.07 0.75 0.21 0.30 0.69 Malaysia 1960-70/1971-83 0.30 0.30 0.99 0.16 0.22 0.71 Group III average 0.18 0.23 0.78 0.16 0.25 0.64 Korea, Rep. of 1960-70/1971-84 0.18 0.78 0.78 0.28 0.27 1.04 Portugal 1960-70/1971-83 0.08 1.03 1.03 0.15 0.16 .92 Group IV average 0.13 0.90 0.90 0.22 0.21 1.05 0.14 0.17 0.82 0.21 0.30 0.69 Total average Source: Authors' calculations based on data provided in country studies. (Table continues on the following page.) 52 Historical Patterns of Price Intervention Effects of Greater Volatility in World Prices We would expect policymakers to be more concerned with price stabi- lization during periods of high price variability than at other times, and consequently we would expect greater price stabilization when world prices are particularly volatile. This is confirmed by a compari- son of price stabilization in the 1960s, when world commodity prices were relatively stable, and in the 1970s and early 1980s, when they were much more volatile. (That world prices were more volatile in the second period is clear from the coefficient of variation ratios in table 3-2, which are larger for most countries in the second period and are almost 80 percent larger on average than in the first period.) The degree of price stabilization was significantly larger in 1971-84 than in 1960-70. The ratio of the coefficients of variation of the relative producer price and the relative world price is smaller in 1971-84 than in 1960-70 for most countries and on average as a result of direct interventions (table 3-2). The average reduction in price variability was 31 percent in 1971-84 (a coefficient of variation ratio of 0.69) and 18 percent in 1960-70 (a ratio of 0.82). So although the variability of international prices was about 80 percent larger in 1971-84 than in 1960-70 (0.30 compared with 0.17), direct interventions reduced the variability of domestic prices more in the second period, so that domestic price variability was only 50 percent larger in 1971-84 than in 1960-70 (a ratio of 0.21 compared with 0.14). This pattern of greater price stabilization in the later period holds for all groups except the protectors (Group IV). The average for that group was dominated by Korea, where price stability fell in the sec- ond period, with the relative (producer to world) price variability ratio increasing from 0.78 in 1960-70 to 1.06 in 1971-84. Consumer Price Variability Direct and total interventions stabilized consumer prices for staples on average for the thirteen countries for which information was avail- able (the average ratios for the variability measures are all smaller than 1; see table 3-3). As with producer prices, the average ratios for the annual price change statistic are smaller than the standard devia- tion ratios, implying that price policies stabilized annual price varia- tions more than price variations around the mean. Direct interven- tions stabilized consumer prices more than indirect interventions, which slightly reduced consumer price variability when measured by the standard deviation and annual price change ratios but increased it significantly it when measured by the coefficient of variation ratio (from 0.72 to 0.85). The Effects of Intervention on Price Variability 53 Table 3-3. Relative Consumer Price Variability of Actual and Counterfactual Scenarios Price Country (products) Period Statistic Direct Total Brazil (rice, wheat) 1970-83 SDP/SDB 0.57 0.46 CVP/CVB 0.39 0.39 ZP/ZB 0.55 0.43 Colombia (rice, wheat) 1960-83 SDPISDB 0.49 0.43 CVP/CVB 0.50 0.59 ZP/ZB 0.34 0.25 Dominican Republic (rice, sugar) 1966-84 SDP/SDB 0.45 0.74 CVP/CVB 0.53 0.85 ZP/ZB 0.29 0.33 Egypt (rice, wheat) 1965-84 SDP/SDB 0.16 0.17 CVP/CVB 0.31 0.41 ZP/ZB 0.14 0.13 Ghana (rice) 1958-84 SDP/SDB 1.93 1.80 CVP/CVB 1.00 1.87 ZP/ZB 2.22 1.33 Korea, Rep. of (beef, pork, rice, 1960-84 SDP/SDB 1.47 1.57 barley) CVPfCVB 1.22 1.72 ZP/ZB 0.78 0,55 Morocco (sugar, wheat) 1960-84 SDP/SDB 0.39 0.67 CVP/CVB 0.34 0.79 ZP/ZB 0.31 0.47 Philippines (rice, sugar) 1960-82 SDPISDB 0.25 0.17 CVP/CVB 0.27 0.28 ZP/ZB 0.33 0.25 Portugal (wheat, silk) 1960-84 SDP/SDB 1.34 1.04 CVP/CVB 1.26 1.01 ZP/ZB 0.65 0.62 Sri Lanka (rice) 1960-84 SDP/SDB 1.33 0.67 CVP/CVB 1.35 1.16 ZP/ZB 1.00 0.60 Thailand (sugar, rice) 1962-84 SDP/SDB 0.36 0.35 CVP/CVB 0.47 0.53 ZP/ZB 0.41 0.38 (Table continues on the following page.) 54 Historical Patterns of Price Intervention Table 3-3 (continued) Price Country (products) Period Statistic Direct Total Turkey (wheat, cotton) 1971-83 SDP/SDB 0.25 0.14 CVP/CVB 0.19 0.19 ZP/ZB 0.20 0.10 Zambia (maize) 1967-85 SDPISDB 0.95 0.58 CVP/CVB 1.57 1.27 ZP/ZB 0.34 0.28 Average SDP/SDB 0.77 0.68 CVP/CVB 0.72 0.85 ZP/ZB 0.58 0.44 Source: Authors' calculations based on data provided in the country studies. All three measures show a reduction in consumer price variability for staples in Brazil, Colombia, the Dominican Republic, Egypt, Morocco, the Philippines, Thailand, and Turkey. Price policies in those countries also stabilized the producer price of staples. In the remaining five countries, consumer price variability either increased or the results were mixed. In Zambia the standard devia- tion and annual price change ratios show decreased price variability from direct and total price interventions, but the coefficient of varia- tion ratios show increased variability because of high direct and total rates of taxation. In Sri Lanka direct interventions increased vari- ability in the consumer price of rice, although they lessened producer price variability. Total interventions lessened consumer price vari- ability according to the standard deviation and annual price change ratios but increased variability according to the coefficient of variation ratio because of high total taxation. In Korea and Portugal the standard deviation and coefficient of variation ratios showed increased consumer price variability from intervention, while the annual price change ratios showed a decline. Producer price variability showed the same pattern in Korea but not in Portugal. In Ghana most measures indicate an increase in price variability for rice; the coefficient of variation ratio, however, shows no change in price variability for direct interventions because rice was heavily pro- tected. Interventions had a similar effect on the producer price of rice. Conclusion On the whole, price interventions have achieved the objective of stabilizing prices, mainly through direct interventions. Indirect inter- The Effects of Intervention on Price Variability 55 ventions, which were the major source of agricultural taxation (see chapter 2), did little to improve price stability, which might have mitigated their harmful overall effects on agricultural incentives. Countries tended to stabilize prices more during the period of volatile world prices in 1971-84 than when world prices were more stable in 1960-70. Producer prices were more stabilized for importables and staples than for exportable and nonstaple products. These results also hold for the variability of prices around their trend. Variability was also reduced for staple food prices for consumers in most countries examined. Appendix 3-1. Demand Elasticity and the Relation between Price and Income Variability If we assume that demand for a commodity is stable and that price variability results entirely from supply shocks, then the degree to which price stabilization measures stabilize income is inversely related to the elasticity of demand. If p is the price of the commodity, q is the quantity, and R = pq is revenue,5 then (3A-1) logR = logp + logq, and (3A-2) Var(logR) = Var(logp) + Var(logq) + 2cov(logp, logq). If (3A-3) q = ap- , where b is the price elasticity of demand (in absolute value), then it is easy to show that (3A-4) cov(logp, logq) = -bVar(logp), so that equation 3A-2 becomes (3A-5) Var(logR) = (1 - 2b) Var(logp) + Var(logq). Equation 3A-5 shows that when the price is stabilized, Var(logp) falls, and the reduction in the variability of revenue (or income) is inversely related to the price elasticity of demand, b. From equation 3A-5, we have (3A-6) dVar(logR) = 1 - 2b. dVar(logp) For b larger than 0.5, price stabilization will increase income vari- ability, and for b smaller than 0.5, price stabilization will stabilize 56 Historical Patterns of Price Intervention income. The lower the value of b, the greater the transmission from price stabilization to income stabilization. For a tradable product with a small share of the world market, price and quantity are uncorrelated, cov(logp,logq) is zero, and (3A-7) Var(logR) = Var(logp) + Var(logq). The derivative in equation 3A-6 equals one, so price stabilization unambiguously stabilizes income. Appendix 3-2. Defining the Z-Statistic The Z-statistic is defined as n 1 Z(qAt q- , t12 (3B-1) Z = '___=2n _I_I_ where qA1 is the relative price of product A at time t. The standard deviation of the price series is the square root of the average squared deviation of the price from the sample mean. The Z-statistic is the square root of the average squared deviation of the price from its value lagged one period (or of the first difference in the price). The two statistics are thus directly comparable. We calculated the Z-statistic for two reasons. First, producers may be more concerned with annual changes in their prices than with price deviations from the twenty-five-year sample mean. Second, two very different price series may have the same standard deviation even though one would be considered more stable than the other, as reflected by the Z-value. For instance, assume P = 100 for ten periods, then rises to 200 for ten additional periods, while P2 varies annually from 100 to 200, = 100,0 t s9 t 200, 10 s t ! 19 and P2t 100, t = 2n, 0 s n s 9 2 200, t = 1 + 2n, 0 5 n 9 Both series have the same mean of 150 and the same standard deviation of 50, but Pit has a much smaller Z-value than P2t-23 com- pared with 100. Thus, the Z-statistic seems to better reflect the rela- tive stability of the two series, in the sense that Pit experiences only one change over the period while P2t changes every year. Part II Output, Consumption, and Trade Effects of Price Intervention Supply Response and Growth in Agriculture and the Economy (in collaboration with Jorge Quiroz) Because agricultural price reform redistributes income among pro- ducers, consumers, and the government, it may involve resistance to the reform. But if improved (that is, more neutral) agricultural incen- tives lead to a more dynamic sector, with large gains in agricultural output and productivity over time and a positive effect on the econ- omy as a whole, the medium- to long-run effects may be substantial and may alleviate the short-run political costs of the price reform- and improve its sustainability as well. Our study shows that the aggregate supply response in agriculture is quite significant, and that the annual rate of agricultural growth would have been substantially higher in the absence of total price interventions. An important related issue is the effect of agricultural taxation on overall economic growth. In many developing countries, the high rate of agriculture taxation has been part of an explicit or implicit policy of industrialization-led growth, justified in part by the belief that industry was the dynamic sector while agriculture was static and not very responsive to incentives. That meant that economic growth could be accelerated by shifting resources from agriculture to indus- try, by taxing agriculture directly, and by protecting the industrial sector. Our analysis and tests show, however, that policies that depress agriculture's terms of trade below international levels are associated with slower economic growth. The efficiency gains of agricultural price reform depend largely on the aggregate response of the agricultural sector to improved incentives-itself a function of the response of both existing products and new products. Few studies have directly analyzed aggregate sup- ply response. Instead, much of the empirical work on supply response has dealt with individual products or groups of related prod- ucts. Most of the debate, however, has dealt with the degree of aggre- gate response over time. The issue is important because if the aggre- gate supply response is low, as has frequently been argued,' then the social (efficiency) cost of taxing agriculture is also low.2 We believe 59 60 Output, Consumption, and Trade Effects of Price Intervention that the methodologies used to estimate the aggregate supply response have often underestimated that response, and we discuss this issue in some detail. Influences on the Supply Response How agricultural supply responds to a price reform depends on how severely interventions have depressed prices, how extensive and credible the reforms are, how responsive output is to a given price change, and what time period is considered. Clearly, some of these elements are related. Responsiveness to a given price change depends on the credibility of the reform, and credibility may depend on the extent and nature of the price reform. We have already demonstrated in chapter 2 that price interventions severely depressed agricultural prices in the eighteen countries stud- ied over 1960-84. Total taxation averaged over 30 percent, reaching almost 40 percent or more in over half the countries. Direct price interventions reduced agricultural prices an average of 8 percent, with the prices of exportables declining by more than the average and importables receiving protection, on average. Indirect price interven- tions led to huge reductions in agriculture's domestic terms of trade in most of the eighteen countries, averaging over 22 percent. Without price interventions, agriculture's terms of trade would have increased by more than 40 percent during 1960-84. With such a large price impact from intervention, the effect of price reform on output is likely to have been significant as well. Extent and Credibility of Reform Both the breadth of price reforms and their credibility strongly influ- ence the response of producers to price changes. If price reform affects only a single product or a small group of products, the supply response for that product will tend to be large because of substitution (changes in output mix), but the aggregate supply response will be smaller than it would be had the reform been more general. That means that reforming the policies that create high rates of indirect taxation, which affect the entire agricultural sector,3 should result in a broader, more substantial response over time than reform of direct interventions alone. Producers' expectations about whether the price change will last will also influence how much supply responds to a given price change. If producers believe that price changes are only temporary, they might use more variable factors to increase the output of the annual crops whose prices increased. But they would not increase investment, in order to increase productivity or to increase the area planted with tree crops, as they might if they believed the new price Supply Response and Growth in Agriculture and the Economy 61 configuration would last, because they would not expect a higher return on their investment. The supply response would therefore be much more limited. Many factors affect the credibility or expected longevity of the new set of prices that result from reform. If the prices of products subject to price control are increased once and for all in an inflationary envi- ronment, producers would expect the nominal price increase to erode in real terms, and they would have little incentive to increase output. Their response would be different if price controls were removed entirely and prices were free to adjust to market forces. Macroeconomic uncertainty will also dampen the response to price reforms, because it creates instability in relative incentives in the economy. When inflation is high and the real exchange rate is sub- stantially overvalued, the rate of indirect taxation is high and highly variable. Removing direct price interventions may not lead to a large supply response under these conditions because of the general uncer- tainty and because the impact of nominal price changes on relative prices is hard to discern in such an environment. In order to generate a substantial response by producers under these conditions, agri- cultural price reform would probably need to be accompanied by broader reform of unsustainable macroeconomic policies. Institutional and infrastructure weaknesses can also limit the out- put response to price reform. Producers are unlikely to respond strongly to a price increase given by a marketing board if the infra- structure and human resources are inadequate to efficiently market and transport inputs and output, or if the marketing board lacks the financial resources to buy the larger output generated by the price increase. To make reform credible, other measures would be needed to strengthen and liberalize marketing and transport. Sometimes a poor supply response to price reform is related to the disparity between official prices and true incentives. Often official prices do not reflect marginal incentives because some transactions are carried out at other prices. In Kenya, when the marketing board was unable to buy the entire output of beans produced in response to its official offer of higher prices, farmers either waited months to get paid by the board or had to sell some of their output on the domestic market at a lower price. In Ghana, the Cocoa Marketing Board often delayed payments because of corruption or lack of funds, issuing promissory notes or checks that farmers had difficulty cashing locally. Under these conditions, farmers may be responding to the lower market price rather than to the "higher" official price. The weak response to an increase in the official price does not mean that farmers do not respond to higher incentives but only that they realize that higher official prices do not necessarily constitute an actual improvement in the incentives they face. The failure of poorer farmers to respond to input subsidies is also 62 Output, Consumption, and Trade Effects of Price Intervention sometimes viewed as an indication of a low responsiveness to improved incentives in the agriculture sector. Often, however, small- scale farmers fail to take advantage of input or credit subsidies not because they are uninterested but because poorly designed programs make it difficult for them to do so-while allowing larger, better- connected farmers or agricultural processors to capture the subsidies. Bates (1981: chapter 3) recounts numerous examples of this phenome- non in a number of African countries, including Ethiopia, Ghana, Kenya, Nigeria, and Sudan. He also notes that when small farmers do manage to receive subsidized fertilizer, they often receive it at the wrong time or receive a variety unsuitable to their soils or crops. Similarly, Avillez, Finan, and Josling (1988: 25, 26) found that Por- tuguese farmers are often reluctant to use "subsidized" credit because of the high cost (travel, time-consuming administrative requirements) of obtaining the funds. Sometimes supply fails to respond as expected to improved incen- tives, such as subsidized inputs, because the subsidies are not really reaching those the authorities thought they were reaching or intended to reach. Brandao and Carvalho (1991) report that in Brazil, subsidized credit went not to its intended recipients but to new "farmers," attracted to agriculture by the credit subsidy, which they used for speculative purposes. Brandio and Carvalho also report that the use of subsidized fertilizer was much lower than reported in 1975 and 1976 because of a brisk market in false receipts for the subsidies- the "paper fertilizer" market (28). Dethier (1989) reports that farmers in Egypt often evaded controls intended to ensure that subsidized inputs were used in major food crops, using them instead in more profitable crops. Hence, the response of the major food crops was considerably less than expected. Under any of these conditions, the supply response is likely to fall considerably short of its potential level. But if measures were taken to eliminate some of the constraints on improved input use and to ensure that prices reflected opportunities at the margin, the supply response to improved incentives might be considerably larger than under conditions in which prices do not reflect true opportunities and scarcities. Supply Response of Individual Products Policymakers are generally concerned with the short-term impact of a price reform, but reform's success and sustainability are often deter- mined by its long-term impact. Consequently, both the short-run and the cumulative output effects of direct and total price interventions were estimated in the country studies for this project.4 We defined the cumulative output effect as the difference between actual output Supply Response and Growth in Agriculture and the Economy 63 and the output that would have resulted if interventions had been removed in the first year of analysis (1960 in most cases) and output had been determined by the price path in the absence of intervention. We used a variant of the distributed lag or Nerlovian approach to estimate the cumulative output effect.5 The long-run Nerlovian out- put effect, or elasticity, is obtained for a given price change that lasts long enough for output to reach its new long-run equilibrium value. Our variant for the cumulative output effect allows for the price effect of interventions to vary over time. (Further details on the approach are presented in the appendix to this chapter.) We used another vari- ant to obtain the cumulative output effect for tree crops, which have a long gestation period. The results present no major surprises. Overall taxation of agricul- ture in most countries examined has been very high, and the output effects of this taxation have been correspondingly large, especially for exportables. In most cases, the cumulative output effects were larger than the short-run effects, whether positive or negative. The excep- tions were in countries where there was a reversal in the impact of price interventions over time from positive to negative (or vice versa), so that cumulative effects were smaller than short-run effects or the sign of the short-run and cumulative effects differed-for example, when direct protection was positive in the 1960s but turned negative in the 1970s as governments tried to stabilize domestic prices during the boom in world commodity prices. In general, the output effects were negative for products that were taxed directly (mostly exportables) and positive for products that were protected directly (mostly importables)-that is, the own-price effects dominated the cross-price effects. In almost all cases, the out- put effect of indirect and total price interventions was negative, espe- cially for exportables. The exceptions were some importables for which total intervention was positive because of high direct protec- tion and low indirect taxation; for them, the output effect of total interventions was positive. Take the case of wheat in Argentina, which was taxed by both direct and total price interventions. Direct price interventions reduced output by 3.5 percent in the 1960s, 19.3 percent in the 1970s, and 4.2 percent in the 1980s as measured by the short-run output effect, and by 9.6 percent, 31.9 percent, and 10.8 percent for the same periods as measured by the cumulative output effect. Total interven- tions reduced output during the same three decades by 13.6 percent, 23 percent, and 21.8 percent for the short-run output effect, or by 26 percent, 39.6 percent, and 36.6 percent for the cumulative output effect. The effects of interventions on agricultural output were clearly substantial, with the cumulative effects larger than the short-run effects. The output effects of direct price interventions were largest in 64 Output, Consumption, and Trade Effects of Price Intervention the 1970s, when direct taxation of agriculture rose during efforts to stabilize domestic food prices as world prices soared. But the output effects were larger for total price interventions than for direct inter- ventions, especially in the 1960s and 1980s, when they were four to five times larger for the short-run effects and three times larger for the cumulative effects. By the 1980s, the cumulative output loss from total price interventions was 36.6 percent, implying that wheat out- put would have been about 60 percent larger in the absence of total interventions over the period (36.6/63.4). The case of sugar in Pakistan demonstrates a different pattern. For direct price interventions, the short-run output effect was high and positive (63.2 percent) in the 1960s, when average protection was very high, but low and negative (-0.7 percent) in the 1970s, when average direct protection was slightly negative. The cumulative out- put effect remained positive in the 1970s (25.6), however, a carryover of the very large positive cumulative effect of the 1960s (91.2 percent). So although the direct protection rate was close to zero in the 1970s, cumulative output was still larger than it would have been had there been no direct protection in the 1960s. Rice in Korea exhibited yet another pattern. Total intervention was negative in the 1960s but positive thereafter as direct protection increased and indirect taxation fell. The effects on output, both short- run and cumulative, were correspondingly negative in the 1960s (about -10 percent) but turned positive in the early 1980s in response to increasing protection, averaging about 20 to 25 percent for the short-run effects and 40 to 45 percent for the cumulative effects. In general, the cumulative impact of total interventions has been extremely large, especially for exportables. The cumulative loss in output averaged about 47 percent for cotton and 21 percent for coffee in Colombia, 17 percent for sugar and 18 percent for coffee in the Dominican Republic, 19 percent for wheat (an importable) in Egypt, 30 percent for cocoa in Ghana (but 60 percent when the impact of failing to plant new trees is also considered), and 17 percent for palm oil and 23 percent for smallholder rubber in Malaysia. Aggregate Supply Response in the Short and Medium Run Even in the short run, aggregate output can expand. Expansion can be achieved through increased use of agricultural labor, capital, and variable inputs. The real value of agricultural gross domestic product (GDP) can also expand through the reallocation of agricultural labor, capital, and land to higher productivity uses in response to changes in relative agricultural prices as a result of price reform. Supply Response and Growth in Agriculture and the Economy 65 The Effect of Changes in Output Mix: Measuring Agricultural GDP A price reform that affects relative prices within agriculture will induce a change in output mix. The impact of that change on real agricultural GDP depends crucially on how it is defined. If real agri- cultural GDP is defined in terms of base or prereform prices (a Las- peyres quantity index), then changes in output mix following a price reform will have no impact on real agricultural GDP (assuming total quantity produced remains unchanged). When some other type of quantity index is used-such as the Paasche index, which uses post- reform (border) prices, or a superlative index, such as the Divisia index, which reflects the effect of substitution possibilities-a change in output mix implies a positive effect on real agricultural GDP. There may be good economic reasons for using the Paasche quantity index, since this would evaluate real agricultural GDP at or closer to border prices, the country's opportunity cost.6 Hence, the reallocation of a given set of resources to more productive uses within the agricultural sector will generate real income gains. This can be illustrated with a simple example. Recall that, on aver- age, direct price interventions protected importables and taxed exportables, resulting in a net tax for the sector as a whole (chapter 2). Removing direct price interventions, then, would lead to an increase in the price of exportables, a fall in the price of importables, and a net increase in prices for the sector as a whole. Assume that the output of exportables (Qx) is 100, the output of importables (Qm) is 100, and their respective prices (relative to non- agricultural prices) are Px = 100 and PM = 100, with the agricultural price level P = 100. The initial value of aggregate output is then 20,000. Further assume that the border prices in the absence of direct interventions are Pi, = 90 and P' = 120, so that after removing direct interventions the price level (at the original quantities) is P' = 105. The agricultural terms of trade have gone up by 5 percent, so at the original quantities, the value of output is now 21,000. In other words, the sector gained 5 percent in terms of the value of output. That is not the entire gain, however, because agricultural producers can move resources to the products whose relative price increased. We assume here that the short-run elasticity of output of exportables (Ex) equals 1 and that it depends on relative prices within agriculture (PXIPM).7 We further restrict the degree of aggregate supply response by assuming that purchased inputs and labor and capital remain unchanged and that total output remains at 200 units, even though agriculture's terms of trade increased. This allows us to focus exclu- sively on the impact of a change in output mix. The relative price PxlPM increased from 1 to P2IP& = 120/90 = 1.33. 66 Output, Consumption, and Trade Effects of Price Intervention With Ex = 1, the new output-of exportables is Q' = 133.3, implying that Q' = 66.7 (since total output is fixed). Now the value of output is 22,000. In other words, the increase in agriculture's terms of trade of 5 percent led to a rise in the value of output of 5 percent (from 20,000 to 21,000), while the change in output mix led to an additional gain of 4.8 percent (from 21,000 to 22,000). The gain in the value of output is 10 percent (2,000/20,000), and the gain in real terms is 4.8 percent as a result of the reallocation of resources to products where their margi- nal productivity is highest. The real value of agriculture's aggregate output increases by 4.8 percent when agricultural GDP is measured at border prices, which values output at the country's opportunity cost. The dollar value of agricultural GDP also increases by 4.8 percent. If the Laspeyres quantity index were used, real agricultural GDP would not change, because total output remained unchanged follow- ing the price reform. According to this definition (used to measure real agricultural GDP in national income accounts), changes in output mix would have no impact on real agricultural GDP. With the Paasche quantity index, however, changes in output mix will have a positive impact on real agricultural GDP, with the effect depending on how much the relative price PxlPM changes for a given change in agricul- ture's terms of trade. In our example, real agricultural GDP increased by 4.8 percent. If, however, for the same change in agriculture's terms of trade (P' = 105), the relative price change were larger, say P = 130 an,d P' = 80, the index of real aggregate GDP would also increase-by 14.9 percent instead of by 4.8 percent, as in our first example. The use of this index reflects changes in the real value of agricultural output in terms of border prices or constant dollars. A few examples from the country studies, using a Divisia quantity index to measure changes in aggregate output,8 can also illustrate the importance of how real agricultural GDP is defined. In the Dominican Republic direct price interventions protected importables (rice) at 19 percent and taxed exportables (coffee and sugar) at 24.8 percent. Removing direct interventions would have increased the relative price of exportables (Px/PM) by 58 percent. Keeping the value of out- put in producer prices constant, and assuming a price elasticity of exportables with respect to PxIPA of only 0.44, we find that the change in output mix resulting from the removal of direct price inter- ventions would have increased real agricultural GDP by 11.8 percent.9 In Malaysia direct protection of importables (rice) was 23.6 percent and direct taxation of exportables (palm oil and rubber) was 12.7. Removing direct intervention would have increased PXIPM by 41.6 percent, and with an elasticity of 0.1, real agricultural GDP would have increased by 4.6 percent because of the change in output mix. As these examples show, when output is measured at border prices, Supply Response and Growth in Agriculture and the Economy 67 changes in agricultural output mix may have a significant impact on real agricultural GDP. Impact of Interventions on Short- and Medium-Run Output Response To examine the impact of price policy on annual growth of real agri- cultural GDP, we performed two tests. We used a nonparametric test to compare the average agricultural growth rate in the group of coun- tries in which nominal protection rates were higher than the average with the rate in the group in which protection rates were lower than the average. This test showed a higher average growth rate for the group with higher protection rates. For the second test, we used a regression estimate of the growth rate of real agricultural GDP to Simu- late its average growth rate in the absence of total price interventions. We also found a significant relationship between total protection and agricultural growth. For at least two reasons, changes in the level of agriculture's terms of trade may be related to changes in the growth rate of agriculture. First, according to standard price theory, changes in the level of out- put are related to changes in the price level. If adjustment to a new price level is slow, however, because of high adjustment costs or adaptive expectations, the growth rate of output may also be affected for a long time. Second, and more important, a higher relative price level for agriculture may permanently reduce labor outmigration from the sector, increase investment, and encourage wider adoption of new techniques, thereby resulting in a higher growth rate for the sector (Mundlak 1985). NONPARAMETRIC TEST. For the nonparametric test, we pooled the 373 annual observations on nominal protection rates and growth rates over the sample period across the eighteen countries. We than classi- fied the observations into the high (above-average) and low (below- average) protection rate groups and compared the average growth rates in the two groups. We used protection rates rather than agricul- ture's terms of trade (or relative price) because relative price indexes are not comparable across countries. We used agricultural GDP rather than the value added for the selected products because we are inter- ested in the impact of price policy on sectoral growth. Variations in average protection rates across countries tend to better reflect longer- term differences in protection rates than do variations in protection rates over time, which may reflect more transitory effects such as variations in border prices. Decomposing the variation in direct and total protection rates showed that, for both measures, about 60 per- cent of the variation is due to variation across countries and about 40 68 Output, Consumption, and Trade Effects of Price Intervention percent to variation over time. Consequently, the relation between growth rates and protection rates discussed below reflects a mix of both short- and long-term effects, but with a higher weight for the long-term effects. For the high and low direct protection rate groups,10 the difference in mean agricultural growth is small (1 percentage point, or 30 per- cent) and statistically not significant, but for the two total protection rate groups, the difference is large (2.5 percentage points, or 90 per- cent) and significant (table 4-1). The test provides strong evidence of an association between high total taxation and low rates of agri- cultural growth, but weaker evidence of such an association for the direct protection rate. This finding seems to confirm that total inter- vention rather than direct intervention is the relevant concept for examining differences in agricultural growth performance because it also captures the substantial impact of industrial protection and exchange rate misalignment. Direct protection, although the more commonly used measure, captures only a small fraction of the impact of price interventions on agricultural incentives (about a quarter of the total impact on average in our sample). There are additional reasons that explain why total protection might be more strongly associated than direct protection with growth in real agricultural GDP. The relation we are interested in is that between agricultural growth and agricultural incentives. Since Table 4-1. Comparison of Nominal Protection Rates and Agricultural Growth (percent) Direct protectiona Total protection Variable High Low High Low Nominal protection rateb 10.8 -29.3 -8.3 -46.2 Annual agricultural growth rate 4.3 3.3 5.2 2.7 a. For this analysis, we used the traditional measure of direct protection (difference between the producer price and the border price in the absence of direct interventions relative to the border price in the absence of direct price interventions) rather than the measure used elsewhere in this study (difference between the producer price and the border price in the absence of direct interventions relative to the boider price in the absence of total price interventions). The traditional measure reflects the actual change in price resulting from direct interventions and so is the relevant measure for evaluat- ing output response using supply elasticity; see note 10 for this chapter. b. The average nominal direct protection rate was -10 percent, with 197 observa- tions below and 176 above the average. The average total protection rate was -30 percent, with 216 observations below and 157 observations above the average. The rates in the table are the averages for the high (above-average) and low (below-average) groups. Source: Authors' calculations as explained in the text. Supply Response and Growth in Agriculture and the Economy 69 indexes of agricultural incentives (relative prices) cannot be compared across countries, we use protection rates instead. And differences in total protection rates tend to reflect differences in agricultural incen- tives more closely than do differences in direct protection rates. This is so for several reasons. First, direct price interventions are used in part to stabilize domestic prices, and so rates of direct protec- tion tend to move inversely with border prices. Consequently, changes in direct protection rates do not always reflect changes in producer prices. Indirect protection rates, however, would not be expected to be closely (and inversely) related to agricultural border prices, and so changes in total protection tend to reflect more closely changes in producer incentives. This is confirmed by our finding that most of the reduction in producer price variability (relative to that of border prices) is due to direct price interventions, with indirect price interventions having a negligible impact (see chapter 3). Second, the impact of direct taxation on agricultural GDP may be ambiguous. Some of the impact may result from changes in output mix, since the impact of direct price interventions varies by product or product group-some are taxed and others are protected." Hence, even a low average rate of direct taxation may have a considerable effect on output if the tax structure has a strong effect on relative prices within agriculture. This is less likely to occur with indirect intervention: industrial protection policies reduce relative incentives for the entire agricultural sector, and overvaluation of the real exchange rate taxes all agricultural tradables, which constitute the bulk of the agricultural sector in most countries. Third, changes in direct intervention are not always positively cor- related with changes in total intervention. And the latter constitute a better measure of the overall impact of price intervention than the former. Hence, it should be no surprise that agricultural growth is more strongly associated with total than with direct intervention. Consequently, we use the total protection measure in the regression analysis that follows. PARAMETRIC TEST. The model used for simulating the impact of total price interventions on agricultural GDP Consists of two equations; esti- mation is based on the pooled data of 373 observations for the eigh- teen countries. The first equation relates the long-term level of agri- cultural GDP (in logarithm) with a trend variable and with the rate of total protection. The second equation relates the percentage change in agricultural GDP to the percentage change in GDP, the change in the rate of total protection, and the lagged residual from the first equation. In the first equation, the coefficient of the trend variable reflects the exogenously determined agricultural growth rate, say because of 70 Output, Consumption, and Trade Effects of Price Intervention growth in the agricultural labor force or increased public investment. The coefficient of the total protection rate reflects the impact of higher incentives on the long-term path of agricultural GDP. The results for this first equation for the long-term path of agricultural GDP are logYAj, = - 72.2 + (0.0393 - 0.0003D1 (-19.6) (21.1) (-15.6) (4-1) - 0.0002D2)t + 0.424NPRT(i) (-12.8) (9.09) R2 = 0.70, D. W = 0.23, where YAj, is agricultural GDP in country i in year t, t is the trend variable, D1 and D2 are dummy variables for countries with low growth rates and average growth rates, respectively, and NPRT6) is total nominal protection in country i in year t. All coefficients were found to be highly significant. The coefficient of the trend variable is 0.039, indicating an annual growth rate of 3.9 percent, which is close to the actual average annual growth rate of 3.86 percent for the eighteen countries (see table 4-2). The coefficient of the total rate of protection is 0.424, indicating that an increase in total protection (that is, a reduction in total taxation) from -30 per- cent (the actual average, as shown in table 2-1) to zero will have an effect on the path of long-term output of over 12 percent. The two dummy variables for low and average growth rates are significant, but their impact is negligible. Countries with high growth show a steady-state annual growth rate of 3.93 percent, those with low growth have a rate of 3.9 percent, and those with average growth have a rate of 3.91 percent. The second equation relates the percentage change in agricultural GDP to the percentage change in GDP, the change in the rate of total protection, and the lagged residual from equation 4-1 (the difference between the actual and the long-term level of agricultural GDP). The results of the equation for this short-term effect on agricultural GDP are AlogYAj, = .028 + 0.666AlogY, + 0.132ANPRTi) (3.73) (6.88) (3.83) (4-2) - 0.031D1 - 0.013D2 - 0.089RESi,,-1 (-2.32) (-1.21) (-3.95) R2 = 0.185, D.W = 1.88, where Y, is GDP in country i in year t, RESj,t-1 is the residual of equation 4-1 in year t-1, and AXj, is equal to Xj, - Xi,t,, the first difference in Xit. (The low R2 is typical of equations estimated in the Supply Response and Growth in Agriculture and the Economy 71 Table 4-2. Effect on Agricultural Growth of Removing Total Price Interventions, 1960-84 (percentage) Twenty- Simulated year Total Actual Simulated growth cumulative Country intervention, growthb growthc effectd eftfect C6te d'Ivoire -49.00 5.20 6.08 0.88 49.97 Ghana -59.50 5.80 7.12 1.32 86.92 Zambia -46.30 1.90 2.94 1.04 32.81 Group I -51.60 4.30 5.38 1.08 56.56 average Argentina -39.10 1.00 1.67 0.67 17.25 Colombia -30.00 3.00 3.80 0.80 30.23 Dominican -39.90 2.40 3.12 0.72 24.17 Republic -44.40 6.90 7.75 0.85 65.19 Egypt -32.40 2.70 3.30 0.60 21.05 Morocco -39.50 2.60 3.81 1.21 44.15 Pakistan -27.40 4.70 5.31 0.61 30.87 Philippines -40.10 2.40 3.13 0.72 24.53 Sri Lanka -40.10 5.10 5.84 0.74 40.74 Thailand -31.80 2.80 3.36 0.56 19.94 Turkey -36.47 3.36 4.11 0.75 31.81 Group II average -8.30 7.60 7.88 0.28 23.09 Brazil -21.60 1.00 1.35 0.35 8.74 Chile -17.60 5.80 6.12 0.32 19.23 Malaysia -15.83 4.80 5.12 0.32 17.02 Group III average 13.20 7.40 6.82 -0.58 -42.80 Korea, Rep. of 7.70 1.10 1.06 -0.04 -0.98 Portugal 10.45 4.25 3.94 -0.31 -21.89 Group IV average Average -30.3 3.86 4.47 0.61 27.5 a. Total nominal protection rate, average 1960-84. b. Growth of agricultural GDP, geometric average 1960-84. c. Growth of agricultural GDP in absence of total intervention, geometric average 1960-84. d. Simulated growth minus actual growth. e. Effect in year twenty of removing total intervention (with annual growth equal to the average in each year), as a percentage of base year output, that is, viewed from initial year's perspective). Source: Authors' calculations as explained in the text. 72 Output, Consumption, and Trade Effects of Price Intervention first difference in variables rather than in their level.) The coefficient of the lagged residual is -0.089, indicating that if actual agricultural GDP is below its long-term path (determined by the trend variable), the negative residual will lead to a short-term increase in the level of agricultural GDP in the next period of about 9 percent of the deviation between the actual and long-term value. This means that the adjust- ment to the long-term value of agricultural GDP will be slow-about 50 percent after 7.5 years and 60 percent after ten years. Consequently, an increase in the total rate of protection (or a reduction in the rate of taxation) will result in an increase in the growth rate of agricultural GDP that will last many years.12 The model with its estimated coefficients was then used to simulate the level of agricultural GDP in the absence of total price interven- tions.13 The new agricultural GDP series for each country was then regressed against a trend variable to obtain the growth rate of agri- cultural GDP in the absence of total price interventions. Total protec- tion had a large impact on the growth rate of the agricultural sector (table 4-2): removing all price interventions was associated with an average annual increase in agricultural growth of 0.61 percentage points (from 3.86 to 4.47 percent, or a 16 percent increase) for the eighteen countries. For individual countries, the effects of removing total price inter- ventions vary. In Argentina, where total protection was negative, its removal would have led to a 0.67 percentage point increase (from 1 to 1.67 percent, or a two-thirds increase) in the annual agricultural growth rate. In Korea, which was an exception in our sample with positive total protection, removal would have reduced the agri- cultural growth rate by 0.58 percentage points (from 7.40 percent to 6.82 percent, or a 7.8 percent reduction). In the absence of total price interventions, the annual agricultural growth rate would have been 1.08 percentage points higher for the group of extreme discriminators (Group I), 0.75 percentage points higher for the representative group (Group II), and 0.32 percentage points higher for the mild discrimina- tors (Group III). For the protector countries (Group IV), however, it would have been 0.31 percentage points lower. For individual countries, the effects of removing total price inter- ventions vary. In Argentina, where total protection was negative, its removal would have led to a 0.67 percentage point increase (from 1 to 1.67 percent, or a two-thirds increase) in the annual agricultural growth rate. In Korea, which was an exception in our sample with positive total protection, removal would have reduced the agri- cultural growth rate by 0.58 percentage points (from 7.40 percent to 6.82 percent, or a 7.8 percent reduction). In the absence of total price interventions, the annual agricultural growth rate would have been 1.08 percentage points higher for the group of extreme discriminators Supply Response and Growth in Agriculture and the Economy 73 (Group I), 0.75 percentage points higher for the representative group (Group II), and 0.32 percentage points higher for the mild discrimina- tors (Group III). For the protector countries (Group IV), however, it would have been 0.31 percentage points lower. The cumulative twenty-year effect of removing total price interven- tions depends on both the actual growth rate and the change in the annual agricultural growth rate (growth in the absence of total price interventions minus actual growth; see table 4-2). The output effects were very significant. On average, the cumulative impact would have been 27.5 percent of the initial output, or about 13 percent of the output after twenty years. The effects in individual countries varied substantially. In Ghana the cumulative impact of removing total price interventions would have been 86.9 percent of the original output (or 28 percent of the last year's output), because of both a large impact on the growth rate and a high actual growth rate. By contrast, the cumulative impact in Chile would have been only 8.7 percent, because of a low impact on the annual growth rate and a low actual growth rate. In Korea, which had a positive total protection rate, the cumulative impact of removing all price interventions would have been a reduction of 42.8 percent of initial output (or a reduction of 10 percent of the output after twenty years). From another test that related growth of agricultural GDP to indirect and total rates of protection and to misalignment of the real exchange rate, we found that all three variables had a negative impact on agri- cultural growth, but the significance of indirect protection was low (see chapter 9 for details). Aggregate Supply Response in the Long Run Single-Equation Regressions Versus General Equilibrium Results Further increases in aggregate output are possible beyond those resulting from increased use of existing resources and changes in output mix. But they require bringing in more labor and capital from other sectors and wider adoption of new techniques. These processes take time. Most empirical studies of supply response (including those by Bond 1983, Herdt 1970, Peterson 1979, and Reca 1980, and most of the country studies in this project) have used single-equation regression analysis, with a lagged output variable for estimating Nerlovian long- term output effects. But this approach cannot capture the underlying economic structure or the actual economic phenomena that occur over time in response to a price change.14 As a result, such single- equation regressions generally produce lower aggregate output price elasticities. 74 Output, Consumption, and Trade Effects of Price Intervention Estimating the long-run aggregate supply response requires an economywide perspective that takes into account intersectoral resource movements in response to changes in incentives across sec- tors. A few studies have done this, including the country study on Chile for this project, another study on Chile by Coeymans and Mundlak (1992), and the studies on Argentina by Cavallo (1988) and by Mundlak, Cavallo, and Domenech (1992). These studies attempt to capture the underlying structure by specifying and estimating dynamic general equilibrium models that explicitly include intersec- toral resource reallocation over time through migration and invest- ment in response to sectoral incentives. The studies find that labor migration and investment respond to agriculture's terms of trade. Mundlak (1985, 67) found that "policies which tax agriculture reduce the investment in agriculture, increase the outmigration and reduce the implementation of new techniques." Cavallo (1988) simulated the impact of a 10 percent price increase to obtain aggregate output price elasticities for Argentina, based on the work of Mundlak, Cavallo, and Domenech (1992). The short- and medium-term responses are low, but the full response is large, with an elasticity of 1.78. It takes twenty years to obtain that response, however, and most of it comes only in the second ten-year period (see, also, Binswanger 1990). The price elasticity is 0.07 after one year, 0.16 after three years, 0.36 after five years, 0.71 after ten years, and 1.78 after twenty years. Because the total rate of taxation of agricul- ture during 1960-84 was so high in Argentina (39.1 percent), how- ever, removing total price interventions would have increased agri- culture's terms of trade by 64.2 percent. So even with these low elasticities, the output effect would have been substantial-about 4.5 percent after one year, 10.3 percent after three years, 23 percent after five years, 45.6 percent after ten years, and 114.3 percent after twenty years. Four observations about these results are worth emphasizing. First, even though the short- and medium-term elasticities are small, the short- and medium-term output effects are not because of the very large price increase that would have resulted from removing total price interventions. Second, the long-term effect is significantly larger with an approach that explicitly captures intersectoral resource move- ments. Single-equation regressions, such as those used for the esti- mates presented earlier in this chapter, underestimate aggregate sup- ply response because they do not allow for intersectoral resource reallocation as a response to changes in incentives. Third, if estima- tion of the migration and investment functions underlying the elastic- ity estimates were based on price data reflecting large random price variations around a weak price trend (the price behavior Sturzeneg- ger [19901 found to characterize agriculture in Argentina during 1960- 85), then the parameter values obtained might strongly underesti- Supply Response and Growth in Agriculture and the Economy 75 mate the response to a long-term (permanent) change in price policy. And, finally, other estimates of short- and medium-term elasticities are not as low as those Cavallo found for Argentina. In a study on Chile, Coeymans and Mundlak (1992) estimated short- and medium- term elasticities that are considerably larger than those found for Argentina, while the long-term elasticity is smaller. For Chile esti- mated elasticities are 0.6 after four years (compared to 0.36 after five years in Argentina) but only 1.4 after nineteen years (compared to 1.78 after twenty years in Argentina). Underestimation of Aggregate Response A general problem with simulating the effects of a permanent change in price policy using estimates based on time-series data is that this method underestimates the effects. Time-series data include price variations that are only temporary, or that producers believe are tem- porary. Since the estimates are obtained under conditions of general and relative price instability, they will underestimate the response to a permanent change in incentives under conditions of stability. This problem affects the estimation of both single-equation models and dynamic general equilibrium models. Consider the issue of price policy changes and expectations about their duration. Farmers are less likely to respond strongly to price changes that reflect annual variations around a given mean price than to variations that reflect a change in the mean that is expected to last. Farmers would be likely to respond to temporary price changes only by altering their use of variable factors. If they expect a price change to last, however, they would also adjust the quantity of fixed factors, by investing in machinery and other improvements, and they might also adopt new techniques. In other words, the greater the variability in prices, the lower the response to a given change in price. Bapna (1980), in a study of aggregate supply response in Ajmer district in India, found much lower supply elasticities (from 0.2 to 0.25) when using the volatile lagged price as the expected price than when using a moving average (0.5 to 0.6). Similarly, in a study of Canadian wheat farmers, Schiff (1986) found that the response to a preannounced price increase was about double the response to an expected but uncertain price increase. A general climate of uncertainty can also dampen the response to a price change, and again, models that fail to take this into account will underestimate the effects of credible price reform under more stable conditions. Macroeconomic instability, with high and variable rates of inflation and high variability in relative prices such as the real exchange rate, real interest rates, and asset prices, will discourage investment. Producers will try to hedge the value of their assets rather than invest in new production. 76 Output, Consumption, and Trade Effects of Price Intervention Other types of uncertainty will have a similar effect. Chronic uncer- tainty about the definition and enforcement of property rights will tend to discourage long-term investments. This is the case in many tropical regions in developing countries, where lack of legal title to land has led to slash-and-burn agriculture, with harmful conse- quences for agricultural investment, the biomass, and the environ- ment.15 Periods of temporary uncertainty about property rights prob- ably have a similar, inhibiting effect on agriculture. During the Chilean land reform program under the Frei administration, invest- ment in cattle, tree crops, and structures fell dramatically, and farmers substituted crops with shorter production lags. The aggre- gate supply response in such circumstances would be expected to be low. Another reason that the dynamic general equilibrium studies on Argentina and Chile underestimate output response is that they assume that total investment is given, so that price policy affects only the sectoral allocation of a given investment total. But eliminating price interventions, particularly indirect interventions, should result in a more stable macroeconomic environment and more credible and sustainable policies. This change could well lead to higher domestic and foreign investment, and thus to a stronger supply response than that estimated under conditions of macroeconomic instability. Effects on Economic Growth The high rate of taxation of agriculture in many developing countries has been motivated in part by the belief that industry was the dynamic sector while agriculture was static and not very responsive to incentives. If that were the case, economic growth could be acceler- ated by shifting resources from agriculture to industry. This resource shift was to be achieved by taxing agriculture directly and by protect- ing the industrial sector and maintaining an overvalued real exchange rate (below its equilibrium value). Several of the country studies for this project explicitly mention such an industrialization strategy (Argentina, Chile, Colombia, the Dominican Republic, Ghana, Malaysia, Morocco, Pakistan, and Zambia), and most of the others suggest that such a policy was implicit in the objective of increasing government revenue by taxing agriculture and using it to increase spending outside agriculture. Did these policies accelerate growth? We examined the impact of such price interventions on overall economic growth in three ways. First, we compared the various measures of intervention with the rates of growth of GDP across the four country groups. Second, we used regression analysis to relate the various measures of interven- tion to GDP growth for the eighteen countries. Third, we used a non- Supply Response and Growth in Agriculture and the Economy 77 parametric test to examine the relation between the various measures of intervention and GDP growth for the eighteen countries, dividing them into two groups based on the average size of the various mea- sures of intervention. All three tests16 suggest a negative relation between the rate of total taxation of agriculture and GDP growth: policies that depress agricul- ture's terms of trade below international levels are associated with slower economic growth. We also found a negative relation between the rate of indirect taxation of agriculture and GDP growth across the four country groups. The regression analysis also indicated a signifi- cant negative relation between exchange rate overvaluation and GDP growth, and the nonparametric test showed a significant negative relation between industrial protection policies and GDP growth. Comparison of Price Interventions and GDP Growth by Country Groups Table 4-3, which relates rates of taxation and growth for the four country groups, shows that higher rates of indirect and total taxation have been associated with lower economic growth. The indirect rate of taxation captures the effects of both exchange rate distortion, which measures distortion in the price of tradables in agriculture relative to nontradables, and industrial protection, which measures distortion in the price of agricultural tradables in relation to other tradables. The rate of total taxation also captures the effect of direct intervention on agricultural prices. We see that the growth rate of GDP for the least interventionist group (the protectors) is twice as large (6.5 percent) as that (3.3 per- cent) for the most interventionist group (the extreme discriminators). Table 4-3. Price Interventions and GDP Growth, by Country Group, 1960-85 (percent) Nominal rate of protection Annual GDP Country group Indirect Total growth Extreme discriminators (Group I) -28.6 -51.6 3.3 Representative discriminators (Group II) -24.4 -36.4 5.1 Mild discriminators (Group III) -15.7 -15.8 5.3 Protectors (Group IV) -13.6 10.4 6.5 a. For the countries in each group, see table 4-2. Source: Calculations based on data from country studies. 78 Output, Consumption, and Trade Effects of Price Intervention Table 4-4. Regression of Average Annual Rate of GDP Growth Against Intervention Variables Exchange Total Indirect rate protection protection overvaluation Number of Constant (NPRT) (NPR') (ERD) R2 F observations 6.26 0.043 .. .. 0.17 3.25 18 (7.57) (1.80) 6.21 .. 0.054 .. 0.06 1.03 18 (4.86) (1.02) 5.88 .. .. -0.028 0.12 2.08 18 (7.74) (- 1.44) Note: The numbers in parentheses are t-values. Source: Authors' calculations as explained in the text. We cannot claim causality from high taxation to slower growth, but this inverse relation between taxation and GDP growth is confirmed by our other two tests as well. Regression Analysis of GDP Growth and Intervention In the second test, we ran regressions of the average annual rate of growth of GDP for each country against the average indirect and total rates of protection and exchange rate distortion (table 4-4). GDP growth was positively related to total protection (or negatively related to total taxation), negatively related to exchange rate distortion, and positively related to indirect protection (or negatively related to indi- rect taxation). Our null hypothesis was that the annual rate of GDP growth is positively related to the rate of total and indirect protection (or nega- tively related to the rate of total and indirect taxation) and negatively related to exchange rate overvaluation. To test the hypothesis, we performed one-tailed tests. With 16 degrees of freedom,17 the rate of total protection is significant at the 5 percent level, and exchange rate overvaluation is significant at the 10 percent level. The rate of indirect protection, which is a weighted average of exchange rate overvalua- tion and industrial protection, is significant only at the 17 percent level, because the rate of industrial protection is not significant. The results indicate that a 10 percentage point increase in the rate of total taxation reduces annual GDP growth by 0.43 percentage points and that a 10 percentage point increase in overvaluation of the exchange rate reduces growth by 0.28 percent points. With an aver- age annual GDP growth rate of 5 percent for the eighteen countries, a Supply Response and Growth in Agriculture and the Economy 79 Table 4-5. Values of Various Measures of Intervention for the Eighteen-Country Sample (percentages) Standard Minimum Maximum Number of Variable Mean deviation value value observations Annual GDP growth 5.0 2.0 2 8 18 Total protection -30.3 19.1 -59 13 18 Indirect protection -22.5 8.9 -37 -1 18 Exchange rate overvaluation 32.0 23.8 2 103 18 Industrial protection 44.2 31.5 1 135 18 Direct protection -7.8 16.5 -27 39 18 Source: Calculations based on data from country authors. 10 percentage point fall in total taxation would result in a 10 percent increase in the GDP level after twenty-five years, and a 10 percentage point reduction in the degree of overvaluation would result in a 7 percent increase in the GDP level. Nonparametric Test of the Relation between GDP Growth and Intervention The third test is a nonparametric test examining the relation between GDP growth and the various measures of intervention. For each mea- sure, the sample of eighteen countries was divided into two subsam- ples, one for values higher than the average and one for values lower. (The mean, standard deviation, minimum, and maximum values for GDP growth and various measures of intervention are presented in table 4-5.) We then ran a t-test to determine whether the average GDP growth rate was significantly smaller in the subsample with the higher-than-average values of intervention measures than in the other subsample. The results show that GDP growth falls with an increase in the rates of total and indirect taxation and in the rate of industrial protection (table 4-6). The v statistic18 shows a significant relation between GDP growth and total taxation (at the 2.5 percent level), a nonsignificant relation between GDP growth and indirect tax- ation (and exchange rate overvaluation), and a significant relation between GDP growth and industrial protection (at the 1 percent level). Our analysis of the three tests indicates that policies that lower agriculture's domestic terms of trade below their international level Table 4-6. Results of Nonparametric Test of Relation between Measures of Intervention and Average Annual GDP Growth Rate Subsample I Subsample 2 GDP GDP Intervention growth Number of Intervention growth Number of Measure of intervention rate rate observations rate rate observations v statistic 8 Total taxation 12.0 5.43 7 42.0 4.71 11 2.12 Indirect taxation 15.9 5.15 9 29.0 4.84 9 0.93 Industrial protection 22.9 5.78 10 70.8 4.02 85.97 Note: Rates of total and indirect taxation indicate negative rates of total nominal and indirect protection, respectively. Also, a positive rate of industrial protection implies taxation of agriculture. Source: Authors' calculations as explained in the text. Supply Response and Growth in Agriculture and the Economy 81 are associated with a slowdown in economic growth. This is precisely the opposite of that intended from industrialization-led growth strat- egies based on taxation of agriculture through direct and indirect policies.19 Conclusion Taxation of agriculture has been a major part of a development strat- egy based on industrialization-led growth. The belief that economic growth could be accelerated by shifting resources from agriculture to industry rested on the premise that agriculture was a static sector and received further support from the prevailing pessimism about the prospects for agricultural exports. Our results show that these were gross misperceptions from which both agriculture and the economy as a whole may have suffered substantial harm in many developing countries. Both the rate of taxation and the aggregate supply response of agriculture have been considerably greater than envi- sioned by the proponents of this development strategy. The pro- tracted and extensive taxation of agriculture have been associated with a lower rate of agricultural and overall economic growth. Appendix. A Variant of the Nerlovian Model for Dynamic Output Effects We wish to examine the time path of output generated by alternative paths of output prices and input prices in order to estimate the dynamic output effects of alternative price policies. For that purpose, we use a general form of a Nerlove-type supply function: (4A-1) Xt = f(P,, Py",, W_1, Xt- where X, is output of X at time t, P,,t is the expectation at time t -1 of price P,, at time t, Pe,t is the expectation at time t -1 of price vector PY,, of the vector of substitute crops Y, and W is the price vector of variable inputs at time t-1.20 All variables are in logarithm. All prices should be thought of as being relative to an index of nonagricultural prices. We start with the case where the share of purchased inputs in the value of output is negligible.21 Since we are looking at the impact of price policy on a single crop at this point, we can assume that the prices of primary factors (labor, land, capital) are independent of price policy, and we can abstract from them. (This may not be a reasonable assumption for the agricultural sector as a whole, however.22) We start by examining the output effect of policies affecting the crop's own price, and then we generalize the analysis by including the price of substitute crops. 82 Output, Consumption, and Trade Effects of Price Intervention The assumption of adaptive expectations results in the same output equation as that for adjustment costs. We start with adaptive expectations: X= a + bPe,, and Pe,, - Pe,-1 = 0(,- - ,-1), 0 < 0 < 1. (4A-2) Then, Pe, = 0x,t-1 + (1 - )Pe,t-1 0 - 0Y)-1P't-1 so that (4A-3) Xt = a + bO Z(1 - 0)i_1x'_11 which can be rewritten as (4A-4) Xt = a + bOPt-1 + (1 - 0)Xt_1, where 0 is the coefficient of adjustment of expectations, b is the long- term elasticity, and bO is the short-term elasticity (after one period). The coefficient b is the long-term elasticity in the sense that it mea- sures the output effect of a given price change after the given price change lasted an infinite number of periods. Policies do not in general have constant price effects over time, however. An approach designed to capture the long-term output effect in such a case is developed below. Adjustment costs can also cause lags in the adjustment of output to price changes. Inputs are not perfectly mobile or malleable. Capital and labor in agriculture differ from capital and labor in other sectors, and capital and land (and possibly labor) may differ by crop as well. Moving factors across sectors or crops is costly, and the marginal cost of moving them increases as more of them are moved. This implies that the adjustment to any new long-term equilibrium will be grad- ual. We could thus have X*= a + bPe, (4A-5) Pe,t = X,t-1 X, - Xt = a(X*- Xt-J), 0 < a < 1, where X, is output at period t, X*is long-term equilibrium output at period t, a is the coefficient of adjustment of output, b is the long- term elasticity, and ab is the short-term elasticity. We have assumed that the expected price is the lagged price, or cobweb-type expecta- tions, where Pe = Pt _ 1. Equation 4A-5 can be rewritten as Supply Response and Growth in Agriculture and the Economy 83 (4A-6) X, = aX*+ (1 - a)X,-, or (4A-7) X, = aa + abP,t + (1 - a)Xt-i. Equation 4A-7 is equivalent to (4A-8) Xt = a + aebE(1 - ae)i-1p"'_ i i=1 The concept of long-term equilibrium output (Xf) is of the compar- ative statics type. It assumes that enough time has elapsed since the price change for actual output X to equal long-term desired output x*. Equations 4A-4 and 4A-7 are identical in the sense that one cannot empirically distinguish between a and 0. Since this is the case, we proceed with equations 4A-7 and 4A-8. The parameters a, b and a can be estimated from equation 4A-7, and the parameter values can be replaced in equation 4A-8 to calcu- late the dynamic output effects. From equation 4A-8, we obtain (4A-9) AX, = abE (1 - a);-lAkt-i- i=1 Now assume that the policy change (say, elimination of direct and indirect interventions) occurs at the beginning of the period under study, or time to. Then the output effect AXt at any time t > to can be derived from equation 4A-9 as t-t (4A-10) AXt = ab 1 (1 - a)i-1AP i=1 where AP,r t is the price change caused by the policy, and AXt is the cumulative effect on X, as a result of the policy change. Alternatively, AX, can be obtained from equation 4A-7 as (4A-11) AX, = abAP,, + (1 - a)AXt-1, where abAP,t, is the short-term effect. Thus, a policy change at time to does not effect Xto, but has an effect at to+1 equal to AXo+l = abAP,0, and an effect at to + 2 equal to AXtO+2 = abAPx,to+1 + a (1 - a) bAP, to, and so on, where abAPx,to+, is the short-run effect in time to + 2 and a (1 - a) bP,t, is the effect at to + 2 of the price change at to. In other words, the total effect on Xt is the sum of effects of past price changes, with the effect of any past price change declining with time. 84 Output, Consumption, and Trade Effects of Price Intervention If AP,, = AP is constant, then AXt converges to bAP over time. But in reality, the change in P,t caused by the elimination of direct and indirect price interventions is not constant over time because world prices vary over time (and the effects of price controls, export taxes, and other policies often vary directly with the world price) or because the equivalent tariff or the exchange rate overvaluation varies over time (because of oil price shocks or other external changes). Equation 4A-10 allows us to trace the dynamic output effects of a policy change that results in price changes that vary over time. Next we derive the equation corresponding to equation 4A-10 for AX, for the case of substitute crops Y with price P (j = 1, . . . J). We assume that the dynamic effects of price changes are caused by adjustment costs and that farmers have cobweb-type expectations (Pe = P-) for P and P . Then I X* = a + bP,t + cpe,t j-1 pe, =p1 (4A-12) pe = p Y,t-1 U7 , --1) X, - Xt-1 = a(X*'- X'_J), 0 < a < 1, where P is the price of substitute crop Y., and ci is the cross-price long-term elasticity of X, with respect to Pg (since all variables are in logs). Equation 4A-12 can be rewritten as I (4A-13) Xt = ea + abP,t_+a c P. t-1 + (1-a)Xt-, j=1 which is equivalent to (4A-14) X =a + aZ a)i-1b4,, + Lc iY't- 1=1 - ~ j1 -i The parameters a, b, and cj can be estimated from equation 4A-13, or they can be obtained from other studies. Then, the impact on X, of a price policy change starting at time to can be derived from equation 4A-14 as t-t (4A-15) AX, = aE '(1 - a)i-(b1R,ti + LcjAy,t-i Equivalently, it can be obtained from equation A-13 as (4A-15') AX= a(bAPx,t-1 + L CAPy )t-1 + (1 - a) AXt-1, (=1 Y Supply Response and Growth in Agriculture and the Economy 85 where the first term on the right side, a[., is the short-term (one- period) effect. We assume now that the share of purchased inputs z in the value of output is not negligible. There are two ways of incorporating that element in the analysis. First, if we know the elasticity (d) of X* with respect to the input price, Pz,,-, (where P is the log of the price of z relative to PNA), then equations 4A-15 and 4A-15' can be adjusted accordingly. Then AX, is given by t-t, (h (4A-16) AXt = a (1 - a)i-1 bdPx,t_i+ -1 L Pt-_ + dAP,t_i i=1 j=1 or equivalently by (4A-16') AXt = Abx,t-i + L CA,_1 + dAP + (1 - a) AXI_1, ( j=1 y where the first term of right side c[.] is the short-term (one-period) effect.23 Second, if the parameter d is not known, then value added (VA) should be used for X and for all the yj terms (for all crops). Then AP and AP should be replaced in equations 4A-15 and 4A-15' by AVAx and AVAY, and the elasticities b and cj should be replaced by vxb and vj c,24 respectively, where v is the share of value added in the value of output, so that vx = VAxIP, and vj = VA /P1 25 In this case, equations 4A-15 and 4A-15' become, respectively, t-t, I (4A-17) AXt = ae (1 - a)i- xbAVAx,t_i + LvjcAVAYP-i j=1 j=1 .-1 and (4A-17') AX, = a vxbAVAx,t-1 + LvxcAVA + (1-a)AXt-1, I1=1 where the first term on the right side, a[.], is the short-term (one- period) effect. Consumption and Trade Effects of Intervention By affecting agricultural producer and consumer prices (chapter 2) and production (chapter 4), price interventions have also had an impact on the consumption of agricultural products and on foreign exchange earnings. Where direct price interventions increased con- sumer food prices, as in Brazil, Ghana, Korea, Portugal, and Turkey, they reduced consumption. Where food prices were subsidized through direct interventions, as in Egypt, the Philippines, Sri Lanka, Thailand, and Zambia, they increased consumption. Indirect inter- ventions, which reduced consumer food prices in all countries, increased consumption. The foreign exchange effect depends on how consumption and output were affected and how output changes affected imports of tradable inputs. Direct price interventions created an antitrade bias on average by taxing agricultural exports and food imports and pro- tecting domestic food production. In the mid-1970s, direct price inter- ventions reduced foreign exchange earnings from agriculture because high world prices led to high negative rates of direct protection in those years. The foreign exchange effect of indirect interventions was almost always negative because indirect interventions taxed the pro- duction and subsidized the consumption of agricultural products, lowering relative agricultural prices for both producers and consumers.' Consumption Effects Appreciation of the real exchange rate reduces the price of food prod- ucts,2 leading to a fall in domestic output and an increase in con- sumption, including consumption of imported foods at the expense of domestic production. This effect was particularly strong in the 1970s, when real exchange rates appreciated substantially because of increased foreign aid, higher revenue from oil exports as prices rose, and macroeconomic policies. In some countries, direct price interven- 86 Consumption and Trade Effects of Intervention 87 tions, such as food subsidies, also led to increased consumption of imports. Together, indirect and direct price interventions made imports cheaper and biased incentives against the production of tra- ditional crops. Other factors, including greater urbanization and the increased cost of time, also influenced these changes in consumption patterns: imported food products are often easier and less time- consuming to prepare. Between 1966-70 and 1976-80, per capita consumption of tradi- tional foods in West Africa barely grew (0.27 percent year for maize) or fell (1.5 percent a year for millet and 1.69 percent for sorghum), while per capita consumption of imported foods grew rapidly (2.8 percent a year for rice and 8.5 percent for wheat products; [World Bank 1986: 92]). Two factors accounted for this change: depressed terms of trade for agricultural products and a smaller price difference between imported and domestically produced food products (coarse grains) in West Africa than in world markets. The world price of rice was three times that of sorghum on average over that period, but in West Africa rice was rarely more than two times the price of sorghum, and sometimes the prices were the same. Wheat flour was often twice the price of maize in developing countries with free trade policies, but the prices were nearly equal in C6te d'Ivoire and Nigeria. Specialization in the diet in favor of imported food products increased over time, and tastes adjusted to the new imported foods, thereby increasing the dependence on imports. Consequently, reform of incentives policies has become more urgent over time as some of these effects may become increasingly difficult to reverse. The relation between changes in food prices and changes in food consumption and imports was examined in a study on wheat by the International Center for the Improvement of Maize and Wheat (CIMMYT 1983). The study looked at changes in bread prices and their relation to wheat consumption and imports between 1970 and 1980 in fifty developing countries-including most of the eighteen countries studied in this project (but excluding Chile, Morocco, Portugal, and Sri Lanka). It found a strong negative relation between changes in food prices and changes in imports and consumption (table 5-1). Countries that export food products have tended to tax their pro- duction through direct interventions, increasing consumption above the level that would have prevailed in the absence of these price interventions. For wheat in Argentina, direct price interventions increased consumption an average of 2.2 percent in the 1960s, 7.5 percent in the 1970s, and 2.5 percent in the early 1980s. For rice in Thailand, the effects averaged 4.8, 4.3, and 2.7 percent respectively in the three decades. The effects on consumption were similar when food imports were subsidized, as with wheat in Egypt. But where food imports were taxed, as they were in many countries, consump- 88 Output, Consumption, and Trade Effects of Price Intervention Table 5-1. Effects of Changes in Bread Prices on Wheat Consumption and Imports in Fifty Developing Countries, 1970-80 (annual percentage change) Wheat Price of bread consumption Wheat imports More than 5 0.1 -11.5 3 to 5 0.1 -4.4 0 to 3 0.7 -3.7 -3 to 0 2.1 -1.9 -5 to -3 3.2 4.9 More than -5 3.5 11.7 Source: cIMMYr (1983). tion fell. In Colombia, direct price interventions decreased wheat con- sumption in all three decades: by 16.3 percent in the 1960s, 4.3 per- cent in the 1970s, and 5.8 percent in the 1980s. For rice in Korea in those three decades, consumption was depressed by 0.6 percent, 11.3 percent, and 14.2 percent. The effect on rice consumption in C6te d'Ivoire and Ghana was also negative. Although the consumption effect of direct price interventions was often negative, it was positive in most cases in the 1970s because consumer prices were kept below the high world levels during the commodity boom. In Chile, direct interventions reduced consump- tion in the 1960s and 1980s (by 0.8 percent and 1.7 percent) when wheat was protected, but increased it (by 2.7 percent) in the 1970s when direct interventions taxed production. The same pattern affected cotton in Pakistan; for sugar in Thailand, consumption was reduced by 4.5 percent in the 1960s, increased by 7.5 percent in the 1970s, and reduced again by 34 percent in the 1980s. Similarly, sugar consumption in the Dominican Republic was severely depressed by direct price interventions in the 1960s (by more than 50 percent) but increased (by 25 percent) in the 1970s. For some products, such as basmati rice in Pakistan, direct interventions reduced consumption only in the 1960s (by 8.9 percent); in other periods, the effect was to boost consumption, by 2.7 percent in the 1970s and 12 percent in the 1980s. Wheat in Pakistan followed the same pattern. Indirect interventions, which lowered the relative consumer prices of agricultural products through overvaluation of the real exchange rate and industrial protection, had a positive effect on consumption in all cases. Even where direct price interventions reduced consump- tion, the positive effect of indirect interventions was usually so large that the total effect on consumption was positive. For wheat in Colombia, the direct and total effects on consumption were, respec- Consumption and Trade Effects of Intervention 89 tively, -16.3 and -3.9 percent in the 1960s, -4.3 and 9.6 percent in the 1970s, and -5.8 and 14.2 percent in the 1980s. For rice in the Philippines, the respective effects were -9.4 and -0.5 percent, 1.5 and 9.2 percent, and -0.6 and 9.8 percent. And for cotton in Turkey, they were -8.4 and 2.4 percent, -10.3 and 0.4 percent, and -2 and 8.3 percent. Trade and Foreign Exchange Effects In calculating the foreign exchange effects of intervening in the prices of the agricultural commodities3 examined in this study (table 5-2), we considered three components: the effect of changes in out- put (change in quantity produced times the border price); the effect of changes in consumption (change in quantity consumed times the border price); and the effect of changes in the use of tradable inputs due to the change in output (change in quantity of inputs imported times the border price). We calculated the foreign exchange effects of direct and total interventions were calculated for both the long-run or cumulative output effect and the short-run output effect (see chapter 4). The foreign exchange effects, which are measured as a percentage of total exports, were larger for total price interventions than for direct price interventions because of the especially strong effect of indirect interventions, particularly including exchange rate over- valuation. Other studies have also found that agricultural exports and net exports are highly responsive to the real exchange rate because it affects agriculture's relative price. For a group of fifty-two developing countries in 1965-82, Balassa (1987) found that the response of agri- cultural exports (the agricultural export-to-output ratio) to changes in the real exchange rate was generally large for the group as a whole (an elasticity of 0.68) and even larger in Sub-Saharan Africa (1.35). The response was stronger still when the net (exports-minus-imports) exports-to-output ratio in agricultire was used-an elasticity of 4.96 for the group as a whole and 11.47 for the sixteen Sub-Saharan Afri- can countries. Thus, macroeconomic and trade policies that cause appreciation of the real exchange rate will have a negative impact on exports and on net agricultural exports and will reduce foreign exchange earnings. Size of the Foreign Exchange Effect On average for the entire period, the foreign exchange effect of direct price interventions was negative, equal to a loss in the value of net trade of about 2.5 percent of actual exports for the short run and 7.8 percent for the long run. As with the output and consumption effects, 90 Output, Consumption, and Trade Effects of Price Intervention Table 5-2. Foreign Exchange Effects for All Commodities Relative to Total Export Earnings (percent) Direct Total Country Period Short-run Long-run Short-run Long-run Group I C6te d'Ivoire 1960-69 -8.68 n.a. -13.68 -15.64 1970-79 -10.87 n.a. -20.25 -21.71 1980-82 -7.61 n.a. -20.19 -22.50 1960-82 -9.50 n.a. -17.39 -19.17 Ghana 1963-69 9.43 7.50a -2.42 -7.63 1970-79 15.50 29.10 1.40 -18.80 1980-85 279.67 -149.17 -12.00 -319.40 1963-85 82.57 -46.92 -2.86 -80.26 Zambia 1966-69 -0.40 -0.43 -0.68 -0.78 1970-79 -1.99 -4.52 -2.47 -7.73 1980-84 -2.56 -5.00 -1.70 -12.64 1966-84 -1.81 -3.78 -1.89 -7.56 1960s 0.12 3.54 -4.79 -8.01 1970s 0.88 -16.81 -7.11 -16.08 1980s 89.83 -77.08 -11.30 -118.18 Whole periods -5.65 -3.78 -9.64 -13.37 Group II Argentina 1960-69 n.a. -52.64 n.a. -103.30 1970-79 n.a. -40.28 n.a. -66.56 1980-85 n.a. -15.59 n.a. -80.22 1960-85 n.a. -39.34 n.a. -83.84 Colombia 1961-69 4.53 10.19b -0.41 -6.80b 1970-79 -6.50 -12.48 -10.95 -27.03 1980-83 1.90 1.59 -1.55 -6.15 1961-83 -0.72 -6.13 -5.19 -19.28 Dominican 1966-69 24.92 26.83 2.15 2.20 Republic 1970-79 -4.77 3.81 -15.05 -14.95 1980-84 -3.05 -4.05 -14.49 -19.32 1966-84 1.93 6.59 -11.28 -12.49 Egypt 1965-69 n.a. n.a. -15.25 n.a. 1970-79 n.a. n.a. -25.66 n.a. 1980-85 n.a. n.a. -24.19 n.a. 1965-85 n.a. n.a. -22.76 n.a. Morocco 1961-69 n.a. n.a. 2.78 -6.89 1970-79 n.a. n.a. -17.60 -29.40 1980-84 n.a. n.a. -11.40 -18.00 1961-84 n.a. n.a. -8.67 -17.75 Consumption and Trade Effects of Intervention 91 Table 5-2 (continued) (percent) Direct Total Country Period Short-run Long-run Short-run Long-run Pakistan 1963-69 20.57 14.36 -53.42 -128.77 1970-79 -41.22 -58.63 -78.52 -175.89 1980-86 -18.83 -54.43 -51.33 -128.04 1963-86 -16.67 -36.11 -63.27 -148.19 Philippines 1960-69 7.53 7.40 -2.51 -5.06 1970-79 -3.17 -11.15 -7.11 -13.90 1980-84 3.19 -1.34 12.49 11.25 1960-84 2.38 -1.77 -1.35 -5.34 Sri Lanka 1960-69 -17.31 -12.85 -19.17 -17.88 1970-79 -10.10 -7.56 -13.63 -14.97 1980-85 -7.59 -11.12 -9.99 -15.85 1960-85 -12.29 -10.42 -14.92 -16.30 Thailand 1961-69 -17.98 -36.09c -21.95d -43.88e 1970-79 -6.22 -15.03 -12.05 -30.43 1980-84 2.40 0.41 0.37 -4.43 1961-84 -8.83 -16.44 -12.38 -25.45 Turkey 1962-79 4.36 -15.17 -16.54 -32.40 1970-79 4.18 -30.79 -12.26 -43.94 1980-83 -6.70 -33.28 -14.45 -49.38 1962-83 2.27 -26.06 -14.21 -41.13 1960s 3.80 -7.24 -13.81 -38.09 1970s -9.68 -21.51 -21.42 -46.34 1980s -4.10 -14.73 -12.73 -34.02 Whole period -4.56 -16.21 -17.11 -41.08 Group III Brazil n.a. n.a. n.a. n.a. n.a. 1973-79 -7.81 -3.13 n.a. n.a. 1980-83 -11.00 -12.09 n.a. n.a. 1973-83 -8.97 -6.39 n.a. n.a. Chile 1960-69 n.a. -1.10 n.a. -8.56 1970-79 n.a. -2.25 n.a. -2.86 1980-82 n.a. -0.03 n.a. -0.97 1960-82 n.a. -1.46 n.a. -5.09 Malaysia 1961-69 -0.60 -3.98f -1.63 -6.43f 1970-79 -.02 -3.55 -2.17 -6.42 1980-83 0.34 -2.17 -0.47 -4.58 1961-83 -0.62 -3.50 -1.67 -6.12 1960s -0.60 n.a. -1.63 -7.49 1970s -4.41 -2.98 -2.17 -4.64 1980s -5.33 -4.77 -0.47 -2.77 Whole period -4.79 -3.78 -1.67 -5.60 (Table continues on the following page.) 92 Output, Consumption, and Trade Effects of Price Intervention Table 5-2 (continued) (percent) Direct Total Country Period Short-run Long-run Short-run Long-run Group IV Korea, Rep. of 1962-69 53.03 83.26 -76.86 -65.33 1970-79 8.88 13.05 0.14 -1.36 1980-84 6.04 8.42 4.54 6.36 1962-84 23.62 36.47 -25.69 -21.93 Portugal 1960-69 -0.48 4.29 5.96 4.18 1970-79 -8.36 -6.66 -8.98 -11.68 1980-84 -1.26 0.34 -1.94 -4.04 1960-84 -3.79 -0.88 -1.60 -3.81 1960s 26.27 43.78 -35.45 -30.57 1970s 0.26 3.20 -4.42 -6.52 1980s 2.39 4.38 1.30 1.16 Whole period 9.91 17.79 -13.64 -12.87 Averages 1960s 5.79 1.85 -15.09 -29.02 1970s -6.84 -12.80 -16.18 -31.26 1980s -3.44 -9.17 -9.59 -22.97 Whole period -2.54 -7.80 -14.45 -28.90 a. 1962-69. b. 1968-69. c. 1965-69. d. 1963-69. e. 1967-69. f. 1960-69. g. Average excludes Ghana. Source: Authors' calculations based on data provided in country studies. the direct foreign exchange effects were larger the more world com- modity prices varied relative to domestic prices, or the larger the gap between the two sets of prices. The largest gap (not shown in table 5-2) occurred during the commodity price boom of the early 1970s. In the Dominican Republic the short-run direct foreign exchange effect fell from a gain of about 6 percent of total exports in 1972 to losses of 13 percent in 1973 and 16 percent in 1974. In Pakistan losses increased from 36 percent in 1972 to 82 percent in 1973, implying that had domestic prices for wheat, cotton, rice, and sugar been allowed to rise with world prices in 1973, the combined effect of lower consumption and higher production would have led to an increase in the value of net trade equal to about 80 percent of actual export earnings. The foreign exchange effects of indirect price interventions were invariably negative, since indirect interventions led to a fall in the relative prices of agricultural products for both producers and con- Consumption and Trade Effects of Intervention 93 sumers.4 The foreign exchange effects of total price interventions were thus more severe than those of direct interventions (see table 5-2). For instance, in the 1970s in Argentina, the long-run direct for- eign exchange effect was a loss in the value of net trade of about 40 percent of actual exports, whereas the long-run total effect was a loss of 67 percent. In the Dominican Republic in 1980-84, the short-run direct foreign exchange effect was a loss of 3 percent, and the total effect was a loss of 14.5 percent. In Turkey in the 1960s, the short-run direct effect was a gain of about 4 percent, but the total effect was a loss of about 17 percent. How large the foreign exchange effects are in relation to total exports depends on both the rate of protection or taxation and the relative importance of agricultural trade in total exports. In Argen- tina, which had high rates of direct and indirect agricultural taxation and agricultural exports that contributed from 50 to 70 percent of total exports between 1960 and 1985, the foreign exchange losses from the long-run impact of direct intervention on agricultural trade were large-from a low of about 15 percent of total exports in 1980-85 to a high of 52 percent in the 1960s. The losses as a result of total interven- tion were even higher, from a low of about two-thirds of total exports in the 1970s to 103 percent in the 1960s, for an average of more than 80 percent for 1960-85. In Ghana, too, the foreign exchange losses were large, and they increased dramatically in the early 1980s because of the very high rate of indirect taxation in those years and because total exports (the base in our calculations) fell precipitously from about 20 to about 30 per- cent of agricultural GDP in the 1960s and 1970s to 3 percent by 1982. The total long-run foreign exchange losses, which include the cumulative impact of the tax on tree crop stocks, were much larger than the short-run losses, 320 percent versus 12 percent of total exports for 1980-84. That means that foreign exchange revenues would have been four times larger in 1980-84 had there been no price interventions during 1960-84. In Korea foreign exchange losses as a percentage of total exports diminished over time despite increasing rates of direct protection because of the rapid increase in manufacturing exports and the fall in the rate of indirect taxation. The total long run foreign exchange effect went from substantial losses in the 1960s to a slight gain in the early 1980s. In Pakistan the foreign exchange effects of price interventions were large and negative on average over the entire period. For individual periods, the effects ranged from a direct short-run gain of about 20 percent in the 1960s (because of extremely high protection for sugar) to a total long-run loss of about 176 percent in the 1970s (mainly because of high direct taxation beginning in 1972-73). Total interventions led to foreign exchange losses in all periods, although 94 Output, Consumption, and Trade Effects of Price Intervention the losses were slightly lower in the 1980s as the high rate of indirect taxation fell slightly (from about 40 percent in the 1960s, to 26 percent in 1980-86). For some countries the foreign exchange effects were small. In Chile foreign exchange losses averaged only 1.5 percent for the direct long-run effect and 5 percent for the total long-run effect. In Malaysia, direct short-run foreign exchange losses averaged about 1 percent of total exports and total short-run losses about 2 percent. The average loss in Portugal was about 4 percent or less of total exports for both direct and total short- and long-run effects. Losses were also low in Zambia, where agriculture's share in total exports is less than 1 per- cent, and in Colombia in all periods except the 1970s. For direct price interventions, the cross-country average short-run foreign exchange effect was small in all three subperiods. Losses were highest (7 percent of total exports) in the 1970s, when domestic prices were kept below world prices for several years in many countries, and lowest (3.5 percent) in the 1980s.5 For indirect price interven- tions, average short-run losses were larger in each subperiod, but they fell over time, from about 20 percent in the 1960s to 6 percent in the early 1980s, bringing the average short-run total foreign exchange losses down as well, from about 15 and 16 percent in the 1960s and 1970s to less than 10 percent in the early 1980s. Although the direct effect was small on average, it was large in some countries. It was 15 percent (in absolute value) in Argentina (long-run) and Pakistan in all periods, in the Dominican Republic in 1966-69, in Ghana after 1970, and in Korea, Sri Lanka, and Thailand in the 1960s. The short-run direct foreign exchange effects were lower in the 1980s than in earlier periods (in absolute value) on average for the eighteen countries and in Colombia, C6te d'Ivoire, the Dominican Republic, Korea, Malaysia, Pakistan, Philippines, Portugal, Sri Lanka, and Thailand. The average total foreign exchange effects were larger than the direct ones. Total short-run losses were above 15 percent of total exports in the 1960s and 1970s on average in the eighteen countries and in Egypt and Pakistan. They were also above 15 percent in the 1960s in Korea, Sri Lanka, Thailand, and Turkey, and above 15 per- cent in the 1970s in C6te d'Ivoire, the Dominican Republic, and Morocco. Like the direct short-run foreign exchange effects, the aver- age total effects were lowest in the 1980s. This was also the case in Chile (long-run), Colombia (long-run), Malaysia, Pakistan, Portugal, Sri Lanka (short-run), and Thailand. Impact on the Current Account and the Gross Domestic Product How important were the foreign exchange effects in relation to the current account deficit and the GDP rather than as a share of total Consumption and Trade Effects of Intervention 95 exports? For some countries, the foreign exchange effect of price interventions affecting agriculture was substantial. In Egypt in the 1970s, the current account deficit was about 16 percent of the GDP, and total exports were 24 percent of the GDP. Removing total interven- tions would have had a positive short-run foreign exchange effect of about 26 percent of exports, reducing the current account deficit by 39 percent, or by 6 percent of the GDP. For 1980-85, removal of total interventions would have led in the short-run to a 43 percent reduc- tion in the current account deficit. Ghana's exports in the 1970s were 19 percent of its GDP, and its trade surplus was 1.9 percent of its GDP. The long-run total foreign exchange effect of price interventions was a loss in the net value of trade of 18.8 percent of exports or 3.6 percent of the GDP, so removing total interventions would have increased the trade surplus to 5.5 percent of the GDP in that decade. In 1980-84, the trade deficit was 0.6 percent of the GDP. The long-run foreign exchange effect of total inter- vention in that period was a loss in the net value of trade of 319 percent of exports. Had total interventions been removed at the start of the period, Ghana would have enjoyed a large trade surplus rather than a small trade deficit. Pakistan in the 1970s had exports worth 11.4 percent of its gross national product (GNP) and a trade deficit of 6.6 percent of its GNP. The short-run foreign exchange effect of direct intervention was a loss in net trade of 41.2 percent of exports or 4.7 percent of the GNP, while the short-run total effect was a loss in net trade of 78.5 percent of exports or about 9 percent of the GNP. Removing direct interventions would have reduced the trade deficit from 6.6 to 1.9 percent of the GNP, removing total interventions would have led to a trade surplus of 2.4 percent of the GNP. In 1980-86, when exports were 11.6 percent of the GNP and the trade deficit was 10.3 percent of GNP, the total short-run effect was a loss in net trade of 51.3 percent of exports or about 6 percent of the GNP. Removing total interventions would have reduced the deficit from 10.3 to 4.3 percent of the GNP. Conclusion Agriculture is a highly tradable sector. Our findings indicate that interventions that have taxed agriculture have had a substantial impact on foreign exchange earnings in a large number of countries. Hence, removing those interventions would allow agriculture to make a significant contribution to a relaxation of the foreign exchange constraints and to the current account position in those countries where agriculture has an important share in the GDP and exports. Part III Consequences of Price Intervention for Income Transfers and Equity The Fiscal Impact of Direct Intervention In previous chapters we discussed the consequences of price inter- ventions for agricultural production, consumption, and trade flows. But agricultural pricing policies also have other, wider economic implications through their influence on the government budget, farmers' income, and the real income of urban consumers. All three are likely to be affected by any reform of agricultural price policies. This chapter on budget effects is the first of three that discuss the broader economic implications of price interventions and assess their consequences for income transfers between agriculture and govern- ment, between agriculture and the rest of the economy, and between rural and urban households. In it we identify and quantify the most important budget effects of price policies to see whether they resulted in net outlays or revenues for the government. In the process, we look at changes in these policies over time and at the likely impact of price policy reform on the government budget. In chapter 7 we link these results on budget effects with the intersectoral income transfers between agriculture, the government, and the nonagricultural sector that arise from price interventions. In chapters 7 and 8 we also address the issue of winners and losers-which sectors and income groups were the main beneficiaries of the income transfers resulting from price interventions and which were the losers. This analysis helps clarify the motivations underlying price interventions and the political and economic constraints on policy reform. Many of the fiscal effects of specific agricultural price policies have received considerable attention and have generated intense debate, but there have been few if any systematic attempts to quantify the net fiscal impact of price interventions. Consumer food subsidies, for example, have frequently been cited as a major drain on government budgets. Our analysis shows, however, that although this may be true for some food-importing countries, it is not the case for most countries. Similarly, many countries have a tradition of subsidizing agricultural credit and inputs, which is conventionally interpreted as 99 100 Consequences of Price Intervention for Income Transfers and Equity a means of compensating producers for the heavy taxation of agri- cultural production (the compensation issue is discussed in detail in chapters 7 and 8). Yet our analysis shows that such input measures represented only a small fraction of government expenditure, averag- ing only about 2 percent over 1960-83, while taxation of agricultural exports yielded substantial revenue to the government, averaging 10 percent, as a share of government expenditure, over the period. Indeed, the net effect of direct price interventions in agriculture was a revenue gain to the government of nearly 7 percent over the period- reaching as high as 17 percent during the 1960s. How Price Interventions Affect the Budget Most direct agricultural price interventions affect the government accounts. The effects of some interventions, such as outlays for agri- cultural input and production subsidies or revenues from explicit border taxes, are obvious. The effects of others, such as the effects of price interventions on the operating deficit or surplus of state-owned marketing boards, are more complex or are hidden. Whatever the nature of the price intervention, however, it is likely to have some fiscal impact. And that fiscal impact will, in turn, influence agri- cultural price policy. This dynamic relationship between price policy and fiscal policy is especially clear in the case of agricultural export policy and food and input subsidies. The government's need for revenue to fund expand- ing development programs was probably the major impetus behind taxation of agricultural exports, and it remains a major obstacle to reform. The government revenues deriving from the export opera- tions of the Cocoa Marketing Board in Ghana and from the export tax on grains and beef in Argentina are examples of the importance of the revenue motive and the obstacles it poses for reform. In some coun- tries, the fiscal burden of input and food subsidies escalated so rap- idly that it led to macroeconomic imbalances that could be corrected only through policy reform. Avillez, Finan, and Josling (1990) assert that changes in agricultural policies in Portugal were dictated chiefly by budget pressures; fertilizer and feed subsidies were essentially eliminated in 1983 because of severe fiscal stringencies. Attempts to reduce food subsidies in Egypt, Morocco, and Pakistan also illustrate this link between price policy and the budget. In most industrial countries the major impetus for reform of agri- cultural programs now comes from fiscal pressures, although income transfer seems to be the primary objective of agricultural policies in these countries (OECD 1987). In developing countries, the links between price policies and the government budget are different, aris- ing mainly through the contribution of agricultural exports to fiscal The Fiscal Impact of Direct Intemention 101 revenues and the cost of government outlays for food subsidies to urban consumers. This analysis of the budget effects of price policies considers only direct price interventions in agriculture. But these effects on budget expenditures and revenues are only a part of the overall income gains and losses to agriculture from price interventions. Indirect interventions-industrial protection and overvaluation of the exchange rate-generate income transfers from agriculture to other sectors, but they are off-budget items that yield no revenue to the government.' For that reason they are not considered in the analysis of budget effects. Taxes that apply to other sectors as well as to agri- culture, such as income taxes and value-added taxes, have no direct link with agricultural price policies and would continue independent of agricultural price reforms. These too were excluded from the analysis. There are, however, some indirect links between macroeconomic variables and the budget that could influence the design of direct price policy interventions. Most of these are quite difficult to measure and were not included in our calculations of the budget effects of price interventions because they were not quantified by most of the country study authors. One link involves the nominal exchange rate. As Jansen (1988) illustrates for Zambia, an overvalued exchange rate, especially since 1980, has resulted in substantial fiscal savings in gov- ernment operations involving maize and fertilizer imports, given the government's target domestic prices for consumers. Another complex link sometimes arises between farm prices and the government budget in middle-income countries, where agricul- ture is usually a relatively small sector. Hurtado, Vald6s, and Muchnik (1990) argue in their study on Chile that government lend- ing at concessional rates to the nonfarm private sector was at least partly a compensating mechanism to offset the effects of its agri- cultural policies. The credit program was designed to prevent a decline in nonfarm investment as a result of the substantial rise in nominal urban wages during the 1960s-a consequence of higher farm prices and downward rigidity in urban real wages because of strong trade unions. Measures of the Fiscal Effects of Price Interventions: Classification of Expenditures and Revenues A simple taxonomy of the fiscal implications of farm policies suggests the categories of revenues and expenditures presented in table 6-1. Some price interventions require governments to spend money, and some generate revenue. We are interested here in both these individ- ual effects and their net effect on the budget. It is also useful to 102 Consequences of Price Intervention for Income Transfers and Equity Table 6-1. Classification of Government Revenues and Expenditures Related to Agriculture Category Examples Expenditures Explicit outlays Input subsidies (fertilizer, machinery and machinery operations, pesticides, power tariffs for tubewells) Credit subsidies (interest-free or low-interest loans) Marketing subsidies (storage and railway transportation) Export subsidies (financing, drawback, and others) Consumer subsidies (products sold at less than the market price) Direct product subsidies on output Concealed subsidiesa Subsidies on irrigation (operation and maintenance) Concessional rates on electricity use Operational losses Losses by state-owned companies or parastatals involved in procurement and trade (net of export and of consumer subsidies listed above) Public investment on Agricultural research and extension infrastructurea Construction of irrigation systems, dams, and processing facilities Land reclamation projects Other programsa Agrarian reform programs and others Revenues Direct taxes Land tax Agricultural income taxb Explicit commodity Explicit border taxes on imports and exports of taxes farm products, including intermediate inputs for agricultural production Production taxes on specific commodities Sales taxesb (value added taxes in some countries) Operational gains Profits of state-owned companies or parastatals involved in procurement and trade (net of commodity border taxes listed above) a. These categories are not included in the analysis in this chapter either because of insufficient data (subsidies on irrigation and electricity use) or because they are not price-related interventions. b. Includes the difference between the rates that apply to agriculture and those in the rest of the economy. Source: Classification by the authors. The Fiscal Impact of Direct Intervention 103 distinguish between government outlays and revenues involving producers and those affecting consumers of farm products. Expenditures Estimates of government expenditures related to price policy were placed into two categories: those arising from input policies-credit subsidies and other input subsidies-and those arising from output policies-food subsidies to consumers, production subsidies, and losses of state-owned agencies (table 6-1).2 Full coverage of expendi- ture accounts would include both explicit outlays, such as input and production subsidies, and implicit subsidies, such as concessional rates on electricity and irrigation, public investment on infrastructure, and expenditures on land reform programs. Many of these implicit subsidies, such as agrarian reform programs, public investment (analyzed in chapter 7), and marketing subsidies on storage opera- tions, were excluded from the computations in this chapter because they are not related to price interventions. The implicit subsidy on the operation and maintenance of irrigation systems was excluded because of incomplete information.3 Revenues Government revenues related to price policy were grouped into three categories: export taxes, import taxes, and the profits of state agen- cies. Export taxes include the tax implicit in the sale of agricultural exports by state-owned firms at prices higher than those paid to farmers. Import taxes include explicit border taxes on agricultural imports. In most countries food imports are exempt from border taxes or are a monopoly of state agencies that are exempt from import duties; many other agricultural imports are protected by quotas rather than by tariffs and so do not generate government revenue. The implicit gain from duty-free imports by state agencies was gener- ally classified under the "state agency" category, as were any other profits of state agencies not allocated directly to another category. (For more detail about how the budget effects of price interventions were categorized and calculated, see the appendix to this chapter.) In this quantitative analysis of budget effects, all impacts are reported as a percentage of total government expenditure to provide a common unit of measure. Budget Effects of Direct Price Interventions The pricing policy has had a substantial impact on government reve- nues and budgetary expenditures. Explicit and implicit agricultural export taxes had the single largest impact on the government budget 104 Consequences of Price Intervention for Income Transfers and Equity (table 6-2). The revenue from agricultural export operations ran as high as 20 percent of total government expenditure in 1960-69 and averaged 10 percent for the whole period for the sixteen countrieS4 for which data were available. Over time, however, the trend was clearly declining, with the percentage falling to 11 percent in 1970-79 and to 6 percent in 1980-83. There were wide differences across countries in the revenue effect of export taxes. In Sri Lanka that share was equivalent to 78 percent of government expenditure in 1960-69, but it declined to 57 percent in 1980-85. In Ghana the revenues earned through the operation of the Cocoa Marketing Board were between 20 and 30 percent of govern- ment expenditure in most years. The strong revenue effect of agri- cultural export taxes was also evident in Argentina, Brazil, and Col- ombia, where it represented about 10 percent or more of total government expenditure. At the opposite extreme were Chile, the Dominican Republic, Korea, Morocco, Pakistan, Turkey, and Zambia, where the contribution of export taxes was slight (2 percent or less). In the middle, with revenue contributions averaging 5 to 7 percent, were Egypt, Malaysia, the Philippines, and Thailand. The declining trend in the ratio of export tax revenues to government expenditure is quite clear in this group of countries, and in Sri Lanka as well. The next-largest budget effects came from consumer food subsidies and credit subsidies, but their impact fell far short of that of export taxes. Consumer food subsidies averaged about 2 percent of govern- ment expenditure and credit subsidies about 1.8 percent.5 Although these effects are quite small on average for the sixteen countries- especially considering how frequently the price policy literature refers to the high fiscal burden of these two items-the outlays were sub- stantial in some countries. In Zambia, as a result of the government maintaining a low maize consumer price (below import parity and below the producer price) and as a result of maize-handling subsidies (a consequence of the government's uniform pricing policy across surplus and deficit areas), substantial subsidies went to the leading parastatal and cooperatives (Jansen, 1988). In Brazil agricultural credit subsidies increased from approximately 9 percent of total government expenditure in 1966-69 to 34 percent in 1980-83. In Colombia they averaged 4 percent over 1960-83, and in Turkey they were as high as 6 percent in 1980-83, although the average for 1963-83 was only 1.1 percent. In most of the other countries, however, credit subsidies had very little impact-less than 1 percent of total government expendi- ture. Government outlays for consumer subsidies also had little bud- getary impact in most countries, but they were substantial in Sri Lanka (from 5.1 to 13.4 percent), Brazil (6 to 12.6 percent after 1970), Egypt (3.6 to 4.5 percent), Morocco (5.1 to 6.5 percent), and Pakistan (2 to 4.8 percent). Table 6-2. Budget Effects of Direct Price Interventions, 1960-84 (revenues [ +I and expenditures [-1 as percentage of total public expenditures) Output policies Input policies Consump- Total Credit Input Total Production tion output sub- subsi- input State Net Export Import subsidies subsidies effects sidies dies effects agency- effects Country Period tax (1) tax (2) (3) (4) (5) (6) (7) (8) (9) (10) Cte d'Ivoireb - - - - - - Ghana 1960-69 20.6 0.0 0.0 0.0 20.6 0.0 0.0 0.0 0.0 20.6 1970-79 26.4 0.0 0.0 0.0 26.4 0.0 0.0 0.0 0.0 26.4 1980-84 0.5 0.0 0.0 0.0 0.5 0.0 0.0 0.0 0.0 0.5 1960-84 18.9 0.0 0.0 0.0 18.9 0.0 0.0 0.0 0.0 18.9 Zambia 1960-69 - - - - - 1970-79 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -6.0 -6.0 1980-84 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -8.7 -8.7 1970-84 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 -6.9 -6.9 Group I 1960s 20.6 0.0 0.0 0.0 20.6 0.0 0.0 0.0 0.0 20.6 averagec 1970s 13.2 0.0 0.0 0.0 13.2 0.0 0.0 0.0 -3.0 10.2 1980s 0.2 0.0 0.0 0.0 0.3 0.0 0.0 0.0 -4.3 -4.1 1960-84 9.4 0.0 0.0 0.0 9.4 0.0 0.0 0.0 -3.4 6.0 Argentinad 1960-69 10.5 0.0 0.0 0.0 10.5 0.0 0.0 0.0 0.0 10.5 1970-79 10.6 0.0 0.0 0.0 10.6 0.0 0.0 0.0 0.0 10.6 1980-85 9.0 0.0 0.0 0.0 9.0 0.0 0.0 0.0 0.0 9.0 1960-85 10.2 0.0 0.0 0.0 10.2 0.0 0.0 0.0 0.0 10.2 (Table continues on the following page.) Table 6-2 (continued) (revenues [+1 and expenditures [-] as percentage of total public expenditures) Output policies Input policies Consump- Total Credit Input Total Production tion output sub- subsi- input State Net Export Import subsidies subsidies effects sidies dies effects agencya effects Country Period tax (1) tax (2) (3) (4) (5) (6) (7) (8) (9) (10) Colombia 1960-69 9.9 0.6 0.0 0.0 10.5 -3.1 -1.0 -4.1 0.0 6.5 1970-79 15.2 0.6 0.0 0.0 15.9 -4.7 -1.0 -5.7 0.0 10.2 1980-83 8.9 0.4 0.0 0.0 9.3 -4.5 -0.5 -5.1 0.0 4.2 1960-83 12.0 0.6 0.0 0.0 12.5 -4.0 -0.9 -4.9 0.0 7.6 Dominican 1960-69 - - - - - - - - - Republic 1970-79 2.6 0.0 0.0 0.0 2.6 0.0 0.0 0.0 5.4 8.0 1980-84 1.1 0.0 0.0 0.0 1.1 0.0 0.0 0.0 1.0 2.1 1970-84 2.1 0.0 0.0 0.0 2.1 0.0 0.0 0.0 3.9 6.0 Egypt 1960-69 - - - - - - - - - 1970-79 6.7 0.0 -0.3 -4.5 1.9 0.0 0.0 0.0 0.0 1.9 1980-84 0.9 0.0 0.0 -3.6 -2.7 0.0 0.0 0.0 0.0 -2.7 1970-84 4.8 0.0 -0.2 -4.2 0.4 0.0 0.0 0.0 0.0 0.4 Morocco 1960-69 - - - - - - - - - 1970-79 0.0 0.0 0.0 -5.1 -5.1 -0.2 -0.2 -0.3 0.0 -5.4 1980-84 0.0 0.0 0.0 -6.5 -6.5 -0.1 -0.6 -0.8 0.0 -7.2 1960-84 0.0 0.0 0.0 -5.6 -5.6 -0.2 -0.3 -0.5 0.0 -6.0 Pakistan 1960-69 - - - - - - - - - - 1970-79 1.1 0.0 0.0 -4.8 -3.7 -0.2 -1.8 -2.0 1.0 -4.8 1980-83 0.4 0.0 0.0 -2.0 -1.6 -0.5 -2.5 -3.0 0.6 -4.0 1970-83 0.9 0.0 0.0 -4.0 -3.1 -0.3 -2.0 -2.3 0.9 -4.5 Philippines 1960-69 - - - - - - - - - - 1970-79 5.5 0.4 0.0 0.0 5.1 -0.3 -0.0 -0.3 -0.1 5.4 1980-84 1.2 0.0 0.0 0.0 1.2 -1.5 -0.5 -2.0 -0.7 -1.5 1960-84 4.1 0.2 0.0 0.0 4.3 -0.7 -0.2 -0.9 -0.3 3.1 Sri Lanka 1960-69 77.9 0.0 0.0 -11.1 66.8 -0.1 -1.7 -1.8 0.0 65.0 1970-79 72.6 0.0 0.0 -13.4 59.1 -0.6 -2.2 -2.8 0.0 56.3 1980-85 57.3 0.0 0.0 -5.1 52.2 -0.1 -1.9 -2.0 0.0 50.2 1960-85 71.1 0.0 0.0 -10.6 60.5 -0.3 -1.9 -2.2 0.0 58.2 Thailand 1960-69 12.3 0.0 0.0 0.0 12.3 0.0 -0.3 -0.3 0.0 11.9 1970-79 5.7 0.0 -0.5 0.0 6.1 0.0 -0.9 -0.9 0.0 4.3 1980-83 3.0 0.0 -0.7 0.0 3.7 0.0 -1.1 -1.1 0.0 1.2 1960-83 8.0 0.0 -0.3 0.0 7.5 0.0 -0.7 -0.7 0.0 6.9 Turkey 1963-69 0.0 0.0 0.0 0.0 0.0 0.0 0.6 0.6 -3.3 -2.7 1970-79 0.0 0.0 0.0 0.0 0.0 0.0 -1.4 -1.4 -2.1 -3.5 1980-83 4.0 0.0 0.0 0.0 4.0 -5.8 -1.4 -7.2 -0.5 -3.7 1963-83 0.8 0.0 0.0 0.0 0.8 -1.1 -0.7 -1.8 -2.2 -3.2 Group II 1960s 22.1 0.1 0.0 -2.2 20.0 -0.6 -0.5 -1.1 -0.7 18.2 averagec 1970s 12.0 0.1 -0.1 -2.8 9.2 -0.6 -0.8 -1.2 0.4 8.3 1980s 8.6 0.0 -0.1 -1.7 6.8 -1.3 -0.8 -2.1 0.0 4.8 1960-85 11.4 0.1 -0.0 -2.4 9.0 -0.7 -0.7 -1.3 0.2 7.9 (Table continues on the following page.) Table 6-2 (continued) (revenues 1+1 and expenditures [-] as percentage of total public expenditures) Output policies Input policies Consump- Total Credit Input Total Production tion output sub- subsi- input State Net Export Import subsidies subsidies effects sidies dies effects agency, effects Country Period tax (1) tax (2) (3) (4) (5) (6) (7) (8) (9) (10) Brazil 1966-69 39.0 0.0 0.0 2.1 41.1 -8.9 0.0 -8.9 0.0 32.2 1970-79 20.8 0.0 0.3 -6.0 14.8 -19.7 0.0 -19.7 0.0 -4.6 1980-83 3.1 0.0 0.5 -12.6 -9.1 -34.2 0.0 -34.2 0.0 -43.2 1966-83 20.9 0.0 0.3 -5.7 15.5 -20.5 0.0 -20.5 0.0 -5.0 Chilee 1960-69 -0.0 0.4 0.0 -0.2 0.1 -0.9 -0.1 -1.1 -0.0 -1.0 1970-79 -0.0 1.0 0.0 -0.4 0.6 -0.5 0.2 -0.3 -0.0 0.2 1980-82 0.0 0.6 0.0 -0.2 0.4 -0.4 0.4 0.1 0.0 0.4 1960-82 -0.0 0.6 0.0 -0.3 0.3 -0.7 0.1 -0.6 -0.0 -0.3 Malaysia 1960-69 8.3 0.0 -0.1 0.0 8.3 0.0 0.0 0.0 0.0 8.3 1970-79 7.4 0.0 -0.0 0.0 7.4 0.0 -0.1 -0.1 0.0 7.4 1980-83 3.0 0.0 -0.6 0.0 2.3 0.0 0.0 0.0 0.0 2.3 1960-83 7.1 0.0 -0.2 0.0 6.9 0.0 -0.0 -0.0 0.0 6.9 Group III 1960s 15.8 0.1 -0.0 0.6 16.5 -3.3 -0.0 -3.3 -0.0 13.2 averagec 1970s 9.4 0.3 0.1 -2.1 7.7 -6.7 0.0 -6.7 -0.0 1.0 1980s 2.0 0.2 -0.1 -4.3 -2.1 -11.5 0.1 -11.4 0.0 -13.5 1960-83 9.3 0.2 0.0 -2.0 7.6 -7.1 0.0 -7.0 -0.0 0.5 Koreaf 1960-69 - - - - - 1970-79 0.0 0.7 -1.2 -1.2 -1.6 -1.6 -1.0 -2.6 0.0 -4.2 1980-84 0.0 1.0 -0.4 -0.4 0.2 -1.0 -0.6 -1.6 0.0 -1.4 1970-84 0.0 0.8 -0.9 -0.9 -1.0 -1.4 -0.9 -2.3 0.0 -3.3 Portugal" - - - - - - - - - Group IV 1960s - - - - - - - - 0.0 - averagee 1970s 0.0 0.7 -1.2 -1.2 -1.6 -1.6 -1.0 -2.6 0.0 -4.2 1980s 0.0 1.0 -0.4 -0.4 0.2 -1.0 -0.6 -1.6 0.0 -1.4 1960-84 0.0 0.8 -0.9 -0.9 -1.0 -1.4 -0.9 -2.3 0.0 -3.3 Sample 1960-69 19.8 0.1 -0.0 -1.02 18.9 -1.4 -0.3 -1.7 -0.4 16.8 average' 1970-79 10.9 0.2 -0.1 -2.21 8.8 -1.7 -0.5 -2.3 -0.1 6.4 1980-83 5.8 0.1 -0.1 -1.90 3.9 -3.0 -0.5 -3.6 -0.5 -0.2 1960-83 10.0 0.1 -0.1 -1.95 8.2 -1.8 -0.5 -2.3 -0.3 6.6 indicates that data were not available. Note: See appendix 6-1 for a detailed discussion of how budget effects were allocated to each category. a. A value of 0 for state agency does not necessarily mean that the agency had no profits or losses; rather, it may indicate that any gains or losses were allocated to other categories. b. Data for C6te d'1voire and Portugal were not supplied in sufficient detail by the authors of the country studies to permit calculations of budget effects. c. Simple, unweighted group average. d. Figures are expressed as a percentage of total revenue rather than of total expenditure. e. Figures are expressed as a percentage of total public expenditure including the profits and losses of decentralized agencies. f . The study on Korea reported the subsidies to consumers and producers in grains and livestock (from product price interventions) as a combined estimate; for the purpose of this analysis, this figure was allocated half to producers and half to consumers. Input subsidies are reported separately. Source: Authors' calculations based on country studies. 110 Consequences of Price Intervention for Income Transfers and Equity Calculations of the effect of food subsidy programs on the budget illustrate the complexity of budget flows and raise several measure- ment issues. Bread and flour subsidies in Egypt provide an instruc- tive illustration. Wheat flour and bread are sold to consumers in unlimited quantities at subsidized prices. The wheat subsidy, of mod- est proportions before 1973, reached 15 percent of government expen- diture in 1974 (a year of exceptionally high world prices) and has continued to create enormous fiscal strain. What is not usually recog- nized, however, is that before 1973 the government gained consider- able revenue from its sales of imported grain,6 much of it from the sale of grain imported under food aid programs on highly concession- al terms (Dethier 1990). To compute the true cost to the government budget of subsidies on purchases of imported wheat flour, Dethier developed four alternative measures. For 1981, for example, esti- mates of the budget cost of food subsidies using the four measures ranged between 2.5 and 5.0 percent of total government expenditure. Dethier selected 3.7 percent as the "best" measure of the budget cost in 1981, for an average net value of 3.6 percent during 1980-84. Sugar policies in Egypt also illustrate the complex budget flows that need to be considered in estimating the budget effects of price inter- ventions. The government agency in charge of sugar distribution incurs a cost (an expenditure effect) from the sale of rationed sugar, but it makes money (a revenue effect) from the sale of nonrationed sugar when international prices are low and from export sales of refined sugar. The net effect of these three operations on the state sugar agency's budget fluctuates considerably and depends on world prices. The net effects were negligible in 1975-79, resulted in net outlays of 2.2 percent of government expenditure in 1980 and 3 per- cent in 1974, and generated net revenues in 1972-73. So the fiscal burden of consumer food subsidies depends on many factors. In several countries the national food authority or its equiva- lent has become a real profit center, using the revenue from duty-free monopoly imports to support some of its money-losing domestic operations. An example is the National Food Authority in the Philip- pines (described in Intal and Power 1990). In some countries, too, food is subsidized implicitly through agri- cultural export taxes. In Argentina, Pakistan (basmati rice), and Thai- land, the export orientation of the food sector allows the government to keep domestic food prices low and to gain revenue in the process- constrained only by the extent to which it is willing to reduce produc- tion incentives. In contrast, governments in food-importing countries that wish to maintain low consumer prices generally must incur a net outlay to do so. They receive no revenue from border taxes on food imports, since food imports are rarely subject to tariffs, and, in any case, the state agencies that have a monopoly on imports of major The Fiscal Impact of Direct Intervention 111 staples are exempt from trade taxes. In addition, most countries pro- tect the production of import-competing nonstaple agricultural prod- ucts through import quotas rather than revenue-yielding tariffs, so most agricultural imports provide no direct revenue. The only source of revenue from importables comes from domestic sales at higher than import parity prices, including imports through food aid. Changes in Agricultural Price Policies and Their Budget Effects Policies of selective intervention in a single sector such as agriculture are particularly susceptible to pressures (including fiscal pressures) for change from many sides. The country studies show that the objec- tives of specific policy instruments changed over time, as did their effects on the budget. Many of these shifts represented temporary changes designed to achieve a specific objective; others signaled the beginning of a new policy trend. In both cases, the dynamic relation- ship between price policies and their budget effects influenced the nature of these changes. The link between price stabilization policies and the budget was a typical source of change in policy objectives. The Philippines intro- duced an explicit variable export tax in the 1970s as a stabilization measure following devaluation of the peso. The tax soon began to generate substantial revenue, however, and the revenue motive came to dominate the stabilization objective (Intal and Power 1990). As a result, this taxation later became a regular feature and remained in place until approximately 1980. A change in the opposite direction occurred in Thailand, where the revenue contribution of the export tax on rice declined from about 12 percent in the 1960s to only 3 percent in the early 1980s as the variable export tax came to be used mainly to stabilize domestic prices (Siamwalla and Setboonsarng 1991). Changes also occurred in the pattern and size of government spending on agriculture. In Malaysia the government spent very little before 1981 on direct production subsidies, concentrating instead on agricultural infrastructure-mainly irrigation and processing facilities. After 1981, however, the government began to subsidize local paddy production and agricultural inputs (Jenkins and Lal 1990). In Chile the land reform program of 1966-73 absorbed a very high share of rising total government expenditure on agriculture, reducing public expenditures on interventions related to price policy (Hurtado, Vald6s, and Muchnik 1990). Overall most countries experienced a declining long-term trend in the net budget effect of price interventions. The net effect diminished substantially between the 1960s and early 1980s, mainly because of 112 Consequences of Price Intervention for Income Transfers and Equity the declining share of taxation as a percentage of government expen- diture. The only item in the government accounts whose weight increased over the period was credit subsidies. (Why these shifts occurred is discussed in the country studies and in volume 5 of this study.) Conclusion In most countries the net budget effect of price interventions in agri- culture was a positive gain in revenue, mainly through taxes on agri- cultural exports. For the representative group of countries (Group II), price interventions in output and input markets during 1960-84 pro- duced positive net revenues to the government equivalent to almost 8 percent of total government expenditure. The revenue effects of out- put taxation alone reached as high as 20 percent of total government expenditure in the 1960s. Over time, however, the net contribution of the price interventions to government revenue for the representative group of countries fell substantially, from 18 percent of total govern- ment expenditure in the 1960s to 5 percent in 1980-84. In some coun- tries, the revenue contribution remains important and undoubtedly presents a considerable obstacle to policy reforms intended to reduce direct agricultural taxation. In those countries, sectoral reforms that would eliminate or reduce agricultural export taxation would need to be accompanied by economywide tax and other fiscal reforms. For a few countries agricultural price interventions resulted in net outlays. This was the case in Brazil, Korea, Morocco, Pakistan, Tur- key, and Zambia and, on average, for the mild discriminators (Group III) in 1980-83 and the protectors (Group IV) in 1960-84. In Brazil the net budget effects of price interventions have varied dramatically, from a revenue gain of 32 percent during 1966-69 to a net outlay of 43 percent during 1980-83, mainly because of a large increase in credit subsidies to agriculture. The single most important source of revenue from government interventions in agricultural markets was export taxes, including the profits from the operations of state agencies involved in export activ- ities. On average during 1960-84, export taxes were equivalent to approximately 10 percent of government expenditure (20 percent in 1960-69). And in three countries-Brazil, Ghana, and Sri Lanka- export revenues amounted to 20 percent or more of total government expenditure. Credit and food subsidies had the next largest impact on the bud- get, although at less than 2 percent of government expenditure over the period, their impact was considerably lower than that of export revenues. In some countries, however, credit or food subsidies were substantial. Credit subsidies represented outlays equivalent to 20 per- The Fiscal Impact of Direct Intervention 113 cent of government expenditure in Brazil and 4 percent in Colombia, and consumer food subsidies constituted a fairly high share of gov- ernment expenditure in Brazil, Egypt, Morocco, Pakistan, and Sri Lanka (table 6-2). The price intervention-budget effect interaction differed system- atically for food exporters and food importers. Governments in food- exporting countries, such as Argentina and Thailand, were able to keep domestic food prices lower than border prices through export taxes and to gain revenue in the process. But governments in food- importing countries, such as Egypt, Morocco, Pakistan, and Sri Lanka, incurred substantial fiscal costs for food subsidies to keep consumer prices low. In most of these countries, food imports gener- ated no revenues because the state agencies with a legal monopoly on staple imports paid no trade taxes, while other agricultural import- ables were protected through quotas rather than revenue-generating tariffs. Finally, it is important to remember that all the revenue and expen- diture effects described in this chapter were the result of direct sec- toral interventions. And although the net effect of these interventions was to provide revenue to the government, a far larger impact on agriculture resulted from off-budget items not quantified here-the effects of indirect price interventions, which do not provide revenue to the government. These effects are considered again in the follow- ing chapter, which looks at the income transfers between agriculture and the rest of the economy. Appendix 6-1. Breakdown of Government Expenditure and Revenue Effects of Agricultural Price Interventions Government Expenditures: Input Policies Because the time-series data on government expenditures in the countries in our sample differed in coverage, categories, and defini- tions, we regrouped the primary data presented in the country reports for the cross-country comparison and numerical analysis pre- sented in this chapter. The following categories of expenditures were used in this classification: CREDIT SUBSIDIES. The authors of the country studies computed the annual implicit subsidy flows on agricultural credit in real terms, using a nominal reference interest rate (in the absence of subsidy).7 The computations assume that all credit is repaid at the end of the year and, therefore, make no adjustment for "soft" repayment 114 Consequences of Price Intervention for Income Transfers and Equity schemes. The selection of the reference cost of credit (i*) is clearly important to the computation of the subsidy. The authors of the coun- try studies used one of three methods to estimate it: the social rate of return to public investment; the international lending rate (plus the risk premium to the country in question); or the domestic market interest rate. The authors made their own choices, and the values used for the calculations in this chapter are those presented in the country reports. INPUT SUBSIDIES. This category includes subsidies for fertilizer, pesticides, and machinery operation. Because fertilizer and pesticides were imported in some countries, for convenience of presentation in table 6-2 input subsidies include both expenditures on input subsidies and revenues from import taxes on tradable inputs. Government Expenditures: Output Policies CONSUMER SUBSIDIES. This category includes explicit food sub- sidies; it also includes implicit food subsidies when they were calcu- lated by the authors of the country studies. Implicit subsidies on food consumption include the deficits of state-owned marketing boards that resulted from selling at a financial loss (to wholesalers and retailers) food that it purchased from domestic producers and foreign suppliers. When the authors did not include these implicit subsidies as consumer subsidies, the subsidies were assigned to the "state agency" category. (The losses of state agencies were defined as pur- chases plus operational expenses minus sales. No adjustment was made for possible inefficiencies in the management of the parasta- tals.) For food-importing countries that received substantial amounts of food aid, such as Egypt, adjustments were made in the border price. For example, Dethier (1990) presented four estimates of border prices for wheat and flour in Egypt that adjusted the border price of commercial imports for the grant element of food aid for concessional imports. Thus, for countries that received food aid imports, the food subsidy category includes the net revenues to the government from food aid imports. PRODUCTION SUBSIDIES. This category includes both explicit sub- sidies and implicit subsidies to government agencies. For countries for which the study authors did not present disaggregated figures for production subsidies and parastatal accounts, production subsidies are reported under "state agency." For example, only a general bal- ance that also included input subsidies was available for the parasta- tal accounts for Turkey and Zambia; this was reported under "state agency." The Fiscal Impact of Direct Intervention 115 LOSSES OF STATE AGENCIES. The way items were assigned to this category is discussed in the government revenue section that follow under "state agencies." Government Revenues Revenues were grouped into three categories: export taxes, import taxes, and profits of state agencies. EXPORT TAXES. This category includes both explicit export taxes and implicit export taxes that were classified as such by the authors of the country study. Implicit export taxes were calculated as the difference between the export prices received by state-owned companies and the prices paid to farmers (at the wholesale level). Otherwise, implicit export taxes were classified under "state agency." In some cases- Colombia is one-export taxes were calculated as the net of total taxes and subsidies. IMPORT TAXES. In most developing countries, food imports are exempt from border taxes, generally because of a state monopoly on food imports. For example, despite a statutory tariff on corn imports in the Philippines of about 86 percent during the period, the National Food Authority imported corn duty free. The implicit gain in such cases was classified under "state agency." Similarly, in Ghana, the schedule of custom and excise tariffs lists a 50 percent ad valorem duty on maize imports and a specific tax on rice imports. These taxes do not apply to food aid, however, and cereal imports are exempt from sales taxes. So for Ghana, there is a loss in revenue from the exemptions but a gain to the state agencies from their sales in the domestic market. For each country, these budgetary flows were allo- cated to "import taxes" or "state agency" (table 6-2) according to the same criteria used to measure explicit and implicit export taxes. STATE AGENCIES. Many agricultural price interventions result from the activities of parastatals and are, strictly speaking, off-budget items if budget items are defined as explicit tax revenues and government expenditures. But the gains and losses of state agencies, after wend- ing their way through various ministries and decentralized agencies, finally surface in the government accounts, where they often repre- sent substantial revenues or expenditures. For the calculations in this chapter, the category "state agency" includes the profits and losses of state agencies when they were not allocated to any other category. So values of zero for "state agency" (table 6-2) could mean that the revenues from trade operations by the state agency (such as revenues from implicit export taxes) were classified as export taxes. 116 Consequences of Price Intervention for Income Transfers and Equity Measuring the operatiornal losses and gains of state-owned com- panies involved in the domestic procurement, processing, and trade of food products was particularly complex. For example, when the operational losses seemed to be attributable largely to higher than "normal" operating margins, these were in most cases monopoly activities for which there was no domestic reference marketing mar- gin. Moreover, the data on government operations were frequently aggregated over several different functions. Where data on the fiscal receipts of marketing boards were unavailable, we estimated the bud- get costs or revenues indirectly by subtracting the parastatal costs plus farmers' gross revenue from the f.o.b. value of the marketed product (the case in Ghana). NET EFFECTS. This category measures the net effect on the govern- ment budget of the price policies affecting inputs and outputs. Other Measurement Issues All the flows in table 6-2 are measured from the government's side, so that positive figures mean government revenues and negative figures mean government outlays. Also, because most input subsidies and other outlays are not crop-specific, the budget effects in table 6-2 include budget flows for all agricultural commodities, not just the selected products. Finally, so that all budgetary flows are expressed in a comparable unit across countries, they were calculated as a percent- age of total public (central government) sector expenditure. Excep- tions were Argentina, where they are expressed as a percentage of government revenue, and Chile, where total public expenditure included profits and losses of decentralized agencies. State or provin- cial government accounts were excluded because information on these political jurisdictions was not available for many of the countries. The Effects of Intervention on Income Transfers to and from Agriculture In most industrial countries the main objective of agricultural price policies is to maintain farm income and employment in the face of declining real world prices for cereals. The result has been significant net income transfers to agriculture. In most developing countries, however, the primary objectives have been food self-sufficiency, domestic price stability, low food prices for urban consumers, and government revenue. Although the relative weights given these often conflicting objectives vary across countries and through time, on average the net effect of total interventions has been an enormous income transfer out of agriculture, averaging 46 percent of agri- cultural gross domestic product (GDP) a year over 1960-84. For the four country groups in our analysis, transfers out of agriculture from total price and nonprice interventions ranged from 2 percent for the protectors group (Group IV) to 140 percent for the extreme discrimi- nators (Group I). The average for the representative group (Group II) was 37 percent. Such enormous transfers out of agriculture must have had a severely depressive effect on private investment in agri- culture and so on agricultural growth over the period (see the discus- sion of agricultural incentives and output growth in chapter 4). In this chapter we estimate the size and direction of the income transfers between agriculture and the rest of the economy and then look at how these transfers were distributed among agriculture, gov- ernment, and the rest of the economy. An analysis of these income transfers and of who gained and who lost from them helps to reveal some of the motivations for price interventions and to identify the forces arrayed against reform. Agriculture was clearly the loser, while the big winners were government (net revenue gain), urban con- sumers (lower food prices), and industry (cheap raw materials and other inputs). In this chapter we also examine the hypothesis that governments have compensated for the adverse income effects of price policies on agriculture by increasing agricultural investment. 117 118 Consequences of Price Intervention for Income Transfers and Equity That hypothesis is rejected in most cases, which. means that agri- cultural price policy has been the primary influence on farm income. On the Concept of Income Transfers between Agriculture and the Rest of the Economy Transfers between agriculture and the rest of the economy included those related to price policies and those related to other policies affecting agriculture. Price-related transfers are defined as the changes in real income of the agricultural sector resulting from direct and indirect price interven- tions affecting output and input prices and the prices of consumer goods purchased by farm households. Specifically, these transfers are measured as the change in value added resulting from both direct and indirect price interventions, measured at the actual level of produc- tion and adjusted for the change in the rural consumer price index. The change in the price index accounts for the effect of interventions on the cost of the rural consumption basket, which comprises the farm household's output and its purchases of other agricultural and nonagricultural tradable products. Thus, price-related transfers ulti- mately measure the overall net movement in real purchasing power of agriculture that results from price interventions. Government expenditures and taxes that are not directly related to price policy, such as public investment in agriculture and government revenues from sector-specific taxes, also result in income transfers between agriculture and the rest of the economy. These non-price related transfers were defined to include public investments that can be considered public goods, such as irrigation and roads, and other investments, as in research and extension, that help to shift the agri- cultural production frontier and so affect farm income. Marketing- related investments by state agencies, such as in the storage infra- structure, were excluded on two premises: first, such expenditures influence the marketing margin and so are reflected in the prices received by farmers or paid by consumers (transfers for such outlays appear under price interventions); second, they do not clearly consti- tute a transfer to agriculture (in fact in many countries those activities are undertaken by private traders). Price- and non-price-related transfers (hereafter referred to as price transfers and nonprice transfers respectively) together constitute net transfers. Nonprice transfers are considered to be unaffected by indi- rect price interventions, and so the same value of nonprice transfers is used in computing net transfers from direct and total (direct plus indirect) interventions. (The methodology for measuring price policy-related transfers, including adjustments for changes in the rural consumer price index, is presented in appendix 7-1.)1 In inter- The Effects of Intervention on Income Transfers to and from Agriculture 119 preting the results on transfers, one ought to note that, in terms of farmer welfare, one dollar lost through price transfers in year t is unlikely to be fully compensated by one dollar invested in public investment in year t. The benefits of such public investment do not generally reach producers for several years, and even then the bene- fits are usually geographically concentrated and crop-specific and so do not necessarily compensate farmers outside the affected regions or crops. The annual flows of price policy-related transfers were computed at the actual level of production for the selected products, so these measures represent only the very short-term impact of price policies on farm income; they ignore the potential effects on the level of production. (As explained in chapter 4, no reliable measures of the long-run output effects from changes in price interventions are avail- able from the country studies.) Because the response of farm output to a given change in price policy interventions grows over time, our measure of short-term transfers could differ significantly from the longer-term income effect of price interventions. Ideally, both the short- and longer-term transfers should be computed. The short-term transfer represents the impact at the time of a policy change and so is likely to be influential to the policy debate. The longer-term transfer measures the more lasting effects on the sector's income, but these effects are hard to measure empirically, a factor that may reduce their influence on the policy debate. Empirical Evidence from Previous Studies Empirical studies of transfers affecting agriculture in developing countries fall into two broad groups. One group is represented by Ishikawa's (1967) analysis of intersectoral flows in Taiwan, China, India, and Japan.2 Ishikawa's study was concerned with the view, dominant in the 1950s and 1960s, that emphasized the extraction of the "surplus" from agriculture to finance industrialization, presuma- bly at a low social cost in terms of agricultural growth. More recent work by Ishikawa (1987) documents the great diversity in the patterns of intersectoral resource flows in these Asian countries and concludes that heavy net outflows of resources from agriculture took place in Taiwan and India, whereas net inflows occurred in China and Japan.3 But there is no simple or obvious relationship between the patterns of resource flows and the rate of growth in agricultural production in the four cases, as Myint (1987) argues in his assessment of Ishikawa's findings. Interpreting these outcomes would require an examination of how productively the intersectoral resource flows were used and how policies in various countries affected the efficiency of resource use in agriculture. 120 Consequences of Price Intervention for Income Transfers and Equity Studies in the second group are based on a partial equilibrium analysis in the Marshallian economic surplus framework. Originally applied to U.S. agriculture by A. C. Harberger in the 1950s,4 this approach was subsequently expanded by Wallace (1962) and applied by Dardis (1967) and Josling (1969) to the United Kingdom. Studies applying this approach to developing countries began to emerge in the late 1970s, and this literature is now vast (see, for example, Bale and Lutz 1979; Lutz and Saadat 1988; Tolley, Thomas, and Wong 1982; and Timmer 1986). These studies found that agricultural price policies in developing countries have taxed agriculture and favored consumers over producers. They also found that these policies' major objectives were to raise revenues for the government and to keep food prices down for urban consumers-and that the second objective frequently defeated the first (Lutz and Saadat 1988). This literature on transfers has provided ample documentation of the short-run income transfers from direct price interventions, but it has ignored the effect of indirect interventions on transfers and has contributed little to the analysis of the effects of price policy on agri- cultural efficiency and long-term growth. Thus, these studies found that the static dead-weight efficiency loss is small relative to the income transfers and budget effects of price interventions, reinforcing the perception that the short-term transfers and budget effects may have been the dominant political-economy forces behind farm price policies. The analysis presented in this chapter, by incorporating the effects of indirect price interventions, remedies one shortcoming of these studies, but the weakness with respect to longer-term effects remains because of the difficulties of measuring long-term output effects (see chapter 4). Measuring Income Transfers PRICE POLICY-RELATED TRANSFERS. Income transfers resulting from price interventions in agriculture include transfers through interven- tions in output and input markets. Transfers through input markets cover all agricultural inputs, not just those purchased for the produc- tion of the products analyzed in this project, since most input price interventions are not earmarked for specific crops (for the list of prod- ucts analyzed, see table 1-1). We provide two measures of transfers through output market inter- ventions. The first measures the income change from direct and total price interventions on the output of the selected products only (first column in tables 7-1 and 7-2).5 For consistency with the full-sector coverage of transfers through input markets and nonprice transfers, the second measure of transfers through output policies covers the rest of the sector.6 Since we did not have measures of direct price Table 7-1. Net Income Transfers to (+) and from (-) Agriculture as a Result of Direct Price and Nonprice Interventions, 1960-86 (percentage of agricultural GDP) Direct price transfers Output of other Sum of direct price and nonprice transfers agricultural Sum of direct price transfers Avg. of Output of products Assumption assumption selected All (assumption Assumption 2 (1) + (2) Nonprice Assumption Assumption I & 2 (half products- inputSb 2)c 1 (1) + (2) + (3) transferSd 1(4) + (6) 2 (5) + (6) of (7) + (8)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) C6te d'Ivoire 1960-69 -10 0 -43 -10 -53 6 -4 -47 -25.5 1970-79 -24 2 -93 -22 -115 18 -4 -97 -50.5 1980-82 -10 1 -55 -9 -64 20 11 -44 -16.5 1960-82 -16 1 -67 -15 -82 13 -2 -69 -35.5 Ghana 1962-69 -13 0 -76 -13 -89 3 -10 -86 -48 1970-76 -14 3 -127 -11 -138 3 -8 -135 -71.5 1976-84 1962-76 -14 1 -100 -13 -113 3 -10 -110 -60 Zambia 1960-70 1971-79 -5 4 -35 -1 -36 5 4 -31 -13.5 1980-84 -5 4 -27 -1 -28 4 3 -24 -10.5 1971-84 -5 4 -32 -1 -33 5 4 -28 -12 Group I 1960s -11.5 0.0 -59.5 -11.5 -71.0 4.5 -7.0 -66.5 -36.8 averagee 1970s -14.3 3.0 -85.0 -11.3 -96.3 8.7 -2.7 -87.7 -45.2 1980s -7.5 2.5 -41.0 -5.0 -46.0 12.0 7.0 -34.0 -13.5 1960-84 -11.7 2.0 -66.3 -9.7 -76.0 7.0 -2.7 -69.0 -35.8 Argentina 1960-69 -10 1 -14 -9 -23 0 -9 -23 -16 (Table continues on the following page.) Table 7-1 (continued) (percentage of agricultural GDP) Direct price transfers Output of other Sum of direct price and nonprice transfers agricultural Sum of direct price transfers Avg- of Output of products Assumption assumption selected All (assumption Assumption 2 (1) + (2) Nonprice Assumption Assumption 1 & 2 (half productsa inputsb 2)c 1(1) + (2) + (3) transfersd 1(4) + (6) 2 (5) + (6) of(7) + (8)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) 1970-79 -13 6 -20 -7 -27 0 -7 -27 -17 1980-83 -10 3 -13 -7 -20 0 -7 -20 -13.5 1960-83 -12 4 -16 -8 -24 0 -8 -24 -16 Colombia 1960-69 -1 2 -4 1 -3 3 4 0 2 1970-79 -8 3 -15 -5 -20 5 0 -15 -7.5 1980-83 0 4 -1 4 3 4 8 7 7.5 1960-83 -4 3 -8 -1 -9 4 3 -5 -1 Dominican 1960-69 - Republic 1970-79 -17 -1 -13 -18 -31 13 -5 -18 -11.5 1980-84 -5 -1 -3 -6 -9 10 4 1 2.5 1960-84 -13 -1 -10 -14 -24 12 -2 -12 -7 Egypt 1966-69 -19 -3 -3 -22 -25 10 -12 -15 -13.5 1970-79 -11 -4 -2 -15 -17 6 -9 -11 -10 1980-84 4 -8 -1 -4 -5 7 3 2 2.5 1966-84 -9 -5 -2 -14 -16 7 -7 -9 -8 Morocco 1963-69 -5 0 -18 -5 -23 9 4 -14 -5 1970-79 -4 0 -15 -4 -19 8 4 -11 -3.5 1980-84 2 0 4 2 6 8 10 14 12 1963-84 -3 0 -12 -3 -15 8 5 -7 -1 Pakistan 1960-69 - - - - - - - - 1970-79 -16 0 -9 -16 -25 14 -2 -11 -6.5 1980-86 -21 3 -14 -18 -32 14 -4 -18 -11 1960-86 -18 1 -11 -17 -28 14 --3 -14 -8.5 Philippines 1960-69 -1 -1 0 -2 -2 2 0 0 0 1970-79 -10 -1 -7 -11 -18 6 -5 -12 -8.5 1980-86 -8 0 -4 -8 -12 4 -4 -8 -6 1960-86 -6 -1 -4 -7 -11 4 -3 -7 -5 Sri Lanka 1960-69 -5 2 -7 -3 -10 10 7 0 3.5 1970-79 -10 3 -14 -7 -21 15 8 -6 1 1980-85 -15 0 -15 -15 -30 42 27 12 19.5 1960-85 -9 3 -12 -7 -19 19 12 0 6 Thailand 1962-69 -14 0 -25 -14 -39 7 -7 -32 -19.5 1970-79 -15 0 -26 -15 -41 8 -7 -33 -20 1980-83 -7 1 -10 -6 -16 10 4 -6 -1 1962-83 -13 0 -23 -13 -36 8 -5 -28 -16.5 Turkey 1961-69 8 0 6 8 14 4 12 18 15 1970-79 9 1 10 10 20 4 14 24 19 1980-83 -5 9 -7 4 -3 5 9 2 5.5 1961-83 6 2 6 8 14 4 12 18 15 Group II 1960s -5.9 0.1 -8.1 -5.8 -13.9 5.6 -0.1 -8.3 -4.2 averagee 1970s -9.5 0.7 -11.1 -8.8 -19.9 7.9 -0.9 -12.0 -6.5 1980s -6.5 1.1 -6.4 -5.4 -11.8 10.4 5.0 -1.4 1.8 (Table continues on the following page.) Table 7-1 (continued) (percentage of agricultural GDP) Direct price transfers Output Of other Sum of direct price and nonprice transfers agricultural Sum of direct price transfers Avg. of Output of products Assumption assumption selected All (assumption Assumption 2 (1) + (2) Nonpnce Assumption Assumption I & 2 (half productsa inputsb 2) 1(1) + (2) + (3) transferSd 1(4) + (6) 2 (5) + (6) of (7) + (8)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) 1960-86 -8.1 0.5 -9.2 -7.6 -16.8 8.0 0.4 -8.8 -4.2 Brazil 1966-69 - - - - - - - - 1970-79 4 14 -8 18 10 13 31 23 27 1980-83 -1 33 2 32 34 11 43 45 44 1960-83 2 20 -5 22 17 12 34 29 31.5 Chile 1960-69 0 2 -6 2 -4 4 6 0 3 1970-79 1 3 -2 4 2 4 8 6 7 1980-82 1 1 3 2 5 3 5 8 6.5 1960-82 0 2 -3 2 -1 4 6 3 4.5 Malaysia 1960-69 -10 0 -3 -10 -13 6 -4 -7 -5.5 1970-79 -14 1 -7 -13 -20 10 -3 -10 -6.5 1980-83 -9 2 -4 -7 -22 14 7 3 5 1960-83 -12 1 -5 -11 -16 9 -2 -7 -4.5 Group III 1960s -5.0 1.0 -4.5 -4.0 -8.5 5.0 1.0 -3.5 -1.3 averagee 1970s -3.0 6.0 -5.7 3.0 -2.7 9.0 12.0 6.3 9.2 1980s -3.0 12.0 0.3 9.0 9.3 9.3 18.3 18.7 18.5 1960-83 -3.3 7.7 -4.3 4.3 0.0 8.3 12.7 8.3 10.5 Korea, 1962-69 11 1 6 12 18 6 18 24 21 Rep. of 1970-79 18 1 18 19 37 7 26 44 35 1980-84 27 1 21 28 49 9 37 58 47.5 1962-84 17 1 15 18 33 7 25 40 32.5 Portugal 1960-69 8 -4 13 4 17 1 5 18 11.5 1970-79 9 -2 5 7 12 1 8 13 10.5 1980-82 10 3 6 13 19 2 15 21 18 1960-82 8 -2 9 6 15 1 7 16 11.5 Group IV 1960s 9.5 -1.5 9.5 8.0 17.5 3.5 11.5 21.0 16.3 averagee 1970s 13.5 -0.5 11.5 13.0 24.5 4.0 17.0 28.5 22.8 1980s 18.5 2.0 13.5 20.5 34.0 5.5 26.0 39.5 32.8 1960-84 12.5 -0.5 12.0 12.0 24.0 4.0 16.0 28.0 22.0 U Sample 1960-69 -3 0 -10 -3 -13 4 1 -9 -4 averagee 1970-79 -7 2 -20 -5 -25 8 3 -17 -7 1980-83 -3 3 -7 0 -6 9 10 3 6 1960-83 -6 2 -16 -4 -19 7 4 -12 -4 - indicates that data were not available. a. The change in the gross value of output of selected agricultural products as a result of direct price interventions (in relation to the counterfactual simulation without intervention). b. Transfers resulting from the price interventions on inputs, including credit subsidies (and replanting subsidies for rubber), on all agricultural products. c. The additional effect of price-related transfers on the gross value of output for the rest of agriculture. Under assumption 1, the rate of nominal protection for the rest of agriculture is assumed to be 0 (not shown); under assumption 2, it is assumed to equal nominal protection (or taxation) for the selected products. d. Nonprice transfers include public investment in irrigation, agricultural research and extension, and land improvements. e. In computing the average for the total sample, the unit corresponds to the value of each variable in a specific country in a specific year. This amounts to giving more weight to countries for which more observations were available. The results following this procedure exhibit a slight numerical difference with the alternative procedure of a simple, unweighted group average from the averages for each country, particularly during the 1960s and to a lesser extent during the 1980s, years in which there were more missing values. Source: Authors' calculations based on data from country studies. Table 7-2. Net Income Transfers to (+) and from (-) Agriculture as a Result of Total (Direct and Indirect) Price and Nonprice Interventions, 1960-84 (percentage of agricultural GDP) Total price transfers Sum of total price and nonprice transfers Output of other agricultural Sum of total price transfers Avg. of Output of products Assumption Assumption assumption selected All Assumption Assumption 1 (1) + (2) 2 (1) + (2) Nonprice Assumption Assumption I & 2 (half products- inputsb Ic 2c + (3) + (4) transfersd ](5) + (7) 2 (6) + (7) of (8) + (9)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) C6te d'Ivoire 1960-69 -13 1 -10 -55 -22 -67 6 -16 -61 -38.5 1970-79 -32 3 -42 -126 -71 -155 18 -53 -137 -95 1980-82 -15 2 -27 -78 -40 -91 20 -20 -71 -45.5 1960-82 -21 2 -26 -89 -45 -108 13 -95 -95 -63.5 Ghana 1962-69 -28 1 -65 -154 -92 -181 3 -89 -178 -133.5 1970-76 -25 4 -60 -218 -81 -239 3 -78 -236 -157 1976-84 1962-76 -26 2 -63 -184 -87 -208 3 -84 -205 -144.5 Zambia 1960-70 1971-79 -19 9 -80 -144 -90 -154 .5 -85 -149 -117 1980-84 -71 36 -304 -411 -339 -446 4 -335 -442 -388.5 1971-84 -37 19 -160 -239 -178 -257 5 -173 -252 -212.5 Group I 1960s -20.5 1.0 -37.5 -104.5 -57.0 -124.0 4.5 -52.5 -119.5 -86.0 averagee 1970s -25.3 5.3 -60.9 -162.7 -80.7 -182.7 8.7 -72.0 -174.0 -123.0 1980s -43.0 19.0 -165.5 -244.5 -189.5 -268.5 12.0 -177.5 -256.5 -217.0 1960-84 -28.0 7.7 -83.0 -170.7 -103.3 -191.0 7.0 -96.3 -184.0 -140.2 Argentina 1960-69 -37 -16 -10 -38 -63 -91 0 -63 -91 -77 1970-79 -41 -19 -7 -38 -67 -98 0 -67 -98 -82.5 1980-83 -51 -8 -32 -62 -91 -121 0 -91 -121 -106 1960-83 -41 -16 -12 -42 -69 -99 0 -69 -99 -84 Colombia 1960-69 -14 2 -30 -30 -42 -42 3 -39 -39 -39 1970-79 -20 3 -20 -38 -37 -55 5 -32 -50 -41 1980-83 -15 4 -27 -25 -38 -36 4 -34 -32 -33 1960-83 -17 3 -25 -32 -39 -46 4 -35 -42 -38.5 Dominican 1960-69 Republic 1970-79 -37 0 -11 -28 -48 -65 13 -35 -52 -43.5 1980-84 -22 1 -9 -12 -30 -33 10 -20 -23 -21.5 1960-84 -32 0 -10 -23 -42 -55 12 -30 -43 -36.5 Egypt 1966-69 -26 0 -5 -73 -31 -99 10 -21 -89 -55 1970-79 -14 -1 17 -55 2 -70 6 8 -64 -28 1980-84 3 -6 36 0 33 -3 7 40 4 22 1966-84 -12 -2 18 -45 4 -59 7 11 -52 -20.5 Morocco 1963-69 -7 0 0 -24 -7 -31 9 2 -22 -10 1970-79 -6 1 2 -24 -3 -29 8 5 -21 -8 1980-84 0 1 10 -5 11 -4 8 19 4 11.5 1963-84 -5 1 3 -19 -1 -23 8 7 -15 -4 Pakistan 1960-69 1970-79 -42 7 -7 -26 -42 -61 14 -28 -47 -37.5 1980-86 -44 9 -5 -28 -40 -63 14 -26 -49 -37.5 1960-86 -43 7 -6 -27 -42 -63 14 -28 -49 -38.5 Philippines 1960-69 -11 -1 -11 -7 -23 -19 2 -21 -17 -19 1970-79 -22 0 -9 -17 -31 -39 6 -25 -33 -29 1980-86 -19 1 -7 -11 -25 -29 4 -21 -21 -23 1960-86 -17 0 -9 -12 -26 -29 4 -22 -22 -23.5 (Table continues on the following page.) Table 7-2 (continued) (percentage of agricultural GDP) Total price transfers Sum of total price and nonprice transfers Output of other agricultural Sum of total price transfers Avg. of Output of products Assumption Assumption assumption selected All Assumption Assumption 1(1) + (2) 2 (1) + (2) Nonprice Assumption Assumption 1 & 2 (half products- inputSb e 2c + (3) + (4) transferSd 1 (5) + (7) 2 (6) + (7) of (8) + (9)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Sri Lanka 1960-69 -41 3 -22 -31 -60 -69 10 -50 -59 -54.5 1970-79 -58 5 -39 -60 -92 -113 15 -77 -98 -87.5 1980-85 -60 0 -31 -54 -91 -114 42 -49 -72 -60.5 1960-85 -52 3 -30 -48 -79 -97 19 -60 -78 -69 Thailand 1962-69 -24 0 -2 -33 -26 -57 7 -19 -50 -34.5 1970-79 -28 1 -5 -37 -32 -64 8 -24 -56 -40 1980-83 -24 2 -8 -21 -30 -43 10 -20 -33 -26.5 1962-83 -26 1 -5 -32 -30 -57 8 -22 -49 -35.5 Turkey 1961-69 -9 0 -18 -10 -27 -19 4 -23 -15 -19 1970-79 -6 3 -18 -10 -21 -13 4 -17 -9 -13 1980-83 -22 11 -12 -29 -23 -40 5 -18 -35 -26.5 1961-83 -10 3 -17 -13 -24 -20 4 -20 -16 -18 Group II 1960s -21.1 -1.5 -12.3 -30.8 -34.9 -53.4 5.6 -29.3 -47.8 -38.5 averagee 1970s -27.4 0.0 -9.7 -33.3 -37.1 -60.7 7.9 -29.2 -52.8 -41.0 1980s -25.4 1.5 -8.5 -24.7 -32.4 -48.6 10.4 -22.0 -38.2 -30.1 1960-83 -25.5 0.0 -9.3 -29.3 -34.8 -54.8 8.0 -26.8 -46.8 -36.8 Brazil 1966-69 - 1970-79 -2 16 -11 0 3 14 13 16 27 21.5 1980-83 07 35 -10 -14 18 14 11 29 25 27 1960-83 -4 21 -11 -4 6 13 12 18 25 21.5 Chile 1960-69 -12 4 -24 -30 -32 -38 4 -28 -34 -31 1970-79 -34 15 -72 -79 -91 -98 4 -87 -94 -90.5 1980-82 -5 2 -17 -12 -20 -15 3 -17 -12 -14.5 1960-82 -21 8 -44 -49 -57 -62 4 -53 -58 -55.5 Malaysia 1960-69 -13 1 -1 -4 -13 -16 6 -7 -10 -8.5 1970-79 -16 1 -2 -8 -17 -23 10 -7 -13 -10 1980-83 -13 3 -5 07 -15 -17 14 -1 -3 -2 1960-83 -14 1 -2 -6 -15 -19 9 -6 -10 -8 Group III 1960s -12.5 2.5 -12.5 -17.0 -22.5 -27.1 5.0 -17.5 -22.0 -19.8 averagee 1970s -17.3 10.7 -28.3 -29.0 -35.0 -35.7 9.0 -26.0 -26.7 -26.3 1980s -8.3 13.3 -10.7 -11.0 -5.7 -6.0 9.3 3.7 3.3 3.5 1960-83 -13.0 10.0 -19.0 -19.7 -22.0 -22.7 8.3 -13.7 -14.3 -14.0 Korea, 1962-69 -43 3 -37 -25 -77 -65 6 -71 -59 -65 Rep. of 1970-79 8 2 -23 4 -13 14 7 -6 21 7.5 1980-84 21 2 -11 17 12 40 9 21 49 35 1962-84 -7 2 -26 -3 -31 -8 7 -24 -1 -12.5 Portugal 1960-69 11 -9 7 19 9 21 1 10 22 16 1970-79 6 1 -7 -2 0 5 1 1 6 3.5 1980-82 -1 16 -20 -13 -5 2 2 -3 4 0.5 1960-82 7 -1 -3 6 3 12 1 4 13 8.5 Group IV 1960s -16.0 -3.0 -15.0 -3.0 -34.0 -22.0 3.5 -30.5 -18.5 -24.5 averagee 1970s 7.0 1.5 -15.0 1.0 -6.5 9.5 4.0 -2.5 13.5 5.5 1980s 10.0 9.0 -15.5 2.0 3.5 21.0 5.5 9.0 26.5 17.8 1960-84 0.0 0.5 -14.5 1.5 -14.0 2.0 4.0 -10.0 6.0 -2.0 (Table continues on the following page.) Table 7-2 (continued) (percentage of agricultural GDP) Total price transfers Sum of direct price and nonprice transfers Output of other agricultural Sum of total price transfers Avg. of Output of products Assumption Assumption assumption selected All Assumption Assumption 1(1) + (2) 2 (1) + (2) Nonprice Assumption Assumption I & 2 (half productsa inputsb Ic 2c + (3) + (4) transfersd 1(5) + (7) 2 (6) + (7) of (8) + (9)) Country Period (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Sample 1960-69 -15 -1 -13 -28 -28 -43 4 -24 -39 -32 averagee 1970-79 -22 3 -22 -50 -41 -69 8 -33 -61 -47 1980-83 -19 6 -27 -43 -40 -56 9 -30 -46 -38 1960-83 -21 3 -24 -48 -42 -66 7 -34 -58 -46 - indicates that data were not available. a. The change in the gross value of output of selected agricultural products as a result of direct price interventions (in relation to the counterfactual simulation without intervention). b. Transfers resulting from the price interventions on inputs, including credit subsidies (and replanting subsidies for rubber), on all agricultural products. c. The additional effect of price-related transfers on the gross value of output for the rest of agriculture. Under assumption 1, the rate of nominal protection for the rest of agriculture is assumed to be 0 (not shown); under assumption 2, it is assumed to equal nominal protection (or taxation) for the selected products. d. Nonprice transfers include public investment in irrigation, agricultural research and extension, and land improvements. e. See note in table 7-1. Same procedure was followed to compute sample average. Source: Authors' calculations based on data provided in country studies. The Effects of Intervention on Income Transfers to and from Agriculture 131 interventions for the products that were not analyzed, we used two alternative assumptions for estimating them. Under assumption 1, there are no direct price interventions in the rest of agriculture (the nominal protection rate for other products is zero); under assumption 2, the rate of direct protection (taxation) for the rest of the sector is equal to the average for the selected products.7 The true value proba- bly falls somewhere between these two extremes, since most coun- tries generally intervene only in certain products, mostly tradables (which are well represented by the-commodities covered in the proj- ect). Thus we also provide an estimate based on the average of the two assumptions. TRANSFERS NOT RELATED TO PRICE POLICY. This category includes public investment in irrigation, research and extension, and land improvement (when reported by the authors of country studies). Investment in rural roads was not included, although conceptually it belongs in this category, because our information was incomplete. The administrative costs of the ministries of agriculture or other pub- lic offices were also excluded. Size and Direction of Income Transfers The average net income transfer out of agriculture as a result of direct price interventions was 4 percent of agricultural GDP during 1960-84 for the selected products in our sample countries (under assumption 1): direct price interventions on outputs reduced agricultural GDP by about 6 percent, and transfers into agriculture through input sub- sidies boosted agricultural income by approximately 2 percent (table 7-1).8 This calculation understates the true transfer out of agriculture, however, because it includes output transfers only on the selected products, whereas input subsidies cover all of agriculture. Expanding the output coverage to the rest of agriculture raises to 19 percent the average net transfer out of agriculture as a result of direct price inter- ventions, under assumption 2. Adding the transfers into agriculture from interventions not related to price policy (7 percent) results in a total net income transfer out of agriculture as a result of direct inter- ventions of about 4 percent (the average of the results under the two assumptions). Finally, taking into account the income effects of indi- rect price interventions raises the net transfers out of agriculture to a staggering 46 percent of agricultural GDP (table 7-2). DIRECT PRICE AND NON-PRICE-RELATED TRANSFERS. Over the period 1960-84, direct price interventions resulted in income transfers to agricultures in only five countries-Brazil, Chile, Korea, Portugal, and Turkey (table 7-3). For the other thirteen countries, direct price 132 Consequences of Price Intervention for Income Transfers and Equity interventions resulted in income transfers out of agriculture ranging from 5 percent (Colombia) to 63 percent (Ghana) of agricultural GDP. For the representative group (Group I), the reduction in the gross value of production from direct price interventions on the output of selected products ranged from 5.9 to 9.5 percent of agricultural GDP in most years (table 7-1). Transfers to agriculture through subsidies on inputs were quite small, ranging from 0.13 to 1.1 percent, for a net value of direct price policy-related transfers out of agriculture of 7.6 percent under assumption 1 and 16.8 percent under assumption 2. Adding the effects of direct price interventions on output for the rest Table 7-3. Pattern of Income Transfers to (+) and from (-) Agriculture as a Result of Direct Price Interventions, 1960-84 Average (percentage of agricultural GDP) Transfers Price Not price Net of policy policy price and Pattern related related nonprice Countries with negative price-related transfers from agriculture Ghana -63.0 3.0 -60.0 C6te d'Ivoire -48.5 13.0 -35.5 Thailand -24.5 8.0 -16.5 Pakistan -22.5 14.0 -8.5 Dominican Rep. -19.0 12.0 -7.0 Zambia -17.5 5.0 -12.0 Argentina -16.0 0.0 -16.0 Egypt -15.0 7.0 -8.0 Malaysia -13.5 9.0 -4.5 Sri Lanka -13.0 19.0 6.0 Morocco -9.0 8.0 -1.0 Philippines -9.0 4.0 -5.0 Colombia -5.0 4.0 -1.0 Averagea -21.2 8.2 -13.0 Countries with positive price-related transfers from agriculture Korea, Rep. of 25.5 7.0 32.5 Brazil 19.5 12.0 31.5 Turkey 11.0 4.0 15.0 Portugal 10.5 1.0 11.5 Chile 0.5 4.0 4.5 Averagea 13.5 5.6 19.2 Note: Countries are ranked in descending order according to price policy-related transfers. Price policy-related transfers represent the sum of transfers resulting from output and input price interventions, using the average of assumptions 1 and 2 on protection for the rest of agriculture (see text and table 7-1). a. Simple, unweighted group average. The Effects of Interoention on Income Transfers to and from Agriculture 133 of agriculture-the impact of which varies from country to country because of differences in the share of agricultural GDP covered by the selected products-increases the rate of transfer out of agriculture. Nonprice-related transfers into agriculture for the twenty-five-year period averaged between 7 and 8.3 percent of agricultural GDP for the extreme discriminators, the representative group, and the mild dis- criminators, and 4 percent for the protectors group. In Brazil, C8te d'Ivoire, the Dominican Republic, Pakistan, and Sri Lanka, transfers not related to price policy were considerably higher than the cross- country group average of 7 percent; they ranged from 12 to 19 percent. Adjusting for income transfers to agriculture from non-price policy interventions reduces the net value of direct transfers out of agricul- ture to 4.2 percent of agricultural GDP for the representative group during 1960-84 (average of assumptions 1 and 2). Although a net annual transfer out of agriculture of 4.2 percent might be impercept- ible year by year, its cumulative effect on agricultural income and investment is likely to have been profoundly harmful to the sector in the longer run. For the three sub-Saharan African countries (the extreme discriminators), the net value of direct transfers out of agri- culture over the period was 35.8 percent-an amount so extraordi- narily high that its cumulative effect over a twenty-five-year period must have induced a major reallocation of labor and savings out of agriculture. By contrast, the equivalent figure for the protectors group (Group IV) was a net transfer into agriculture of 22 percent. Although average direct price and non-price-related income trans- fers affecting agriculture were negative over the period, the long-run trend has been toward declining transfers out of agriculture.9 For the representative group, negative income transfers resulting from direct price interventions on output of the selected products changed little over time, fluctuating between 5.9 and 9.5 percent of agricultural GDP. Positive transfers through input subsidies and nonprice interventions increased from less than 1 percent for inputs and 5.6 percent for nonprice interventions in 1960-69 to 1.1 percent for inputs and 10.4 percent for nonprice interventions in 1980-84. Thus, whereas trans- fers resulting from output policies remained fairly constant over time, increases in public investment and input subsidies led to a major reversal in the direction of the net transfers from direct price and nonprice related interventions, from a net outflow from agriculture in the 1960s to a net inflow in the 1980s. Reversal of income flows occurred in all groups except the extreme discriminators, where transfers remained negative in the 1980s, although far below their earlier values. INDIRECT PRICE INTERVENTIONS AND NET TRANSFERS. When the income transfers resulting from indirect price interventions are also 134 Consequences of Price Intervention for Income Transfers and Equity taken into account, most countries had extraordinarily high total net transfers out of agriculture throughout 1960-84 (table 7-2). The excep- tions were Brazil, Egypt (after 1981), Korea (after 1969), and Portugal, which had positive net transfers into agriculture. Brazil is an atypical case, with large input subsidies and public investment resulting in high transfers into agriculture. 10 Korea was the only country in which most transfers into agriculture came through output-related price interventions rather than through input subsidies or public investment. The other fourteen countries show a systematic pattern of substan- tial net income outflows from agriculture as a result of indirect price interventions. In twelve countries, total net outflows were equivalent to more than 20 percent of agricultural GDP in most years, reaching as high as 100 percent or more in Ghana and Zambia (table 7-2). Even where overall transfers into agriculture from direct price and nonprice interventions were substantial, as in Turkey, the income transfers out of agriculture as a result of indirect interventions generally out- weighed them. Whatever agriculture gained from sector-specific poli- cies was more than taken away by currency overvaluation and indus- trial protection. For most countries in most periods, direct and indirect price inter- ventions reinforced each other. Brazil and Korea are the only coun- tries in the sample in which positive income transfers to agriculture through input subsidies and public investment (Brazil) or output poli- cies (Korea) offset the negative transfers from indirect price interven- tions. In the other countries the domestic currency was so overvalued that its effect dominated the price and nonprice transfers and contrib- uted to a substantial resource flow out of agriculture over the entire period. For the representative group (Group II), the net outflow from agriculture constituted almost 37 percent of agricultural GDP. For the extreme discriminators (group I), total price interventions reduced agricultural income by half (the outflow was 103 percent under assumption 1). Agricultural Price Policy and Government Agricultural Investment: The Compensation Hypothesis Some observers have argued that taxing agriculture does not neces- sarily harm the sector because the government often uses these tax revenues to invest in the sector.1" To examine the hypothesis that government investment compensates for taxation of agriculture, we estimated empirically the relation between government investment and measures of agricultural price policy and agricultural income. Our estimates apply to the sector as a whole rather than to specific crops, since compensation of specific crops does not necessarily imply net compensation to the agricultural sector. 12 The Effects of Intervention on Income Transfers to and from Agriculture 135 Government investment in agriculture might be positively or nega- tively related to agricultural incentives or it might be unrelated. We tested the null hypothesis that government investment expenditures are either unrelated or positively related to agricultural price policy or income (Ho) against the alternative hypothesis of a negative relation- that government investment expenditures compensate for adverse changes in price policy or income. In most cases the hypothesis of compensation is rejected: investment policy does not seek to compen- sate for the negative effects of agricultural price policy on agricultural incentives. That means that agricultural price policy is the major determinant of agricultural gross income. Testing the Compensation Hypothesis For the measure of government investment, we used an index of government investment bias, defined as the share of agricultural investment in total government investment relative to the share of agricultural GDP in total GDP (see appendix 7-2 for a detailed descrip- tion of the model and the hypothesis tests). Our measures of agri- cultural price policy and income included direct and total nominal protection rates, the producer price index for the selected agricultural products relative to the index for nonagricultural products, the effects of direct price interventions on the government budget, and income transfers to or from agriculture as a result of direct and total price interventions (average of the results under assumptions 1 and 2). We ran regressions (ordinary least squares) of the government invest- ment bias index against each of these six measures for each of the fourteen countries for which data were available (see table 7-4). We then estimated a regression for all fourteen countries with pooled cross-section and time-series data and using a country dummy vari- able for the intercept. The null hypothesis (no relationship or a positive relationship) is rejected in favor of the compensation hypothesis in only five countries-Chile, Egypt, Ghana, Morocco, and the Philippines. And only in Egypt and Morocco were the results significant for all six agricultural price policy or income measures. In these countries pro- ducers seem to have received some compensation, through higher government investment expenditures, for the large consumer food subsidies that were partly financed through low producer prices. The opposite effect was seen in five other countries-C6te d'Ivoire, the Dominican Republic, Korea, Pakistan, and Thailand. In these coun- tries decreased net protection and lower agricultural income have been accompanied by lower government investment, or, conversely, higher protection has been accompanied by higher public invest- ment. For the remaining four countries-Malaysia, Portugal, Turkey, and Zambia-the results were inconclusive. The results for the pooled Table 7-4. Effects of Agricultural Price Policy on the Index of Government Investment Bias: Estimation Results Direct Total Producer Direct price Total price protection protection Budget price indexb transfersc transfers, Country rate (NPR,) rate (NPR) effect (B) (P/PNA) (TD) (7y) Chile -2.69387 -1.79970 0.01275 0.01199 -0.05065 -0.00294 (-2.38) (-1.73) (0.09) (1.28) (-3.27) (-1.06) C6te d'Ivoire 0.72352 0.35487 _ -0.00086 0.00251 0.00128 (2.43) (1.23) - (-0.66) (2.74) (2.30) Dominican Republic 1.31814 1.81452 0.04306 0.00084 0.01240 0.01728 (2.18) (2.29) (2.12) (0.31) (2.02) (3.42) Egypt -0.84566 -1.18020 -0.04385 -0.00900 -0.02011 -0.00622 (-2.52) (-2.79) (-4.19) (-2.62) (-4.06) (-3.87) Ghana -0.35921 -0.74629 -0.00385 -0.00645 -0.00194 -0.00038 (-1.69) (-1.99) (-1.21) (-1.99) (-1.68) (-0.86) Korea, Rep. of 0.63168 0.91013 -0.09154 0.00422 0.00225 0.00832 (2.91) (5.52) (-2.85) (8.47) (0.91) (6.43) Malaysia -1.10146 -1.02999 0.02530 -0.01062 -0.01017 -0.00855 (-1.09) (-0.77) (2.23) (-3.53) (-1.15) (-0.76) Morocco -0.35801 -0.55478 -0.10699 -0.00415 -0.00657 -0.00480 (-1.36) (-1.25) (-0.83) (-1.40) (-2.42) (-1.99) Pakistan 0.62937 1.22560 -0.00960 0.00604 0.00559 0.00315 (6.82) (5.97) (-0.43) (1.55) (3.45) (1.90) Philippines -0.92362 -0.98575 -0.05239 0.01251 -0.01215 -0.00817 (-2.12) (-1.80) (-3.03) (4.19) (-1.74) (-1.31) Portugal -0.35690 -0.86405 0.01017 0.10800 -0.01173 (-0.35) (-1.94) - (2.63) (1.00) (-2.14) Thailand 3.13955 3.41878 0.08971 0.01144 0.02334 0.01685 (55.03) (3.38) (6.74) (1.42) (3.80) (1.85) Turkey 0.17947 0.16871 0.00480 0.00029 0.00180 0.00083 (2.02) (1.05) (0.46) (0.30) (1.03) (0.55) Zambia 0.36256 -1.17723 0.02838 0.00816 0.00032 -0.00112 (0.76) (-2.53) (0.88) (2.76) (0.05) (-2.57) All countries 0.32204 0.40652 -0.00807 0.00353 0.00142 -0.00018 (3.24) (3.36) (-1.57) (6.55) (1.31) (-0.39) - Not available. Note: The index of government investment bias is the share of agriculture in total government investment relative to agriculture's share in GDP. The number in parentheses is the t-statistic. a. Government price-related outlays as a percentage of total fiscal expenditures. b. PAPNA is the producer price index for the selected products relative to the producer price index for nonagricultural products; base = 100 in the first period. c. Direct and total transfers to agriculture as a percentage of agricultural GDP. Source: Authors' calculations as explained in the text. 138 Consequences of Price Intervention for Income Transfers and Equity regression show that the null hypothesis cannot be rejected in favor of the compensation hypothesis for five of the six price policy or income measures. What Do the Results Mean? What do the coefficients of the various measures of price intervention (table 7-4) indicate about the size of the impact of these variables on the government investment bias index? For the pooled regression, the (point) estimates indicate that on average for the fourteen coun- tries, a 1 percentage point increase in the nominal rate of direct pro- tection (say, from -21 to -20 percent) is associated with a rise in the government investment bias index of 0.32 percentage points (say, from 20 to 20.32 percent). A 1 percentage point increase in the total rate of protection is associated with a 0.41 percentage point rise in the index. A 1 percentage point increase in price-related budgetary out- lays to agriculture (in relation to total fiscal expenditures) is associated with a 0.008 percentage point fall in the index. And a 1 percentage point increase in the producer price index relative to its base value is associated with a 0.0035 percentage point increase in the government investment bias index. Overall our results seem to confirm those of Hayami and Kikuchi (1978), who found a positive relation in Southeast Asian countries between investment in irrigation and the international (border) price of rice. In addition, our results support the findings of Ghai and Smith (1987), who found that discrimination against agriculture in sub-Saharan Africa has occurred, as evidenced by a low share of investment funds for rural development and a low level of public services in rural areas as well as by low prices. Other tests also generally confirm the finding of no relationship or a positive relationship (our null hypothesis). Using the model of price interventions developed in chapter 9, we found no significant relation between nonprice transfers and direct price interventions, for either importables or exportables. And in another model relating changes in nonprice transfers to changes in agricultural GDP and to an error- correction term'3 in which direct intervention and real exchange rate distortion appear, we found no relation between nonprice transfers and real exchange rate misalignment. A negative relation was found between income transfers through direct price interventions and transfers through nonprice interventions, but the compensation through nonprice transfers was small-only 24 percent of the income loss that resulted from direct price interventions. Price policy clearly remains the main determinant of farm income. Why did we find no compensation for most countries, or rather, why was the hypothesis of a positive or no relation between invest- The Effects of Intervention on Income Transfers to and from Agriculture 139 ment and price policy interventions not rejected in favor of the com- pensation hypothesis? Several political-economy considerations sug- gest some answers. First, if a government decides to tax agriculture- whether because of its large share in the country's international trade (trade taxes are easier to administer and collect than other taxes), or because agricultural producers constitute a relatively weak interest group, or because of a development strategy of industrialization through import substitution and the extraction of a "surplus" from agriculture-it hardly seems likely that it would then turn around and use that revenue to compensate agriculture with higher investment expenditures. Second, if the agricultural lobby, including the ministry of agriculture, wants to sell an investment project to the government, the project's attractiveness and thus its probability of being financed is greater the higher its expected payoff, which rises with agricultural protection and with agricultural prices.14 Also, the higher the expected rate of return, the higher the probability that the authorities will recoup the funds invested, and thus the more attractive those investments will appear. Similarly, on the supply side for investment funds, it seems reasonable to expect the planning and budget authori- ties to prefer to allocate funds to projects with high expected rates of return. Winners and Losers from Price Interventions: Transfers between Agriculture, Government, and the Nonagricultural Sector To determine which sectors bore most of the cost and which obtained most of the benefits of direct price interventions, we disaggregated the economy into three sectors: agriculture, government, and non- agriculture. By using a multisector framework in which the sum of the transfers is 0, we were able to apply our estimates of direct price transfers to and from agriculture'5 (discussed earlier in this chapter) and budget effects (discussed in chapter 6) to calculate non- agricultural transfers as a residual (table 7-5). All flows are expressed as a percentage of agricultural GDP and were computed under the assumption that direct price interventions in products other than the selected products were insignificant (assumption 1 in the analysis of transfers). This zero-sum assumption, in a framework in which transfers are computed at the actual levels of output and consumption, applies only in the very short run and only to the direct effects. From a longer-term perspective, the price distortions imply an economic wel- fare loss to the country, and the longer-run production and consump- tion response would probably imply a different pattern of transfers, which we could not measure. The analysis was not extended to the 140 Consequences of Price Intervention for Income Transfers and Equity Table 7-5. Government Budget Effects and Balance of Direct Transfers between Agriculture, Government, and the Nonagricultural Sector, Country Group Averages, 1960s, 1970s, and 1980s (percentage of agricultural GDP) Budget Balance of direct transfers Group Period effecta Government Nonagriculture Agriculture Group I 1960-70 20.60 - - (extreme 1970-79 10.18 -4.09 50.94 -46.85 discrimi- 1980-84 -4.10 - natorsb) Group II 1960-69 - - - (repre- 1970-79 9.82 8.33 6.07 -14.40 sentative 1980-84 6.09 6.70 3.24 -9.94 groupc) Representa- 1960-69 - - - - tive 1970-79 4.01 1.88 12.57 -14.45 group, 1980-84 6.09 6.70 3.24 -9.94 excluding Sri Lanka Group III 1960-69 13.16 10.60 - - Mild dis- 1970-79 1.00 2.52 -5.04 2.46 crimina- 1980-84 -13.52 -12.33 5.12 6.23 tors 1960-84 0.52 -2.94 - Group IV 1960-69 - - - 12.52 Protec- 1970-79 -4.2 -4.67 -13.96 18.63 torsd 1980-84 -1.4 -11.77 -16.14 27.90 - indicates that data were not available for enough countries in the group to calcu- late a group average. Note: Transfers are calculated from the government budget effects reported in chap- ter 6 (table 6-2) and the direct price transfers to or from agriculture reported in table 7-1; transfers to or from nonagriculture were calculated as a residual. a. Percentage of total government expenditure. b. During 1960s it only includes Ghana. During 1970s it includes Ghana and Zambia. c. During 1960s it includes Argentina, Colombia, Sri Lanka, Thailand, and Turkey. During the 1970s it also includes the Dominican Republic, Egypt, Pakistan, and the Philippines. d. Only includes Korea. Information was not available for Portugal. Source: Authors' calculations as explained in the text. income transfer effects of indirect interventions because the results would be ambiguous. The very high values of the indirect effects suggest that even in the short run the assumption of no change in output and consumption would be unrealistic, and the possible The Effects of Intervention on Income Transfers to and from Agriculture 141 effects on the nonagricultural sector of such large indirect effects are not well captured by our framework. The analysis thus focuses on the direct price effects in the very short run in a zero-sum framework which, although inadequate for capturing the longer-term intersectoral effects, is useful for elucidat- ing the motivations of the policy process. From our analysis, we identified three different outcomes or patterns of winners and losers: government and nonagriculture gained, government and agriculture gained, and government and nonagriculture lost. Government and nonagriculture gained. This case, in which the gov- ernment earned net revenues from direct price interventions and agriculture was the loser, is the most common outcome and applies particularly to food-exporting countries that consume a significant proportion of the output of exportables. The representative group of countries (Group II) falls into this category. Argentina, an exporter of food grains and beef, and the Dominican Republic, an exporter of sugar, are clear examples. Also falling into this category is Sri Lanka, which taxed nonfood exports and subsidized domestic rice consump- tion; Pakistan; and in some years, Colombia and Thailand. These countries taxed agricultural exportables that were also wage-goods in the domestic market. Government and agriculture gained. This case applies particularly to countries that are net importers of agricultural products and that do not tax agricultural exportables, such as Chile, and to countries that more than compensate for output price-related transfers through input and credit subsidies, such as Brazil in 1967-80 and Colombia in 1960-69 and 1980-83. The nonagricultural sector lost because of the protection of agricultural importables, mainly food. The mild discrim- inators group (Group III) falls into this category. Government and the nonagricultural sector. This case applies to newly industrialized countries like Korea, with relatively high income trans- fers to farmers financed primarily by urban consumers and, to a lesser extent, by the government. Food subsidies exist, but they do not fully compensate urban consumers for the higher prices resulting from protection. Brazil falls into this category for 1980-83, when input and credit subsidies led to large government outlays in relation to the revenues received from output price interventions. In the most frequent outcome, the government earned net reve- nues from direct price interventions (countries in the representative group). The bulk of these revenues came from trade taxes, particu- larly on exports, and these flows were considerably larger than the government outlays on input subsidies. These findings support the hypothesis that the revenue motive directed price interventions that created a marked antiexport bias in the trade regime. In countries such as Argentina, the Dominican Republic, and Thailand, where agricultural exportables were also important in the domestic market 142 Consequences of Price Intervention for Income Transfers and Equity as food products and where importables were few, agriculture was heavily taxed and the benefits went to the government and to con- sumers in the nonfarm sector. Where domestic consumption of exportables was small and importables were protected, urban con- sumers and some agricultural producers were taxed, with the net effect for the agricultural sector depending on the relative size of input subsidies. At the other end of the spectrum were the countries in the protectors group that financed relatively high income transfers to farmers, primarily through the losses of urban consumers and, to a lesser extent, the government. Conclusion The most striking finding to emerge from this analysis of income transfers to and from agriculture is that the transfers out of agricul- ture were so large. For the representative group (Group II), direct price interventions on the selected products during 1960-84 reduced agriculture's real income by about 8 percent of agricultural GDP (table 7-1). Expanding the product coverage to the rest of agriculture raises this figure to somewhere between 8 and 17 percent, which represents a substantial transfer into nonagricultural sectors year after year over a quarter of a century. And when we also account for the effects of indirect price interven- tions (table 7-2), income transfers out of agriculture rise tremen- dously, once again corroborating the findings throughout this study of the substantial influence of indirect interventions. For the repre- sentative group, total price and nonprice transfers out of agriculture averaged 37 percent of agricultural GDP. For the extreme discrimina- tors (Group I), the income effects of total interventions were enor- mous: if we assign a value of 100 to agricultural GDP under total price interventions we find that it would have reached a value of 240 had there been no price interventions over the period. For the sample group as a whole between 1960 and 1984, government interventions extracted an average annual income flow from agriculture equivalent to 46 percent of agricultural GDP. Also important is what we did not find: input subsidies either did not compensate or compensated very little for the substantial income outflows resulting from interventions in output markets, and in most cases public investment in agriculture did not compensate for the negative effects of price interventions. For the representative group, the income transfer through input subsidies was never higher than 1.1 percent of agricultural GDP, and for all eighteen countries the average for 1960-84 was only 2 percent (for direct interventions). Government compensation through investment was found, to vary- ing degrees, in only five of fifteen cases, with only Egypt and The Effects of Intervention on Income Transfers to and from Agriculture 143 Morocco showing compensation for all six agricultural price policy or income variables tested. Finally, our analysis of net transfers between government, agricul- ture, and nonagriculture showed that government and nonagricul- ture were the winners in most cases and agriculture the loser. In the most frequent case, the government earned net revenues from direct price interventions, supporting the hypothesis that government reve- nue was a primary objective of direct price interventions, resulting in a strong antiexport bias in the trade regime. Appendix 7-1. Estimating Price Policy-Related Transfers Transfers from price-related policies (TR) were defined as the change in real income of farmers as a result of price interventions. Direct Effects We ignore purchased inputs for the moment and define real transfers for the sum of products i as the change in the gross value of output for these products as a result of direct price interventions, measured as follows: YY' (_Q Pi'__ (7A-1) ZTRiD = - / c cm o m' _CP"t 1 CPI -ZW _PI CPI 1+tiXCPI'' where TRiD is the change in real income from direct interventions on product i, Y is the gross value of output of product i (PiQio), Pi is the domestic producer price of product i, P is the border price of product i, Qio is the actual value of output of product i, CPI is the actual rural cost of living index, CPI' is the adjusted rural cost of living index corrected for the removal of direct price interventions, and the two terms 1/1 + ti and CPI/CPI' represent the price distortions resulting from direct price interventions on agricultural production and on prices paid by rural consumers, respectively. In computing the sum of transfers, the selected products are weighted according to their share in agricultural GDP. TRiD > 0 implies transfers to agriculture from direct price interventions. The transfer from direct price interventions on tradable inputs (TRDi), such as pesticides, fertilizers, machinery, and equipment, was measured by (7A-2) TRDj =I X ) - 1 UIP-C) [ Y-; IPtc' 144 Consequences of Price Intervention for Income Transfers and Equity which is equivalent to equation 7A-1 on output and measures the effect of price interventions on the aggregate value of purchased inputs. Equation 7A-2 is adjusted easily for credit subsidies, if tj is interpreted as (r - r*)Ir*, where r is the interest rate actually charged to farmers and r* represents the reference price for borrowing. Sim- ilarly, PjQj is then interpreted as the total credit outstanding (stock) to agriculture. Total Effects Real transfers resulting from direct and indirect price interventions for the sum of products i are measured by ZR,=Y -Y* Iz Qio Qi " CPI CPI* IL-Qnp CPI*2\EJ (7A-3) = iQio 1 CPI* E*] CCPI [1 1 +j t( CPI* )(EO where Eo and E* represent the nominal and the corrected exchange rate and CPI/CPI* and E*1E0 are common to all products. Similarly, the transfer from total price interventions on purchased input j was measured by (_io E*Qo0o 1 CPI E (7A-4) TRi = ____ - 1 Li +Q- 1/ CPI*-) Eo) / CPI CPI 1l + t,\CPI* (Eo where TRj > 0 represents a transfer to agriculture (we expect net subsidies to agriculture through input pricing). All variables were measured at the actual value of output and input use. Because nonprice transfers and, to a large extent, input price inter- ventions apply to a broader segment of agriculture than the selected products, the product coverage was alternatively expanded to the whole agricultural sector. In measuring transfers through input markets, the value of tf was obtained from different sources in the country studies. Some authors included measures of effective rates of protection, which yielded esti- mates of tj directly for those cases. Others provided estimates of budget outlays on subsidies on individual inputs, which were then used to measure the equivalent tariff on those inputs. Given esti- mates of the effective rates of protection in the same country, it is possible to estimate the share of tradable inputs Z R QIRQi in the production of i. In measuring the input transfers for the rest of the products, their share of tradable inputs was assumed to be the same as that for the selected products. To obtain TRj, the estimates of credit subsidies (and replanting subsidies for rubber) were added to the price-related transfers on inputs. The Effects of Intervention on Income Transfers to and from Agriculture 145 Finally, the effect on agricultural value added was obtained as the sum of the output and input price interventions to which the value of the non-price-related transfers was added to obtain the net transfers. Adjustment for Changes in the Cost of Living of Rural Households The farm household consumer price index (CPI) was defined as (7A-5) CPI = 0CPIA + (1 - 0) CPINA, where CPIA = 3iPAi measures the agricultural component and CPINA measures the nonagricultural component of CPI. When only the direct interventions are removed, CPINA remains unchanged and only CPIA is affected. In this case, CPI is measured as (7A-6) CPI' = CPIA + (1 - 0) CPINA, where CPIA = i3P jA (and E0j = 1). In the absence of total interventions, CPI is measured as (7A-7) CPI* = CPIA + (1 - /) CPIA, where CPIA = E0jPAj. In most cases, PNA was used as a proxy for CPINA and for CPIlA.16 Each product experiences a different real transfer (direct as well as total). The denominators CPI, CPI', and CPI* are common to all prod- ucts, although consumption patterns within agriculture probably dif- fer according to farm size or income levels, which produce a different output mix.17 Appendix 7-2. The Compensation Hypothesis Model The Model The measure of government investment used in our test of compen- sation at the sectoral level is an index of government investment bias (GIB), defined as the share of agriculture in government investment relative to agriculture's share in GDP: AGI I (7A-8) GIB =AGI AGDP ' GDP where AGI is public investment in agriculture, I is total public invest- ment, and AGDP is agricultural GDP (evaluated at prices in the absence of direct intervention). One way to test for discrimination against agriculture in public investment is to determine whether the social rate of return to public investment (evaluated at nonintervention prices) is larger or smaller 146 Consequences of Price Intervention for Income Transfers and Equity in agriculture than in the rest of the economy. If it is larger in agricul- ture, that implies that investment in agriculture is below the optimum and thus that there is discrimination against the sector. We did not attempt to estimate those rates of return because equality of social rates of return across sectors is compatible with GIB larger, smaller, or equal to 1, depending on the productivity of public investment in each sector. We find values of GIB smaller than 1 in most countries, implying either that public investment per unit of output is less productive in agriculture than in the rest of the economy or that there is discrimina- tion against agriculture. Exceptions are the Dominican Republic (where GIB rises from 0.8 in 1960-72 to 1.48 in 1976-84), Korea (GIB rises from 0.59 in 1960-72 to 1.56 in 1976-84), the Philippines (GIB of 1.18), Thailand (GIB rises from 1.35 in 1960-72 to 2.08 in 1976-84), and Chile (GIB is larger than 1 only if the expenditures related to the agrarian reform are included). We are not concerned here with the level of GIB as much as with its evolution in relation to agricultural price policy, however. The GIB index was regressed individually against six measures of agricultural price policy or income: direct and total nominal protection rates (NPRD and NPRT), the producer price index for the products exam- ined relative to the index for nonagricultural products (PAIPNA), the effect of direct output and input price interventions on the govern- ment budget (B), and the transfers to agriculture as a result of direct and total price interventions (TD and TT). The budget effect (B) of direct price interventions includes the effect of export and import taxes, production subsidies, credit subsidies, input subsidies, and the profits or losses of state agencies as a result of those interventions (see chapter 6). Consumer food subsidies are not included because they do not entail transfers to the agricultural sector. The budget effect is defined as net budgetary outlays related to price policy as a percentage of total fiscal expenditure. The transfers due to direct and total interventions (TD and TT) mea- sure the impact of these interventions on the real income of the entire agricultural sector, not only on the subsectors for the products analyzed. Since we achieved similar results for each country whether we used assumption 1 (NRPD = 0 for the rest of agriculture) or assumption 2 (NPRD = the average level of direct protection for the products analyzed) to calculate income transfers, we used an average of the results for the two assumptions. This implies using a level of direct protection for the rest of the sector equal to half the level of direct protection for the products examined. Furthermore, for TT, we assumed that indirect protection (NPRI) for the rest of the sector was the same as for the products analyzed. This is a plausible assumption since most agricultural products are tradables and so are affected in The Effects of Intervention on Income Transfers to and from Agriculture 147 the same way by indirect policies. Finally, both TD and TT are expressed as a percentage of agricultural GDP. The following regression was estimated for the six variables: (7A-9) GIBt = a + Xt + u, where X = NPRD, NPRT, PAIPNA, B, TD, and TT, respectively, for each regression. The null hypothesis (Ho) is that government investment expendi- tures are either unrelated or positively related to agricultural price policy or income (Ho: 0 Z 0), the alternative hypothesis (Hj) is that government investment expenditures compensate for adverse changes in price policy or income (Hj: 1 < 0). The six regressions were estimated separately for each of the four- teen countries for which data were available. Another regression was estimated with pooled cross-section and time-series data for the four- teen countries, using a country dummy variable to allow the country intercepts (a) to vary. Empirical Results We are interested here in the coefficient 1, which shows the impact of the price policy variables on GIB. Compensation was found in five countries, to varying degrees. In Egypt, 1 < 0 and is significant at the 2 percent level or better for all six variables.. (The confidence levels presented are for a one-tailed test of hypothesis Ho.) In Chile, the coefficients of NPRD, TD (significant at the 2 percent level), and NPRT (significant at the 6 percent level) are negative, but the coefficient of PAIPNA is positive and significant at the 11 percent level. In Ghana, the coefficients of NPRT, PAIPNA (significant at the 4 percent level), and NPRD and TD (significant at the 7 percent level) are negative, and the others are not significant. In Morocco, all six coefficients indicate compensation, with those of TD and TT significant at the 5 percent level, those of NPRD and PAIPNA at the 10 percent level, and the other two not significant. In the Philippines, the coefficients of NPRD, NPRT, B, and TD are negative and significant at the 5 percent level or better, and the coefficient of PAIPNA is positive at the 1 percent level. Over the periods 1960-72 and 1976-84, the GIB index increased in the Philippines from 1.01 to 1.26 while NPRD fell from 5.8 percent to -10.9 percent. In Egypt and Morocco, large consumer food subsidy programs were partly financed through low producer prices, and producers seem to have received some compensation through higher government investment expenditures. The opposite results were found for five other countries-C6te d'Ivoire, the Dominican Republic, Korea, Pakistan, and Thailand. In those countries, most coefficients indicate that decreased net protec- 148 Consequences of Price Intervention for Income Transfers and Equity tion and lower farm income have been reinforced by lower govern- ment investment expenditures or, conversely, that higher protection has been accompanied by higher government investment expendi- tures. For instance, in Korea the GIB index increased from 0.59 in 1960 to 1.56 in 1976-84 while MPRD increased from 35.4 percent to 90.8 percent and NPRI increased from -34.6 percent to -13.3 per- cent. In Pakistan, the GIB index fell from 0.91 to 0.54 between 1960-72 and 1976-86, while NPRD fell from 16.1 percent to -30 percent. Finally, our results are inconclusive for the four other countries- Malaysia, Portugal, Turkey, and Zambia. In Turkey, all coefficients are positive, but only that of NPRD iS significant. For the pooled regression, three coefficients (those of NPRD, NPRT, and PAIPNA) are significantly positive at the 1 percent level and one (TD) at the 10 percent level, while one (B) is negative and significant at the 7 percent level. Thus, for most coefficients, Ho cannot be rejected. So, in five of fifteen cases (fourteen countries plus the pooled data) Ho cannot be rejected in favor of H1, the compensation hypothesis. But the results against the compensation hypothesis are even stron- ger than this summary suggests because the test is biased in favor of the compensation hypothesis for the regression with PAIPNA. To show this, assume that government agricultural investment policy is inde- pendent of agriculture's terms of trade (PA/PNA) so that 0 would be expected to equal zero. But when PA/PNA falls because of a fall in world agricultural prices (P'IPNA), the share of agriculture in GDP-the denominator of GIB-falls and GIB increases. We would then observe a negative value for 3, suggesting the presence of a compensation policy, when in fact the investment policy is independent of PAIPNA* Despite this bias, at the 10 percent level 0 is positive in seven of the fifteen regressions (including the pooled one) with PAIPNA, is negative in four, and is not significant in four, so the hypothesis Ho is rejected in favor of compensation in only four of fifteen cases for PAIPNA. We also corrected for possible bias against the compensation hypothesis in the case of NPRD. If price policy is used to stabilize domestic producer prices, so that NPRD rises when the relative border price PA/PNA falls (and conversely), then the test for NPRD might be biased against compensation. To correct this, we ran a regression that separated the effects of NPRD and P'IPNA: (7A-10) GIB, = (o + 1(NPRD)t ± 02(pAIpkA)t 0 *Ot' Equation 7A-10 decomposes the impact of PAIPNA on GIB into the impact of NPRD and that of P[PNA, since (1 + NPRD)PAIPNA = PA PNA- When GIB was regressed on NPRD alone (using equation 7A-9), we found 0 < 0 in five cases (at the 10 percent level), 0 > 0 in seven cases, and 1 not significantly different from zero in three cases- The Effects of Intervention on Income Transfers to and from Agriculture 149 similar to the results obtained when PAIPNA was used. With equation 7A-10, however, we found 01 < 0 in five cases, 01 > 0 in eight cases, and 1 not significant in two cases; and we found 02 < 0 in two cases, 02 > 0 in seven cases, and 02 not significant in six cases. Thus, the regression using equation 7A-10 confirms previous results, with the hypothesis Ho rejected in favor of compensation in five of fifteen cases with NPRD and in two of fifteen cases with PA/PNA. We also attempted to correct for serial correlation of the residuals, but this had little impact on the results. Another possible problem with the results is that the residuals of more than half the ordinary least squares (oLs) regressions exhibited first-order autocorrelation. This could happen for several reasons. One possibility is that there may be a lag between a change in agri- cultural pricing policy or in world commodity prices and a change in agricultural investment. To test for that possibility, we estimated the following regressions, in which the explanatory variables X are lagged one period: (7A-11) GIB, = a' + 0'X,_1 + u'. The estimation results did not change the general finding that Ho is not rejected in most cases. This may be due to a high degree of autocorrelation of the explanatory variables. One difference using equation 7A-11 is that we found lower levels of significance in the countries where compensation was found using equation 7A-9. We also estimated all regressions using generalized least squares (GLS) to correct for the first-order autocorrelation of the residuals. The regressions were estimated with a maximum likelihood procedure. The results under GLs do not affect the general finding that Ho is not rejected in most cases. For Chile, where compensation was found with OLS, the coefficient of PAIPNA becomes significantly positive with GLS. In the five countries where lower agricultural protection, prices, or income were found to be accompanied by lower agricultural invest- ment with OLS (16 > 0), the results with GLS indicate no change for C6te d'Ivoire, Pakistan, and Thailand, and are inconclusive for the Dominican Republic and Korea, although they still indicate that Ho is not rejected. For the four countries with inconclusive results under OLS, the results with GLS show no change except for Turkey, where 0 is signifi- cantly positive in five of the six regressions. Consequently, under GLs, as in the case of OLS, Ho is rejected in favor of the compensation hypothesis in only five cases. Effects of Intervention on Income Distribution and Poverty Because low-income families typically spend half or more of their income on food and because the bulk of the poor in rural areas depend on agricultural employment, government policies affecting agricultural prices can have a direct and powerful impact on the living standards of the poor. To ensure a regular supply of the basic staples at low prices to consumers is a stated objective of agricultural price policies in most developing countries. Food subsidy schemes in Egypt, Pakistan, and Sri Lanka are well-known cases, but most of the countries studied in the project had at least some measures aimed at reducing the mean level of food prices. Government policies cannot protect all consumers, however, and faced with budgetary pressures, most governments try to reduce the fiscal burden of food subsidies by transferring part of the cost to producers of farm products, through price controls. So not all groups have benefited equally from price interventions, and some have borne the costs. In this chapter we look at the effects of price interventions or var- ious income groups in urban and rural sectors and then at the distri- bution of gains and losses for homogeneous income categories coun- trywide. In the first three sections we deal with the immediate or short-run effects of price interventions on real income. In the longer run, however, price interventions will also influence the level of out- put, employment and nominal wages, and profits in agriculture in a complex set of interactions that determine the long-term income effect of price interventions. In the fourth section of this chapter we look at these dynamic effects of price interventions on the poor in rural areas, the very small-scale farmers and landless workers who are considered in most studies on poverty to constitute the bulk of the poor in developing countries. In the longer run, the high levels of total taxation of agriculture (documented in chapter 2) can be expected to have significantly reduced employment opportunities in agriculture compared with the effect of a more neutral structure of incentives. 150 Effects of Intervention on Income Distribution and Poverty 151 Without price interventions the relative incentives for agricultural productivity would have been much stronger, leading to a more dynamic sector with larger output and productivity gains over time and positive effects for the economy as a whole (as discussed in chapter 4). Thus, in assessing the net effect of price interventions on poverty, we look at the combined effect of the short-run gains to consumers from relatively lower food prices, and the cumulative (longer-run) negative effects on overall agricultural output, employ- ment, and rural income. We find that in the long run the net effect of price interventions was, in many cases, regressive, reducing the real income of the poor more than that of other income groups. For the analysis of the income effects of price interventions, con- sumers were grouped into low-, medium-, and high-income catego- ries (see appendix 8-1 for details on how the real income effect was measured). Measuring the very short-run impact of price interven- tions on urban households' income is conceptually quite straightfor- ward. It is assumed that nominal income remains constant, and the adjustment consists of measuring the effect of interventions on the cost of the household consumption basket for the various income classes. Estimating the short-term impact for rural households is con- siderably more complex because price interventions affect both their cost of living as consumers and their nominal income as producers, and because basic information on incomes and expenditures of rural households is lacking for most countries. The analysis is further com- plicated by the wide diversity of rural households with respect to their standard of living, products they produce, their participation in market activities, and whether they are landless workers or farmers. Food Prices and the Poor in Urban Areas Findings of Previous Studies Previous studies of the real income effects of explicit food policies provide some notion of the size of the effect on consumers' income of changes in food prices relative to the actual open market prices. Mel- lor (1978), using 1964-65 household data for India, concluded that the bottom two income deciles reduced their total real expenditures by 5.5 percent in response to a 10 percent increase in food grain prices. For Egypt, which has a long tradition of explicit food subsidies, Alderman (1988a), using 1981-82 household data, concluded that subsidies associated with public food distribution accounted for 6 percent of household income in urban areas and 7 percent in rural areas. He found that the absolute value of the subsidy was slightly higher for upper- than for lower-income households, but that the poor received a much larger percentage of their total income from 152 Consequences of Price Intervention for Income Transfers and Equity subsidies than the rich. In a study of the wheat flour ration-shop system in Pakistan, Alderman (1988b) concluded that the transfer in 1986 was equivalent to 1.3 percent of average total household expen- ditures for the poorest third of urban households and up to 4 percent when only ration-shop users were considered. And Pinstrup- Andersen (1988), after reviewing the evidence on implicit income transfers from consumer-oriented food subsidies for nine countries, concluded that the subsidies accounted for a significant proportion of the real income of the poor. These studies of countries with large subsidy schemes indicate that explicit food policies (direct price interventions) raised the real income of lower-income urban households by about 4 to 6 percent. Before presenting our own findings, however, we need to point out that the measure of the real income effects used in this project differs from that used in previous studies in two ways: we used the retail price equivalent of the border price instead of the actual open market price as the reference price for estimating the direct nominal rate of consumer protection,' and we included the effects of indirect price interventions, whereas the other studies did not (see appendix 8-1 for a discussion and derivation of the measures).2 Also, the results in the studies cited previously and those we pre- sent in the first three sections of this chapter refer exclusively to the partial equilibrium and immediate effects of price interventions. Whether the short-run income transfers to urban consumers are sus- tainable in the longer run or decline as a result of budget pressures is a relevant consideration, as the reform of food subsidy programs in Pakistan and Sri Lanka makes clear. Furthermore, the longer-term general equilibrium effects of price interventions on poverty allevia- tion could differ significantly over time, as adjustments occur in out- put, employment, and wages in the various sectors.3 The focus on the short-term impact, however, although clearly not presenting the whole picture, is still valuable for its insights on the political economy of price interventions, since this short-run dimension is likely to influ- ence the political calculations that enter into price policy formulation and reform. Size and Pattern of Short-Run Income Effects Contrary to the widely held view that cheap-food policies are preva- lent in developing countries, the impact of direct price interventions on the real income of urban households during 1960-84 suggests that food price policies penalized urban consumers (at the official exchange rate) in about 43 percent of the sample countries (tables 8-1 and 8-2). Most countries protected the production of agricultural importables, mostly food (see chapter 2), but only a subset of these Effects of Intervention on Income Distribution and Poverty 153 countries also reduced food prices to consumers through explicit food subsidies. And although direct interventions subsidized urban con- sumers in eight of the fourteen countries for which data were avail- able, in only four of them (Argentina, Egypt, Pakistan, and Turkey) were the real income gains on both low- and middle-income house- holds higher than 3 percent of their income. Indeed, the income gains or losses from direct interventions were small in most countries for all three income levels. In some cases, such as Cte d'Ivoire during 1970-82, low-income consumers experienced a substantial income loss as a result of the protection of rice production. So despite wide- spread interventions in food prices, the impact on the real income of urban households was generally small. There is, however, a small subset of countries in which the income gains of poor urban consumers were substantial-over 20 percent in Egypt and between 3 and 10 percent in Argentina, Pakistan, and Turkey. In Thailand, the only other country in which gains were at least 10 percent, only the upper-income groups benefited substan- tially; lower- and middle-income groups gained less than 2 percent. In Egypt, urban consumers benefited-especially during the years of relatively high world prices for cereals in the 1970s-from the govern- ment's policy of adjusting food prices with a considerable lag and by a much smaller amount compared with the increase in world prices (Dethier 1990). In the years after the food riots of 1977, consumer prices of wheat in Egypt fell in real terms while world prices increased sharply. In Pakistan the combination of an export tax on basmati rice and a subsidy on wheat flour resulted in a significant income transfer to lower-income urban households. In Argentina consumer gains were about 3 to 5 percent for the low- and medium-income groups before 1980 and about 2.8 percent after that, and in Turkey these groups gained between 6 and 12 percent in the 1970s and 1980s (but less than 2 percent in the 1960s). Looking at the income distribution effects on low-income con- sumers for the four country groups shows that on average over the period direct price interventions had a positive and modest effect in the representative group (Group II)-5.6 percent of household income, or 2.5 percent excluding Egypt-and a small negative effect for the other groups-an income loss of less than 1.7 percent (table 8-1). The representative group includes countries that taxed exports of staples, such as Argentina and Thailand, and countries with large explicit food subsidies, such as Egypt and Pakistan. The income effects for Egypt are so influential on this group that the average income effect declines from 5.6 to 2.5 percent when Egypt is excluded. The income effect is small for the mild discriminators (Group III) and the protectors (Group IV) because,. with average incomes four times the level of the extreme discriminators (Group I), Table 8-1. Short-Run Real Income Effects of Price Interventions on Urban Households, by Income Levels, 1960-84 (percentage of annual real income) Direct interventions Total interventions Low- Medium- High- Low- Medium- High- Country Period income income income income income income C6te d'Ivoire 1960-69 21.40 4.10 2.60 32.10 18.40 14.80 1970-79 -16.30 -2.80 -4.50 25.10 39.10 34.90 1980-82 -35.00 -7.00 -7.67 16.00 59.67 55.67 1960-82 -2.35 -0.35 -1.83 26.96 32.78 28.87 Zambia 1967-69 1.65 1.70 0.90 2.22 2.27 1.20 1970-79 3.65 3.72 1.96 5.06 5.15 2.73 1980-84 1.74 1.76 0.94 5.10 5.22 2.76 1967-84 2.79 2.84 1.50 4.60 4.69 2.48 Group I average 1960-69 11.52 2.90 1.75 17.16 10.33 8.00 1970-79 -6.33 0.46 -1.27 15.08 22.13 18.81 1980-84 -16.63 -2.62 -3.36 10.55 32.45 28.21 1960-84 0.22 1.25 -0.17 65.78 18.74 15.68 Argentinaa 1960-69 5.04 5.04 2.39 9.89 9.89 8.54 1970-79 3.33 3.33 1.59 7.63 7.63 12.87 1980-85 2.83 2.83 1.34 14.31 14.31 9.81 1960-85 3.87 3.87 1.85 10.04 10.04 10.52 Colombia 1960-69 -2.18 -0.70 -0.89 1.67 0.50 0.69 1970-79 1.46 0.42 0.63 3.71 1.14 1.52 1980-83 -1.03 -0.30 -0.43 2.85 0.89 1.20 1960-83 -0.47 -0.17 -0.18 2.72 0.83 1.12 Dominican Republic 1966-69 -0.01 -1.42 0.00 0.38 -1.01 -0.01 1970-79 -0.88 -0.32 0.00 -0.58 -0.12 -0.03 1980-84 -0.95 -0.21 0.00 -0.26 0.26 0.21 1966-84 -0.72 -0.52 0.00 -0.29 -0.20 0.04 Egypt 1960-69 7.95 6.78 3.22 22.92 19.10 8.82 1970-79 32.47 27.41 12.44 49.75 41.59 20.82 1980-84 55.20 47.28 21.14 74.06 63.64 31.78 1960-84 27.21 23.13 10.49 43.88 37.00 18.21 Pakistan 1960-69 -5.05 -3.96 1.71 1.85 1.76 0.18 1970-79 3.51 3.65 2.10 7.25 4.67 0.26 1980-86 2.01 7.19 1.03 4.77 3.01 -0.66 1960-86 -0.13 1.54 0.39 4.60 3.17 0.01 1 Philippines 1966-69 -3.20 -2.10 -1.20 2.37 1.88 1.34 1970-79 0.98 0.68 0.47 6.24 4.20 2.61 1980-85 0.38 0.24 0.14 5.80 3.16 1.59 1966-85 -0.04 -0.01 0.04 5.33 3.42 2.05 Thailand 1981 0.19 1.84 10.15 0.22 3.62 12.57 Turkey 1961-69 1.59 0.77 9.24 11.38 9.95 8.99 1970-79 8.49 6.23 4.82 15.78 13.79 12.19 1980-83 11.59 9.60 8.06 18.76 17.20 15.57 1961-83 6.33 4.68 3.59 14.58 12.88 11.52 Group II averageb 1960-69 1.51 1.32 6.03 7.16 6.00 4.13 1970-79 8.61 3.62 3.73 12.91 10.60 7.50 1980-85 9.83 8.19 5.57 14.63 12.98 9.09 1960-85 5.57 4.90 3.67 10.08 8.87 7.17 (Table continues on the following page.) Table 8-1 (continued) (percentage of annual real income) Direct interventions Total interventions Low- Medium- High- Low- Medium- High- Country Period income income income income income income Group II averageb without Egypt 1960-69 0.44 0.41 1.67 4.54 3.82 3.35 1970-79 4.63 3.14 2.28 6.77 5.43 5.28 1980-85 3.35 3.17 3.35 6.14 5.74 5.85 1960-85 2.47 2.29 2.69 5.25 4.85 5.59 Brazil- 1967-69 0.70 0.70 0.50 1.93 1.93 1.60 1970-79 0.30 0.30 0.22 0.19 0.19 -0.03 1980-83 0.73 0.73 0.23 2.65 2.65 3.00 1967-83 0.47 0.47 0.27 1.08 1.08 0.97 Chilea 1960-69 0.37 0.36 0.37 -5.39 -5.06 -4.74 1970-79 -0.79 -0.79 -0.80 -30.83 -29.13 -28.23 1980-83 -1.16 -1.17 -1.18 -2.56 -2.56 -2.57 1960-83 -0.37 -0.38 -0.38 -15.52 -14.68 -14.16 Malaysia 1960-69 -0.48 -0.41 -0.25 0.08 0.06 0.04 1970-79 -0.74 -0.58 -0.36 0.00 0.00 0.00 1980-83 -1.95 -1.55 -0.94 -1.40 -1.11 -0.68 1960-83 -0.83 -0.67 -0.41 -0.20 -0.16 -0.10 Group III averageb 1960-69 0.20 0.22 0.62 -1.13 -1.02 -1.03 1970-79 -0.41 -0.36 -0.31 -10.21 -9.65 -9.42 1980-83 -0.79 -0.66 -0.63 -0.44 -0.34 -0.08 1960-83 -0.24 -0.19 -0.17 -4.89 -4.59 -4.43 Korea, Rep. of 1962-69 -0.31 -0.46 -0.58 9.59 5.84 3.35 1970-79 -3.72 -2.73 -1.87 0.10 0.03 -0.15 1980-84 -6.06 -4.22 -2.84 -4.30 -3.08 -2.04 1962-84 -3.04 -2.27 -1.63 2.44 1.37 0.66 Portugal 1960-69 -1.23 -1.03 -0.54 0.25 0.53 1.08 1970-79 0.88 0.55 0.34 -0.65 -1.06 0.22 1980-84 -1.42 -1.18 -0.60 -6.40 -6.28 -2.10 1960-84 -0.42 -0.43 -0.20 -1.44 -1.47 0.10 Group IV averageb 1960-69 -0.77 -0.75 -0.56 4.92 4.92 3.19 1970-79 -1.42 -1.09 -0.77 -0.28 -0.28 -0.52 1980-84 -3.74 -2.70 -1.72 -5.35 -5.35 -4.68 1960-84 -1.73 -1.35 -0.92 0.50 0.50 -0.05 Sample averageb 1960-69 1.87 0.67 0.36 6.52 4.72 3.20 1970-79 2.73 2.79 1.22 6.34 6.21 4.20 1980-83 1.92 3.86 1.37 9.24 11.21 8.00 1960-83 2.31 2.27 0.96 7.05 6.48 4.40 a. The budget share information provided in the country study did not distinguish between low- and middle-income urban households, so the income effects for these two income groups are the same. b. Simple, unweighted group average. Source: Authors' calculations based on data provided in the country studies. Table 8-2. Impact of Price Interventions on the Real Income of Urban Households, by Income Levels and Size of Impact, 1960-84 Direct interventions Total interventions Effect Low-income Medium-income High-income Low-income Medium-income High-income Consumers gain (lose) more than 10 percent Egypt Egypt Egypt Argentina Argentina Argentina (C6te d'Ivoire) Thailand Egypt Egypt Egypt Turkey Turkey Turkey C6te d'Ivoire COte d'Ivoire Thailand C6te d'Ivoire Consumers gain (lose) between 3 and 10 percent Argentina Argentina Pakistan Philippines Philippines Zambia Turkey Pakistan Turkey Zambia Zambia (Chile) (Korea, Rep. of) Turkey (C6te d'Ivoire) (Chile) Thailand (Korea, Rep. of) Pakistan (Chile) (Korea, Rep. of) (Korea, Rep. of) Pakistan Effect less than 3 percent Brazil Brazil Argentina Pakistan Pakistan Pakistan Philippines Philippines Brazil Brazil Brazil Brazil Thailand Thailand Philippines Colombia Colombia Philippines Zambia Zambia Zambia Malaysia Malaysia Colombia Chile Chile Chile Thailand Dominican Republic Malaysia Colombia Colombia Colombia Dominican Republic Portugal Korea, Rep. of Dominican Republic Dominican Republic Korea, Rep. of Portugal Dominican Republic Portugal Portugal Portugal Portugal Pakistan Pakistan Pakistan Pakistan Malaysia COte d'Ivoire Malaysia Malaysia Note: Countries were included in a category when the results for at least two of the three subperiods analyzed corresponded to that category. Parentheses indicate that consumers suffered real income losses. The studies for C6te d'Ivoire, Ghana, Morocco, and Sri Lanka did not provide estimates of the real income effects on urban households. Source: Authors' classification based on table 8-1. Effects of Intervention on Income Distribution and Poverty 159 the share of food in the consumer price index (ci) is relatively low, reducing the income effect of a given price intervention. The 1970s marked a period of change in the effects of direct inter- ventions on income in most country groups. C6te d'Ivoire (among the extreme discriminators), the mild discriminators, and the protec- tors went from income gains in the 1960s to income losses thereafter; the losses grew particularly large during the early 1980s. The pattern is reversed for the representative group, which went from very small income gains in the 1960s to substantial gains in the 1970s and 1980s. Thus, this group differs from the others in the level, the sign, and the evolution of income effects. For most countries the income effects of indirect price interventions intensified the effects of direct interventions. In Argentina, Brazil, Chile, Colombia, the Philippines, Turkey, and Zambia the effects of indirect interventions were so large that the total effects were at least double the direct effects.4 Among food exporters like Argentina and Thailand, the effects of subsidies to consumers represented by export taxes on food products were reinforced by the effects of indirect inter- ventions. However, in some countries that taxed consumers through direct interventions, such as Korea, Colombia, and Portugal, the indi- rect effects led to an implicit subsidy that counteracted the tax in several subperiods. When indirect price interventions are included, the extreme dis- criminators and the representative group show a marked increase in the real income of low-income consumers-the total effects are almost twice those of the direct effects (table 8-1). For the mild discriminators the indirect effects generally intensify the income losses from the direct effects. Only for Cte d'Ivoire among the extreme discrimina- tors and for the protectors group do the indirect effects have an oppo- site effect to the direct effects, serving to reduce some of the income loss from the direct effects. The total income gain for low-income urban consumers ranged from 0.22 to 5.3 percent in seven of fourteen countries and was 10 percent or more in three-Argentina, Egypt, and Turkey (table 8-2). On average for the urban poor in the thirteen countries, the static distributional gains from total intervention (the cpi impact) averaged 6.0 percent. Overall, these results confirm that in most countries the effects of what are commonly referred to as cheap-food policies take place mainly through exchange rate misalignment and not through food policies or other direct price interventions. The relatively small effect of direct government interventions on the mean level of con- sumers' real income in most countries, together with the greater sta- bility of domestic food prices compared with border prices (see chap- ter 3), suggests that the principal motive underlying food price interventions may have been to prevent sudden large real income 160 Consequences of Price Intervention for Income Transfers and Equity losses in years of higher-than-mean food prices rather than to raise the standard of living of the urban poor. Unstable world market prices for food products, if passed on directly to consumers, can lead to significant fluctuations in the real income of consumers, a politically sensitive outcome in most developing countries. One of the principal findings of the analysis on intersectoral income transfers (chapter 7) was the relatively large income transfers from agriculture toward the nonagricultural sector. In contrast, the analysis in this chapter on income distribution effects at the household level concludes that the real income effect for urban households was quite modest. The reader should note that although the income effects from direct interventions are low, they are also higher for total inter- ventions. For example, for low-income urban households they increase from 2.3 to 7.0 percent. But the main difference lies in the fact that part of the intersectoral transfers were captured by the gov- ernment, as revenues, and not directly by urban households. Another factor that explains the differences observed is that transfers were expressed as percentage of agricultural gross domestic product (GDP), while urban income effects were measured as percentage of household income levels. Short-Run Income Effects in the Rural Sector In chapter 7 we showed that direct interventions resulted in a system- atic pattern of large income flows from the agricultural sector and that indirect interventions intensified these effects. Even when the direct effects resulted in transfers into agriculture, as in Turkey, the negative effects of indirect interventions generally more than counteracted these positive flows, except in the case of Brazil, Korea (after 1969), and Egypt (after 1981). In this section we look at these income effects at the farm or rural household level, by size of holding (small, medium, or large) or income level (low, medium, high). Differences in consumption patterns, production mix, and mar- ketable surplus as a proportion of farm output account for most of the differences in income effects among farm or rural income categories. As for urban income effects, the results represent only the short-term effects, that is, at given levels of production and given consumption patterns. Because of incomplete information (no data on consump- tion patterns of rural households and no data linking size of holding and output mix), rural income effects could be analyzed for only seven countries. Other limitations of the data made it impossible to adjust farm categories or income groups so that income effects could be compared across countries; for that reason, cross-country compari- sons are unreliable, and the focus of the analysis is the results for individual countries. Effects of Intervention on Income Distribution and Poverty 161 Effects in Individual Countries In Ghana and Korea, and to a lesser extent in Brazil, small farmers (or poorer households)5 benefited from direct price interventions (table 8-3). In Ghana, direct interventions benefited smaller farmers and taxed larger farmers; in Korea and Brazil larger farmers benefited even more than smaller farmers. Ghana was the only country of the seven analyzed in which the income effects (as a percentage of real income), whether positive or negative, were greater for small farmers than for larger farmers. The income from farming operations seems to have been affected relatively more for larger farms, which lost from interventions, than for smaller farms, because larger farms tend to sell a higher proportion of their output and tend to specialize more in exportables.6 In Korea, where the typical farm is very small, larger farmers benefited more because they sell proportionately more of their output in product markets where prices are occasionally above border prices. For Brazil we were unable to analyze the effects of different cropping patterns; we did find, however, that in relation to household income, poorer farmers benefited relatively more than wealthier farmers by consuming more of the cheap food, although wealthier farmers captured a higher percentage of the overall income transfer. In Egypt, Thailand, and Turkey, direct interventions taxed both small and larger farmers during most of the period, although in Thai- land (for which we had the most complete information on rural income distribution), the income effect was very small-less than 1.8 percent for all income groups.7 In all three countries, large farmers were taxed more than smaller farmers. For Egypt, the difference by farm size reflects mainly the difference in output mix and consump- tion patterns. Poorer farmers tend to produce maize, while larger farmers tend to specialize in cotton, which was heavily taxed. In Turkey the differences in income effect result mainly from differences in consumption patterns and in the proportion of output sold.8 In Portugal larger farms benefited from direct price interventions and smaller farmers were taxed. The difference in impacts reflected strong regional differences in product mix, which are associated with differ- ent farm sizes and differences in consumption patterns. In almost all cases, indirect interventions worsened the income effects for rural households, reducing or obliterating the income gains from direct interventions, as in Brazil, Ghana, and Korea, or increas- ing the losses, as in most other countries. Portugal is the primary exception, with an undervalued real exchange rate that intensified the positive income effects of direct interventions. The indirect effects reduced the income losses for some periods and some income groups in Ghana (larger farmers) and, to a much smaller extent, in Thailand (all farm size groups). Table 8-3. Short-Run Real Income Effects of Price Interventions on Rural Households, by Income Levels, 1960-84 (percentage of real income) Direct effects Total effects Small Large Small Large Country Period (poor) Medium (rich) (poor) Medium (rich) Brazil 1960-69 - - - - - - 1970-79 8.60 - 11.80 -1.60 - -1.70 1980-83 -0.20 - -0.80 -14.50 - -14.70 1970-83 6.10 - 8.21 -5.30 - -5.40 Egypt 1964-72 -23.40 -34.70 -36.50 -36.30 -48.50 -52.60 1970-79 -21.40 -41.70 -46.90 -35.20 -53.00 -61.90 1980-85 -10.40 -28.80 -32.40 -31.60 -48.00 -54.90 1964-85 -19.20 -35.30 -38.60 -34.70 -49.80 -56.20 Ghana 1962-69 62.20 - -29.00 -28.90 - -36.70 1970-76 150.20 - -37.80 44.10 - -16.90 1962-76 103.20 - -33.10 5.20 - -27.50 Korea, Rep. of 1962-69 8.30 12.70 14.90 -0.60 -12.20 -18.50 1970-79 14.00 25.80 37.60 9.20 14.80 19.90 1980-84 15.70 37.10 57.60 14.00 30.40 47.90 1962-84 12.40 23.70 34.10 6.80 8.80 12.60 Portugal 1960-64 0.30 2.90 64.80 9.30 12.10 86.50 1965-73 -1.00 0.60 72.80 0.30 1.90 76.30 1974-75 -0.50 0.50 -1.50 7.60 8.30 3.70 1976-84 0.00 0.00 20.30 3.20 2.50 19.30 1960-84 -0.40 0.50 34.10 3.30 3.80 37.50 Thailand 1980-81 -1.03 -1.07 -1.76 -0.86 -1.01 -1.57 Turkey 1961-69 -2.56 -3.31 -4.03 -2.82 -3.69 -4.53 1970-79 -2.93 -3.80 -4.63 -2.75 -3.76 -4.80 1980-83 0.42 0.22 -0.35 -0.03 -0.54 -1.01 1961-83 -2.20 -2.91 -3.65 -2.27 -3.14 -4.01 - means that data were not available. Note: The income effect is measured as [(Y-Y*)/Y*]100, where Yis actual real (deflated) income and Y* is income in the counterfactual scenario, with removal of direct or total interventions. See appendix 8-1 for details. Source: Authors' calculators based on data provided in the country studies. 164 Consequences of Price Intervention for Income Transfers and Equity Patterns of Short-Run Income Effects Price interventions had a relatively larger real income effect on farm producers than on urban consumers. Direct price interventions had considerably more impact on the real income of larger farms (or wealthier rural households), whether these effects were positive (Korea, Brazil-except 1980-83-and Portugal) or negative (Egypt and Turkey). Farms of all sizes were taxed heavily in Egypt, especially larger farms, which are more export-oriented and tend to sell a higher proportion of their production. Ghana was an exception, with small farmers gaining from direct price interventions because they produce more rice, which is protected, and larger farmers losing because they produce more cocoa, which is heavily taxed. In countries where direct interventions protected farmers, as in Brazil and Korea, richer farmers or households benefited more than poorer ones. In Portugal direct interventions had almost no effect on the income of small and medium-size farms while benefiting larger farmers. Thailand was the only case in which the real income effects of both direct and total price interventions were negligible during 1960-84. On the whole, the income effects for rural households were sub- stantially higher and in the opposite direction from those for urban households in the same countries. Removing all price interventions would have brought real income benefits to farm households at all levels of living and by a substantial proportion of household income in Brazil (around 15 percent in 1980-83), Egypt (between 35 and 56 percent for all groups throughout 1964-85), Ghana (17 percent to 37 percent for larger farmers during 1962-76), and, to a lesser extent, Turkey (2.8 to 4.8 percent for all income levels during 1961-79). Because of the pattern of direct interventions, removing total price interventions would have reduced the real income of farmers in Korea (except 1962-69), Portugal (for larger farmers), and Ghana (small farmers in 1970-76). Comparing Gains and Losses of Poor Households Across Sectors Low-income urban families would be expected to gain and low- income farm families to lose from direct price interventions that tax food production-and the opposite pattern would be expected when food production is protected. But calculating the effects of price inter- ventions on the real income of the poor countrywide requires that income categories be adjusted to correct for income inequalities between urban and rural sectors. Most studies on poverty in develop- ing countries conclude that the bulk of the poor are in rural areas, Effects of Intervention on Income Distribution and Poverty 165 among very small-scale farmers and landless laborers. To measure the impact of price interventions on the real income of the poor country- wide, we first established homogeneous income categories for the rural and urban areas. Then we consolidated the results for both sectors to assess what share of the income transfers resulting from price interventions went to the lowest-income households countrywide. We used a zero-sum, three-sector framework similar to that used in chapter 7 to measure net income transfers between government, agri- culture, and the rest of the economy. We took the absolute income transfers resulting from direct price-related interventions on outputs and inputs (derived in chapter 6), expressed as a percentage of agri- cultural GDP, and allocated them to the government, urban house- holds, and rural households.9 This framework allows us to determine which income groups have borne most of the cost and which have captured most of the gains of direct price interventions. Where the net gains have gone to the government, we can see how much of the transfer came from the poor and how much from higher-income cate- gories. And where the government lost, we can see how much went to the poor and how much to the higher-income classes. The approach was applied only to Colombia, Chile, Egypt, and Thailand, and only for selected years (table 8-4), although it could be extended to other countries and periods if the required information on household income were available (a more detailed explanation of the methodology is presented in appendix 8-3). The economywide breakdown of income transfers in Egypt in 1984 (left panel of table 8-4) shows that all income groups gained and the government lost (last panel) from direct interventions. The two lowest income quin- tiles gained 2.5 percent of agricultural GDP, but the three highest quintiles captured 7.4 percent, or more than twice as much. If the objective of direct price interventions was to transfer income to the poor, it was achieved at a substantial cost to the government: a trans- fer of 2.5 percent of agricultural GDP to the two lowest quintiles cost the government 9.9 percent of agricultural GDP because of the 7.4 percent leakage to higher-income households. Looking at transfers by urban and rural sectors (middle panel of table 8-4) shows that rural households lost while urban households gained, especially the two highest income quintiles. The results for 1980 are even more skewed toward the higher-income households, and quite different from the effects in 1975. A very different pattern of income transfers is found for Colombia in 1970. The government captured the full transfer, with low and high income quintiles contributing approximately equal shares (2.2 percent for the two lowest quintiles and 2.2 percent for the two highest). In all years examined for Colombia, income transfers went to the govern- Table 8-4. Income Effects of Direct Interventions by Income Classes Economywide and by Rural and Urban Sectors, Selected Years (percentage of agricultural GOP) All sectorsa Rural Urban quintiles quintiles quintiles Country Years (1) (2) (3) (4+5) (1) (2) (3) (4+5) (1) (2) (3) (4+5) Government Colombia 1961 2.50 -0.83 -1.06 -1.93 3.18 1.06 0.43 1.84 -0.68 -1.09 -1.48 -3.78 1.32 1970 -1.33 -0.83 -1.24 -0.99 -1.34 -0.86 -0.26 -1.05 0.01 0.03 0.02 0.06 3.40 1980 -0.29 -1.74 -1.61 -3.95 0.49 0.42 0.08 0.37 -0.78 -2.16 -1.69 -4.32 7.59 1983 0.66 -0.31 -1.19 -2.62 1.32 1.61 0.24 1.02 -0.65 -1.81 -1.42 -3.63 3.45 Egypt 1970 -0.35 -1.03 -0.44 -4.80 -2.24 -3.30 -4.74 -10.99 1.99 2.25 4.30 6.18 6.52 1975 2.40 3.32 2.96 -4.50 -2.08 -1.77 -6.76 -18.46 4.50 5.10 9.72 13.98 -4.20 1980 0.00 0.02 2.35 0.37 -0.27 -0.30 -0.24 -0.50 0.27 0.30 0.59 0.85 -0.74 1984 1.30 1.20 3.53 3.85 -0.09 0.36 -0.44 -0.42 1.37 1.56 2.97 4.27 -8.87 Chile 1961 0.48 -0.10 -0.16 -0.06 0.59 0.07 0.08 0.82 -0.10 -0.17 -0.24 -0.87 -0.19 1970 -0.39 0.03 0.05 0.14 -0.46 -0.09 -0.11 -0.43 0.07 0.11 0.16 0.57 0.19 1980 0.74 -0.13 -0.21 -0.57 0.94 0.21 0.27 1.14 -0.20 -0.30 -0.47 -1.72 0.16 1982 0.47 -0.06 -0.12 -0.48 0.58 0.14 0.16 0.53 -0.12 -0.20 -0.28 -1.01 0.19 Deciles Deciles Deciles Govern- Rest of (1-3) (4-7) (8-10) (1-3) (4-7) (8-10) (1-3) (4-7) (8-10) ment world Thailandb 1962 0.73 0.95 -1.05 -1.15 -2.95 -8.72 1.88 3.90 7.67 0.76 -1.39 1970 0.47 0.61 -0.64 -0.73 -1.86 -5.51 1.20 2.48 4.90 0.34 -0.78 1980 0.57 0.68 -1.25 -1.02 -2.60 -7.72 1.59 3.30 6.50 0.28 -0.28 1982 0.03 0.06 0.10 -0.01 -0.01 -0.04 0.03 0.07 0.07 -0.01 -0.17 a. Net of rural and urban effects by quintile. b. As a result of the effect of export controls on the world price of rice. Source: Authors' calculations based on data provided in the country studies. Effects of Intervention on Income Distribution and Poverty 167 ment. The shift in the burden between lower- and higher-income groups was apparently associated with changes in export prices of coffee, an important exportable taxed by the government. Higher coffee prices raised the income of farm households, switching fami- lies from lower- to middle-income quintiles, whereas lower prices had the opposite effect. In Chile the government also gained during most years, but the transfers to the government were smaller (0.16 to 0.19 percent of agricultural GDP) than they were in Colombia. When agriculture was protected (1980 and 1982 in table 8-4), the gains to the poorest farm households more than compensated for the losses of the poorest urban households, mainly because more poor people lived in rural than in urban areas and possibly because a higher share of labor income went to the poorest households in rural areas as a result of higher employment and wages in agriculture. Finally, for Thailand, the results show that throughout the period 1962-82 all farm households lost (and the richer households lost a higher share of agricultural GDP than the poorer) and all urban house- holds gained (and the richer households gained more than the poorer). Economywide, the lowest- and middle-income households (deciles 1-7) gained slightly, but the sectoral breakdown shows that the gains went only to urban households. In 1980, for example, the lowest seven deciles captured 1.25 percent of agricultural GDP (urban households in these deciles gained 4.9 percent and rural households lost 3.6 percent) and the government captured 0.28 percent. These transfers came from the rest of the world (0.28 percent, as a result of the effect of rice export restrictions on world prices) and from higher- income households (1.25 percent, or the difference between the loss of 7.7 percent for high-income rural households and the gain of 6.5 percent for high-income urban households). These tentative findings on the balance of income transfers in four countries suggest that the lowest- and middle-income households captured part of the short-run income transfers from direct price interventions (left panel of table 8-4) in Chile, Egypt, and Thailand, whereas the government captured most of the transfers in Colombia (except in 1961). But if we assume, hypothetically, that the main objective of price interventions in Chile, Egypt, and Thailand was to transfer income to the poorest households (first quintile or first three deciles), how much additional income could these households have received had no transfers gone to upper-income groups? In Egypt the poorest quintile would have received 9.9 percent instead of 2.5 per- cent of agricultural GDP in 1984 had there been no leakage to quintiles 2 and 3, which received 7.4 percent; in 1975 they received 5.74 per- cent with no leakage to higher income groups, but with a consider- able transfer from the rural poor to the urban poor. In Chile and 168 Consequences of Price Intervention for Income Transfers and Equity Thailand, direct price interventions did not result in substantial trans- fers as a percentage of agricultural GDP. In Chile, the relatively small transfers to the lowest income quintile in 1980 and 1982 were not accompanied by any leakages to higher-income groups, while in Thailand, there were leakages in all years, but they were considerably smaller than those in Egypt. Longer-Term Effects on the Poor Responsiveness of Agricultural Wages and Employment to Agricultural Prices In the long run, the net effect of price interventions may in many cases have been regressive. To the extent that the benefits from food subsidies were captured mainly by urban households, and predomi- nantly by formal-sector and government employees at the expense of the rural sector, the cost may have been a reduction in the real income of both the rural poor and, indirectly, the urban poor in the informal sector. If taxing agriculture reduces rural demand for labor, rural employment and real wages will fall, leading to increased migration to the cities and increased competition for employment, and thus to a fall in income (or an increase in unemployment) in the informal urban sector as well (Harris-Todaro). The net effect would thus be regres- sive, since the bulk of the poor in developing countries are in rural areas, while the poor in urban areas are largely in the informal sector, which is unprotected by minimum wage or other labor legislation. Almost no empirical studies have been done on the responsiveness of agricultural wages to food prices, and the few theoretical studies that have addressed this issue have reached very different conclu- sions. For example, de Janvry and Subbarao (1986) assume that nomi- nal wages in agriculture are exogenously fixed, which implies that higher farm prices hurt rural households. In contrast, Sah and Stiglitz (1987) argue that it is plausible to assume a food price elasticity close to unity for the agricultural wage rate, implying that higher farm prices would lead to higher income for the rural poor. Ravallion (1990), in a study of the rural welfare effects of food price changes in Bangladesh, concludes that the welfare of the typical poor household is likely to be neutral to the price of rice in the long run-although the effect of higher prices is likely to be more positive for the poorest households than for others. These studies ignore several important factors. Ravallion's compet- itive model, as he acknowledges, omits an important channel through which rural households may benefit from higher food prices-the increase in employment at any given wage rate that would take place if involuntary unemployment existed in lean sea- sons. Furthermore, most authors do not seem to recognize the Effects of Intervention on Income Distribution and Poverty 169 employment effect that removal of price interventions could have through changes in the composition of farm output (see chapter 4). Removing taxes on exports, the most heavily taxed agricultural sector (see chapter 2), could induce a reallocation of labor to food and non- food exportables, including nontraditional exports such as hor- ticultural products. The rural employment effect during the harvest season could be considerable, as was the case in Chile during the first three to four years of expansion during the early 1980s before rural real wages began to increase. For this project, we examined empirically the agricultural wage response to farm prices in Brazil, Chile, and Colombia (see appendix 8-2). In Brazil the rural real wage was found to depend positively on agriculture's terms of trade (coefficient of 0.11, significant at the 10 percent level in a one-tailed test) and even more strongly on the land rental price (coefficient of 0.18, significant at the 1 percent level) and agricultural GDP (coefficient of 0.19, significant at the 2 percent level). A reduction in agricultural taxation would have resulted in higher agricultural prices, higher land rental prices, higher agricultural GDP, and so in higher real rural wages. In Chile, improving agriculture's terms of trade were found to have a positive impact on labor demand in fruits, vineyards, livestock, and annual crops, with a positive impact on real rural labor income (defined as the real wage rate times hours of work). Removing total interventions would have had a small positive impact on rural real labor income up to 1972, and a substantial impact thereafter. The Colombia study found an elasticity of the rural real wage rate to agriculture's terms of trade of 0.54 (with a t-value of 16.4).10 Since total price interventions in Colombia resulted in a fall in the rural real wage rate-by 8.8 percent in 1966-70, 11.2 percent in 1971-75, 18.6 percent in 1976-80, and 24.3 percent in 1981-83-the impact on the rural sector and on the rural poor in particular must have been sub- stantial. These policies, through their impact on rural-urban migra- tion, would also be expected to lead to a fall in the real wage rate in the informal urban sector. These studies show that removing interventions would have raised the real income of landless labor, the poorest group in the rural sector and generally the poorest group in the economy. And, by reducing rural migration, the removal of interventions might also have raised the income of the urban poor, although this effect is less clear because of the possible benefits of food subsidies for the urban poor. Estimates of Cumulative Net Income Effects In the long run, the cumulative net effect of price interventions on the real income of the poor depends on the combined effect of the very short-run impact on the cpi and the impact on overall economic 170 Consequences of Price Intervention for Income Transfers and Equity growth, which affects employment and nominal income. (As chapter 4 showed, total interventions reduced both overall economic growth and agricultural growth, which had a negative impact on the income of both the rural and urban poor over time.) These short-run cpi and overall growth effects were estimated for fourteen countries reporting income distribution effects. These estimates provide a rough indica- tion of the direction of change in the real income of the poor as a result of price interventions. On average, total interventions for the fourteen countries taxed agriculture at a rate of about 26 percent, resulting in an average annual loss in GDP growth of 1.1 percent, or about 23 percent after twenty years. For the urban poor, static distributional gains from total interventions (the cpi impact) averaged 6.0 percent. Assuming a con- stant share of labor income in total income, the net effect (static gains plus cumulative impacts on growth) of total interventions on the urban lower-income group from the early 1960s to the early 1980s was a real income loss of over 15 percent. For the urban poor in all coun- tries except Egypt, the loss resulting from the negative impact on growth was higher than the static income gains from price interven- tions, so the urban poor lost over time. In Egypt after twenty years, the net effect on the urban poor was slightly positive because of high static income gains of 43.9 percent and a cumulative loss due to slower growth over twenty years of about 40 percent. Over twenty- five years, however, the net effect on the urban poor is negative, and the effect becomes increasingly negative over time, because the static gain is a once-and-for-all gain whereas the loss in growth accumulates over time. The only country in which the rural poor gained from interventions is Korea where agriculture is a small share of the econ- omy. In Korea, the static effects on the rural poor were about 7 per- cent and the growth impact of interventions was small. Although these results suggest that substantial benefits would accompany the removal of price interventions, efforts to do away with food subsidies have faced strong resistance from urban con- sumers. The loss to urban consumers as a result of such reforms is immediate and easy to see, whereas the benefits of higher growth appear only over time. In Egypt, Morocco, Zambia, and other coun- tries, government attempts to remove food subsidies led to riots and forced a reversal of the reforms. This and other experience with policy reforms demonstrates the importance of gaining a better understand- ing of the political economy of agricultural pricing policies (this topic is examined in the companion volume to this one, volume 5). Appendix 8-1. Measuring the Real Income Effect To analyze the differential impacts of price interventions on various income-level groups, we classified consumers in each country into Effects of Intervention on Income Distribution and Poverty 171 low-, medium-, and high-income groups." This grouping is based on the classifications prepared for the country studies, and the income groups do not necessarily represent the same decile classification in each country. For a few countries, the urban household categories reported were not based on an explicit income class stratification, although they correlated very closely with income levels. For exam- ple, the 1985 Pilot Household Budget Survey for Zambia, which was used by Jansen (1988), distinguished among township "site and ser- vice," and low- and high-cost urban households rather than among explicit income levels. The results presented for total price interventions reflect the short- run effects only. Even in the medium run, it is unrealistic to assume that the major transformation of the trade and exchange rate regimes implied by the removal of indirect price interventions will leave the nominal income of households in the various income groups unchanged, since wages and employment are likely to change in the trade-liberalizing manufacturing sectors. To estimate the price effect of interventions on consumers, we com- pared the prevailing prices paid by urban consumers with the esti- mated border equivalent prices at the retail level (or, in a few cases, at the wholesale level). Only the products selected for analysis in each country (see table 1-1) are covered here, and the price changes are those reported by the study authors. This means that the measures in this chapter report the income effects of changes in the prices of the selected products only, rather than of all agricultural products. The combined share of the selected products in the cpi differed across countries because of differences in consumption patterns and in product coverage in the country studies. How the effect of price interventions on consumer prices is mea- sured can make a considerable difference in the results. For example, how does government intervention in farm-gate prices affect food prices at the retail level in urban areas? Relevant issues include the price chain between the price of raw materials sold by farmers (wheat, live cattle, sugar cane) and the price of processed products consumed in urban areas (bread, beef, refined sugar). The analysis is further complicated by the possibility that the internal marketing sys- tem may not be fully competitive for some products and there may be direct consumer price interventions. This link between farm and retail prices is usually quite specific to each product, so the level of price interventions for consumers can diverge among products even if the level for producers is the same. To frame the problem in terms of a price system that links the various processes between the farm-gate and retail prices, the analyst needs the corresponding conversion factors and extraction rates and the handling, milling, and other processing-cost functions. Here we present a few illustrations of the types of adjustments 172 Consequences of Price Intervention for Income Transfers and Equity required for products for Argentina, Chile, Egypt, Ghana, and Morocco. For Argentina the proportional impact on consumer prices for beef and bakery products resulting from changes in the prices received by farmers was relatively easy to estimate because an explicit export tax on beef and wheat has been the principal direct price intervention on these food markets. Marketing has been handled by the private sector, and the government has intervened very little at the consumer level. Milk in Chile presents a somewhat more complicated case because of processing methods that combine domestic and imported milk products. Farmers sell unpasteurized fluid milk, a nontradable that competes with imported powdered milk. The milk powder is pro- cessed and blended with the domestic pasteurized fluid milk for sale in the urban market. An interregional econometric model was esti- mated and a price model was solved to get nonintervention prices linking the farm-gate price for fluid milk with pasteurized milk sold in urban centers. The dual-price system in Egypt represents a case in which protec- tion to producers of importables such as wheat and sugar coexisted with subsidies to consumers of bread or flour purchased in ration shops. Moreover, most households purchased additional quantities of the same goods at unsubsidized prices in the open market. Con- sumer prices were measured as the average of the prices paid by consumers in the ration shop and in the open market, weighted by the percentage purchased at each outlet. In Ghana, for products in which the government intervened at various levels, such as maize, rice, and sorghum, Stryker (1991) fol- lowed the food chain, including collection, processing, and distribu- tion. He linked the border prices and producer prices with the whole- sale prices for each product and applied a retail margin to the wholesale price to estimate the nonintervention retail price for consumers. To get the nonintervention consumer price for wheat in Morocco, the reported average import price was adjusted for handling charges and losses, domestic transport costs, and a milling margin to get the wholesale price. An effort was made to use an "efficient" marketing margin. Urban Households The short-term impact of agricultural pricing policies on the real income of urban households can be approximated by comparing the urban cost of living under existing pricing policies and without policy-induced distortions. Two estimates were made, one for the effects of direct price interventions and one for the effects of total Effects of Intervention on Income Distribution and Poverty 173 (direct plus indirect) interventions. To compute these effects, we con- structed consumer cost of living indexes (cpis) for each urban income group. The indexes included the prices of the food products being examined, other agricultural goods, and nonagricultural goods, weighted according to their share in total consumption for each income group and normalized to add up to one. Most of the weights were obtained from the national cpi and, when the relevant database was available, varied by subperiod. Assuming that in the very short run there is no response in a household's nominal income to changes in food prices, the instan- taneous impact of a food price change on household real income for any given income class is measured by the percentage change in the cost of the consumption basket, defined as (8A-1) cP, - 1 CPI for the direct effects, and as (8A-2) =7 - 1 for the total price effects, where CPI' and CPI* represent the cost of the basket of consumer goods and services that would prevail in the absence of the direct and total price interventions, respectively. Nom- inal household income (NY) is spent on a basket of consumer goods and services whose cost is denoted by the corresponding CPI. Actual real income is defined as NY (8A-3) Y = CPI' real income in the absence of direct interventions is defined as NY (8A-4) YD C7 and real income in the absence of total interventions is defined as NY (8A-5) CPI* For this calculation, the consumer price index is disaggregated into the agricultural (A) and nonagricultural (NA) component of CPI: (8A-6) CPI = OPA + (1 - )PNA, where 0 is the share of agricultural products (mostly food) in total consumption expenditures, measured at retail prices, and P, = O j3PAj is an index of the various agricultural products entering the 174 Consequences of Price Intervention for Income Transfers and Equity cpi. The cpis depend on 0, which differs by income group, and thus CPI'ICPI and CPI*ICPI will vary according to income levels. The ci resulting from the removal of direct price interventions is expressed as (8A-7) CPI' = OPA' + (1 - )PNA, where OPA = 1PA. The cpi resulting from the removal of total i price interventions is expressed as (8A-8) CPI* = OPA* + (1 - O)PNlA, where OPA = Z jPJ*. For computational convenience, the direct effect for each income class was computed as the percentage change in the nominal cost of the consumer basket, as measured by -'CPI, (8A-9) D = CP - 1 = dlogCPI = ZOjNPRDj, CPI i which is roughly equivalent to the percentage change in household real income, as defined in equation 8A-1. In most of the country studies, the effects on urban income were estimated using equations 8A-1 and 8A-2, based on data from house- hold surveys. Exceptions were the studies for Colombia, the Domini- can Republic, and Zambia, which used equation 8A-9. To compute the total effect (YT), NPRT was used instead of NRPD, and the procedures used for computing the results reported in the table 8-1 do not adjust for P*A. Thus the results for YT using equations 8A-8 and 8A-9 could differ. Rural Households Because farmers generally produce more than one product, the impact of price changes on the real income of farmers was computed for each of the selected products. The unit of analysis is the farmer, not the individual product. The magnitudes of the income effects among the various farm categories vary mainly according to differ- ences in output mix, consumption patterns, and marketable surplus as a proportion of farm production. We start the analysis by assuming that farmers produce only one product. At t = to, output is given, so the impact of the interventions on nominal income (Y) of farmers (owners) can be measured by the effect on price, or by the effect on value added if purchased inputs are important. Effects of Intervention on Income Distribution and Poverty 175 Part of the output, especially in the case of food crops, is consumed on the farm, so marketed surplus may be smaller than output (Qs) and some farmers may be net buyers of the crop they produce (they need another source of income for that). The real income effect may thus be smaller than the nominal income effect and may even be in the opposite direction for some. Nominal income (Y) is the sum of farm income (Y) and off-farm income (YNF): (8A-10) Y = YF + YNF. Assume that YF is obtained from the production Q of product i, with price Pi: (8A-11) Y = Pi Q7. If purchased inputs are important, then farm income YF is (8A-12) Y = VA, Qs = (P, - i aF9Qt, where VAj = li - 1 aij is the value added of product i, and La; 14 is the value of purchased inputs per unit of output.12. Real income is defined as (8A-13) y = PI ZCi1 + (1 - Z)PNA Y + YF Zai li + (1 - ZCOPNA where ai is the share of product i in farmers' consumer price index P1, and PNA is the price index of the nonagricultural sector. If we assume that at t - to (the period when the price change occurs), the price policy has no impact on off-farm income, then YNF is given. The proportional change in real income (y) due to direct price inter- ventions is (8A-14) 9D where ( Y_ + YNF Y + YNF PI' ZaiPt + (1 - Za,)PNA and (8A-16) Y = PiQ7 or 176 Consequences of Price Intervention for Income Transfers and Equity (8A-17) Y = VAQ =-(P - Zaj,PJ)Q , where P,' is the price of product i and P is the price of input j in the absence of direct price interventions. In the case of total price interventions, PNA also changes, and the effect on real income y is: (8A-18) PT - y where . F+ YurY + Yr _F_NF YF ~NF (8A19) y PI* ZUiP I+ (1 - ei)PA and (8A-20) YF= PiQs or (8A-21) YF = VA"Qf = (P1- ZaI.P)Q , I where Pfis the price of product i and P*is the price of input j in the absence of all interventions,Y13 and PA is the price for the non- agricultural sector in the absence of all interventions. If we exclude from purchased inputs and from changes in the prices of products other than product i, and if (small-scale) farmers are net buyers of product i, it follows that the share of product i in consumption, ai, will be larger than the share of product i in income, Y1Y = (PQ5 /Y), so that the change in real income will be in the opposite direction of the change in P. The assumption that YNF does not change at t - to when agri- cultural prices change may not be realistic because some rural employment is related to the provision of goods and services con- sumed by farmers. If YNF changes at t - to to YL, or YN,, then this change should be incorporated into equations 8A-15 and 8A-19. If farmers produce several products, then farm income becomes: (8A-11) Y= ZPiQ'i or (8A-12) YF =Z VAi Q i, i Effects of Intervention on Income Distribution and Poverty 177 Then Y becomes (8A-16) Y = ZPiQ, or (8A-17) Y = ZVAiQ , and YFbecomes (8A-20) YF = i fQ or (8A-21) YF= Z VAfQs. For hired laborers on the farm, wages would not be expected to be affected at time t - to, so for them y changes only because of a change in the price index PI, and thus PI' (8A-22) YD and PI* (8A-23) PT - - 1. If hired laborers are paid in kind, their nominal income will vary with i, and we are back to equations 8A-16 and 8A-20. Note that the shares ai will probably differ for small-scale farmers, large-scale farmers, hired labor, by region, and so on. Appendix 8-2. Rural Wage Equations We,estimated equations for Brazil, Chile, and Colombia for the impact of intervention on the rural real wage, where both the nominal wage and the rural cpi were allowed to adjust. For Brazil, the estimated equation is W, = 1.54 + 0.11PAIPNA + 0.18PL + 0.19AGDP + 0.0PT (1.57) (1.42) (2.67) (2.98) (0.0) R2 = 0.96, D. W = 2.27, n = 17, where PL is the real land rental price, AGDP is agricultural GDP, and PT is the tractor rental price. 178 Consequences of Price Intervention for Income Transfers and Equity For Chile, the estimated equation is part of a system of simul- taneous equations that determine investment in various forms of cap- ital, migration, agricultural employment, acreage, output, the rural wage, and other variables. The specific rural wage function in the system of equations is W, = 0.19 + 0.99LD + 0.23A,-, - 0.11P - 0.69NR + 0.15Wu (-0.41) (1.48) (2.31) (-1.98) (-3.03) (4.3) R2 = 0.92, D. W = 2.12, where WR is the average real income per worker (real wage times hours of work), LD is the labor demand shifter in fruits, vineyards, and livestock production, A,-, is lagged acreage in annual crops (another labor demand shifter), PF is the price of fertilizer, NR is the rural labor force, and Wu is the nonagricultural wage index. Agricul- ture's terms of trade do not enter directly but they affect WR through their impact on LD and A,-1. Moreover, PF is affected by direct and indirect intervention, as is NR over time because of effects on migration. For Colombia, the estimated equation is derived from a model of labor demand and supply that incorporates a migration function. The regression results for Colombia are logW, = 32.8 + 0.54 logPAIPNA + 1.46logKA (4.0) (16.4) (13.1) + 0.27logWu - 0.94Un - 4.25logNR (6.26) (-2.8) (4.3) R2 = 0.995, D.W = 2.29, where W, is the rural real wage rate, PAPNA is agriculture's terms of trade, KA is the capital stock in agriculture, W. is the urban real wage rate (proxied by the wage in the construction sector), Un is the urban unemployment rate, and NR is the rural labor force. Appendix 8-3. The Sectoral Framework for Comparing Income Gains and Losses The starting point for the results of the analysis presented in table 8-4, was the imposition of some macroeconomic consistency on the data. Thus, the sum of the income gains of the various income brackets in each sector (urban and agriculture) was assumed to be equal to the corresponding direct transfers accruing to the sector, as obtained from the balance of direct income transfers (table 7-5). The implicit assump- tion here is that the income distribution effects of pricing policies are the outcome of a zero-sum game, where the sum of the net transfers Effects of Intervention on Income Distribution and Poverty 179 is zero. In a sense, this is an incorrect assumption, because there are dead-weight losses in the process; however, this assumption is better than estimating the income effects without imposing any bounds (as is done when the income effects are calculated through their impact on the cpi, leaving nominal income constant). Direct transfers captured by the nonagricultural sector were assumed to be fully captured by urban households, as consumers. These transfers were then distributed across income brackets accord- ing to each group's (absolute) level of spending on food. The expen- diture data for the various income groups were obtained from house- hold consumption surveys. Typically, such surveys were available for only one year during the 1960-84 period. It was assumed that the share of expenditures for food for each income strata remained con- stant throughout the period. To the extent that income distribution remained constant within the urban sector during these years, this assumption introduces a slight bias, since there has been some income growth and since the income elasticity of food is probably different from one and may vary with the level of income. Further- more, there is another source of bias in the case of Colombia and Egypt because the estimated income transfers include cotton, an item not directly consumed by urban households. Ideally, future work on this topic could disaggregate the data further and explicitly add the transfers to the processing industry. Consumption surveys also provide the thresholds that separate income quintiles. Those thresholds were extrapolated for all years (1960-84) using an index of nonagricultural per capita GDP (total GDP minus agricultural GDP divided by total urban population). Again, this assumes that the income distribution within the urban sector remains unchanged during the period under study. Once we had calculated the transfers accruing to each quintile in the urban sector and the thresholds' points for each year, we allocated the transfers within agriculture using comparable threshold levels. Land distribution surveys were available during 1960-1984 for all the countries undtr study for one to three years. Basically, the surveys provided information about the number of agricultural holdings in each farm size. So for each country and for specific years, information was available about the number of holdings of a certain size category and on the corresponding percentage of total land accounted for by that category. Information for the remaining years was obtained by linear interpolation. For Egypt and Chile, which experienced massive land reforms, the interpolation presents some problems for the "tran- sition years," in the sense that the change in land distribution might have been less smooth than assumed here. So that the data on land distribution could be used to make infer- 180 Consequences of Price Intervention for Income Transfers and Equity ences about income distribution, the following crucial assumptions were made: * One "owner" household is associated to each holding. * The labor force per household is constant across households. * The amount of capital per holding is appropriately indexed by the amount of land it occupies. * The output mix is the same across different holdings, and all holdings use the same technology. * All agricultural labor is paid the marginal value of production. * Removing direct interventions does not change the functional distribution of income within agriculture. The first two assumptions are rather simple benchmarks, and their main role was to simplify the estimation. Although the assumptions are probably false for bigger holdings because of the skewed distribu- tion of land and the broad income categories used in the study, the impact of these assumptions on the final results are probably negligible. The third assumption on capital may seem reasonable, but the way in which capital is determined depends crucially on the fourth assumption on output mix and technology, which is clearly incorrect. Very small holdings tend to produce a high share of food crops, and, because of transaction costs and possible imperfections in the capital market, they tend to apply more labor-intensive techniques of pro- duction. The main implication of the first five assumptions is that the ratio of land to labor is equalized across holdings of various sizes. It is important to stress that if the production mix and the technology are different, the direction of the bias is unclear. That seems to be the case in some instances. The main problem with the first five assumptions is that they rule out the existence of a pure agricultural labor force that does not own any land. In that case, however, the bias is predictable: the number of poor people will be underestimated-and so will their share in the gains or losses. Finally, the last assumption was intended simply as a useful benchmark, motivated largely by the econometric results obtained for Chile (Hurtado, Vald6s, and Muchnik 1990). The methodology can be suitably adjusted to different assumptions derived from prior estimations of aggregate production functions. The main implication of the first five assumptions is that the ratio of labor and land will be equalized across holdings. Where 0 is the share of labor in total agricultural value added, Ni is the number of holdings in category size i, and Li is the corresponding amount of land. The share of total agricultural income ai, going to households in the ith category, was calculated as where (8A-24) ai = Ni0 + (1 - 0). Effects of Intervention on Income Distribution and Poverty 181 Labor's shares in total agricultural value added were obtained with different degrees of accuracy for each country. For Chile the data come from the national accounts, while for Egypt and Colombia the data were estimated using the techniques employed in the different activities presented by the country authors in their imputation of the effective rates of protection. The average level of household income in each category, Y1, was simply estimated as (8A-25) Yi = YAai( , where YA is agricultural GDP. Having obtained the average income levels for each size category, the remaining problem was to make inferences about the boundaries (thresholds) in terms of size of holdings that was associated with the corresponding thresholds obtained for the urban sector. To this end it was assumed that the mean income level within each category was obtained at the mean holding size. Linear interpolation between those mean levels gave the corresponding threshold points. The out- come was distribution of agricultural income into categories compara- ble to those for the urban sector. The gains and losses due to direct interventions were distributed among categories using the shares of each of these categories (implicitly using the final assumption above). Some Remarks * The case of Thailand was estimated in a different way. The coun- try study for Thailand showed very detailed matrix of income transfers for 1965. These data were extrapolated to the other years, assuming that the share of each income category was con- stant over time. * For the urban sector, the income declared in the consumption surveys was used, while for the agricultural sector, the income measure came from national accounts. It is almost a stylized fact that the income derived from consumption surveys does not match the income estimated from national accounts. Unfor- tunately, the direction of the bias is far from clear. * This methodology provides a rough initial benchmark for the income distribution effects. Future adjustments might include a different assumption with respect to aggregate production func- tions or the working of the labor market. Policy statements that assume that price incentives for agriculture are regressive tend also to assume (implicitly) that nominal wages in the agricultural and nonagricultural sector are rigid. Here, the functional income distribution was assumed to be constant instead. Intermediate assumptions are also clearly possible within this framework. ���b� ��-�� и � н. �' �• � А � и А, е�.�, и А� н о .� An Empirical Analysis of Patterns of Intervention In Part I of this volume we examined the historical patterns of direct and indirect agricultural price interventions. In this chapter we look at several alternative models of agricultural price behavior and then develop a model of direct interventions that best explains these pat- terns in a variety of situations in both the short and long run. Whereas the model was specified with a dynamic structure common to all countries and periods, its structure allows the parameters to vary according to particular structural variables in each country and period. The model is able to replicate our results on average rates of protection (chapter 2) and price stabilization (chapter 3) and to explain most cases encountered in the study. The model predicts that in the short run direct price interventions in agricultural exportables will result in an average reduction in the variability of domestic prices relative to world prices of about one- third-about the same decrease in variability found with the standard deviation and annual price variation measures used in chapter 3. The reduction in variability was found to be directly related to the degree of exchange rate overvaluation, itself the result of macroeconomic industrial protection policies (according to the concept of exchange rate misalignment described in chapter 2). In other words, countries that were most successful in achieving short-term price stabilization also followed macroeconomic and trade policies that resulted in the highest degree of overvaluation of the real exchange rate. A plausible explanation for this relationship might well be called an "addiction" to intervention: widespread import controls, usually through quan- titative restrictions, tend to go hand in hand with controls in other markets, including agricultural markets. The model predicts that in the long run the rate of direct taxation of agricultural exports will be positively related to agriculture's share in a country's total exports, a finding consistent with our results on the fiscal impact of price interventions (chapter 6). When most exports come from the agricultural sector, direct taxation of agricultural 185 186 Determinants of Agricultural Taxation; Conclusions exports will tend to average slightly more than 25 percent in the long run; when most exports come from other sectors, agricultural exports will tend to be subsidized at a rate of about 10 percent. As countries develop, agriculture's share in exports tends to fall, and as govern- ments find new sources of revenue and establish a wider menu of tax instruments, the need to tax agriculture diminishes. For importables, the rate of direct protection was found to be pos- itively associated in the long run with agriculture's share in total imports. The higher the share of food in a country's total import bill, the stronger the pressure to achieve a degree of self-sufficiency, and the higher the rate of protection of agricultural importables. The model also shows that adjustment to the long-run "target" rate of protection or taxation of agricultural tradables after a change in world prices was slower the larger agriculture's share in gross domes- tic product (GDP)-and faster the smaller the share. When a change in world prices causes a deviation from the long-term target rate of protection or taxation of tradables, lower-income countries, with a larger share of agriculture in GDP, tend to return more slowly to the target level than higher-income countries, possibly because of higher adjustment costs. Motives for Intervention and Alternative Models of Price Behavior Governments intervene in the prices of agricultural products for a variety of reasons. They may tax export crops explicitly to generate revenue for expenditures in other sectors, as in the case of rubber in Malaysia and Thailand. Or the export tax may be implicit, adminis- tered by a state marketing board that pays farmers a low price for their products and exports them at a higher price. This implicit tax- the gain over and above "reasonable" marketing and distribution costs-is an alternate way of generating government revenue; the Cocoa Marketing Board in Ghana is an example. If exportables are food products, an export tax achieves the double objective of generat- ing revenue and reducing food prices to consumers, as in the case of wheat and beef in Argentina and rice in Thailand. We found that many countries protect the production of import- competing food products through direct interventions (chapter 2), probably in order to increase self-sufficiency. The rate of protection rises with the share of agriculture in total imports, probably because higher imports imply a greater drain on foreign exchange. Protection rises to boost domestic output, reduce demand, and so to reduce imports. Revenue may be another motive for protecting importables and may also explain why protection increases with the share of food in total imports. Revenue can be generated through an import tariff, An Empirical Analysis of Patterns of Intervention 187 as in the case of wheat in Chile, or through domestic sales at higher than the import price by a state marketing agency with a monopoly on food imports. But a dual pricing policy, with higher prices for producers than for consumers, may create severe fiscal pressures that are unsustainable in the long run. A few countries subsidize food consumption through import sub- sidies, as Egypt and Pakistan do for wheat, and the effect on the budget depends on the share of food aid in total food imports and on the coverage of the subsidy program. The average rate of direct protection is larger-and positive-in developed countries than in developing countries, where it is nega- tive on average.' It has also been found to increase as a country develops. These differences across countries and over time have been explained in many different ways, but most hypotheses have been implicitly or explicitly based on the underlying assumption that gov- ernments (political leaders, ministries, bureaucracies) intervene to further their own interests-especially to increase their chances of staying in power (Downs 1957). According to this theory, interventions are the equilibrium out- comes of the interaction of demand and supply. Interest groups will lobby for an increase in interventions that benefit them-whether through an increase in the price of the products they sell or a reduc- tion in the price of products they buy-or for a reduction in interven- tions that harm them. The more an interest group stands to gain from an intervention, the larger its demand as expressed by the amount of resources spent to affect policy. And the more a group stands to lose from an intervention, the more resources it will devote to resisting it, the higher the cost of supplying the intervention, and therefore the lower its supply. As an economy develops, agriculture typically tends to fall as a share of GDP, employment, and consumption expenditures, and rural wages tend to lag behind urban wages. Demand for intervention in agricultural prices increases because rural income tends to fall further behind urban income (at least absolutely, if not always relatively) and because the costs of organizing (including the free rider problem) fall as the agricultural sector shrinks. The "supply" of interventions also increases (or the resistance to intervention falls) as the economy grows, because food constitutes a smaller share of consumer expendi- tures and because the need to tax agriculture diminishes as other sectors grow and become more important sources of fiscal revenue. Consequently, agricultural protection would be expected to increase as an economy develops. A few studies have attempted to model the behavioral relationship between the degree of agricultural price intervention and its principal determinants. Using five-year interval data for 1955-80 for fifteen 188 Determinants of Agricultural Taxation; Conclusions industrial and industrializing (Korea and Taiwan) countries, Honma and Hayami (1986) estimated an equation of the determinants of vari- ation in the nominal protection coefficient. They found that nominal protection rises as agriculture's comparative advantage falls,2 as the share of agriculture in GDP (or in the labor force) falls, and as the international terms of trade between agricultural and manufacturing fall. In a study of the wheat market in Egypt from 1949 to 1979, Scobie (1981) found that domestic consumer and producer prices rose as the world wheat price rose and fell as import capacity (opening reserves plus receipts of foreign exchange) rose. They also found that an increase in wheat aid shipments led to a fall in the producer price (relative to the consumer price), and that an increase in food price inflation resulted in higher producer prices for wheat (to expand food supplies). In a study of the beef market in Brazil from 1947 to 1971, Lattimore and Schuh (1976) found that the level of the export tax was directly related to the rate of inflation (to reduce the price of beef), to the balance of payments position (with higher foreign exchange reserves, the government is willing to trade off some beef exports to achieve lower domestic prices), and to the international price of beef (to achieve price stabilization). We tried several alternative models before selecting our final model. The main structural variables we considered were measures of comparative advantage and of the relative size of the agricultural sector (both variables are positively correlated in our sample). We expected a negative relation between those variables and the level of protection (as found by Honma and Hayami 1986). We also checked for the presence of any compensation pattern in the data, in the sense of a negative relation between direct and indirect intervention (or between direct intervention and exchange rate misalignment) or between direct intervention and nonprice transfers (government investments in agriculture). We also examined the degree of price stabilization. When the domestic price was used as the dependent variable, most explanatory variables were significant, but residuals showed a high degree of autocorrelation in most versions, suggesting problems in the dynamic specification of the model. When the direct rate of inter- vention was used as a dependent variable, the autocorrelation prob- lem was reduced, but confidence in the results was lowered by hav- ing the world price on both sides of the equation (possibly causing spurious correlation). Also, in this form, the lagged domestic and international prices became significant, again raising doubts about the dynamic specification. (These problems are addressed later in this chapter). And finally, in most regressions for agricultural exportables An Empirical Analysis of Patterns of Intervention 189 or importables, we found that the share of agricultural exports in total exports or agricultural imports in total imports became significant. Honma and Hayami (1986) do not present information on autocor- relation of residuals, and their use of data with five-year intervals suggests that their results reflect long-run rather than short-run price intervention behavior. Scobie (1981) corrected for autocorrelation, suggesting that there may have been dynamic forces at play that were not captured explicitly. Our model distinguishes between short- and long-run price behavior and captures the speed of adjustment to the long-run equilibrium. A Model of Agricultural Price Intervention We developed a dynamic error-correction model of government price intervention. The same model was used for -all countries (and years) studied, but parameters varied to reflect the particular structural or policy variables for each country (and year). This method, which uses endogenized parameters, is known as the variable coefficients approach. We estimated separate models for exportables and import- ables. Only the exportables model is discussed in this chapter, but the model for importables is similar. Let (9-1) Pt = log (Px1PNA)t and (9-2) Pt'= log (PIPNA)t where P is the producer price of exportables, PNA is the price index of nonagricultural goods and services, and P4 is the producer price of exportables in the absence of direct interventions. We assume that policymakers determine P, given P, according to the following decision rule: (9-3) APt = a + OAP' + b(Pt-1 - yP1-). where AX, = X, - Xe, for X = P or P'. Three distinct sets of forces affect AP in any period. One is the short-run transmission from changes in international prices (AP') to changes in domestic prices (AP,). We expect 0 to be positive, and assuming that short-run price stabilization is an objective of policy intervention (as shown in chapter 3), we expect it to be smaller than 1 (0 < 0 < 1). The second force is the long-run target protection rate, that is, the long-run relationship that policymakers want to maintain between the domestic price P and the foreign price P' so that P = -yP'. If y = 1, then P = P' in the long run and there is no intervention. Since 190 Determinants of Agricultural Taxation; Conclusions exportables were taxed on average in our sample countries and importables were protected, we would expect -y to be smaller than 1 for exportables and larger than 1 for importables. And third, whenever P_1 is above its long-run target (,yP_z), we expect a downward adjustment in the price at time t-or an upward adjustment when it is below its target-in order to reduce the gap between the domestic price and its long-run target. In other words, we expect the coefficient 6 to be negative. The faster the adjustment to the long-run target price yP', the larger (in absolute value) the coeffi- cient 6. Assuming no overshooting of the long-run target price, we would expect 6 to be smaller than 1 in absolute value (- 1 < 6 < 0). Thus, 0 measures the degree of short-run transmission from world prices to domestic prices, -y measures the long-run target protection rate, and 6 measures the speed of adjustment to the long-run target price. Finally, in a steady-state situation where P = yP, and AP' = 0, we would also expect AP, to be equal to zero, which implies that a = 0. We will check whether the model satisfies that restriction. Thus, the proportional change in the domestic price3 in any period t (APt) is the result of a short-run transmission from the change in the international price (AP,) and of an error-correction term from the previous period, where the "error" consists of the deviation of the price from its long-run target. Next we include country-specific economic variables affecting the three parameters 3, y, and 6. We postulate that the size of the export- ables sector in agriculture,4 as measured by the share of agricultural exports in total exports (SHAXX), is negatively related to the long- term rate of direct protection (or is positively related to the long-term rate of direct taxation). As explained before, as a country develops, the share of agriculture in total exports tends to fall, and other sectors increase in size, reducing the need to tax agricultural exports. Other forces also come into play that suggest the tax rate will fall as the sector shrinks. One such force is the increase in the sector's supply elasticity, since it faces more elastic factor supplies. To achieve a desired level of tax revenue while minimizing efficiency costs, the tax rate should be lower for sectors with high supply elasticity. Also, as the size of the sector falls, it becomes cheaper for producers to orga- nize and to lobby the authorities for higher prices. Thus we have (9-4) -y = 0 + 7,SHAXX, yo > 0, yl < 0, y > 0. Second, we postulate that moving the domestic producer price toward its long-term target price involves costs (in terms of resource reallocation, changes in income and wealth distribution) and that these costs are directly related to the size of the sector. We use the share of agricultural GDP in total GDP (SHAYY) as the relevant variable An Empirical Analysis of Patterns of Intervention 191 affecting adjustment costs. The higher the share, the lower the speed of adjustment. Thus we have (9-5) 6 = 60(l + 61SHAYY), 6 < 0, 61 < 0, 6 < 0, implying that 1 + 61SHAYY > 0. Third, we postulate that the degree of short-run transmission of changes in international prices to domestic prices is inversely related to the degree of government control in the economy. If interventions are extensive (including trade interventions), we may expect macro- economic policy to be less responsive to international shocks, and domestic agricultural prices to be less responsive to foreign prices. As a measure of the general degree of intervention, we use the degree of exchange rate distortion. Recall that our measure of exchange rate distortion includes the effects of both industrial import controls and macroeconomic policies. Thus we have (9-6) 0 =0o 1 +0 *_ 1 ], j3o > 0, 01 < 0, 0 > 0, implying that 1 + 01(E*/Eo -1) > 0, where E* is the equilibrium exchange rate, Eo is the official exchange rate, and E*/Eo - 1 is the degree of exchange rate distortion. Then, from equations 9-3 to 9-6, we have (9-7) AP, = a + o1 + - 1 AP' + 60(1 + 61 SHAYY)[P,_, - ('Vo + 7y,SHAXX)P'-1] + et, with e = 0, 0o > 0, 01 < 0, )3 > 0, 6o < 0, 61 < 0, 6 < 0, yoa> 0, 'yl < 0, 'V > 0. Econometric Results The results from the estimation of equation 9-7, based on 297 obser- vations, are as follows: Parameters: a 0o 1 60 61 To 71 Estimates: -0.02 0.72 -0.18 -0.34 -1.17 1.02 -0.08 t-values: -0.4 18.4 -4.7 -4.3 -2.3 23.3 -2.5 with corrected R2 = 0.57, and D.W = 2.0. All parameters of interest have the expected sign and are significant at the 5 percent level or better. The constant term a is not significantly different from 0, so the model is consistent with stable domestic prices in steady-state. The model's degree of fitness (R2 = 0.57) is extremely high, considering that the equation was estimated in the first difference rather than in the level of the variable; that is, it was estimated for annual changes in the domestic price level and not for 192 Determinants of Agricultural Taxation; Conclusions the price level itself. The problem of autocorrelated residuals is also resolved (D.W = 2.0), suggesting that the dynamic specification is satisfactory. When the share of agriculture in GDP (SHAYY) is used as a variable in the equation for -y (instead of agriculture's share in exports, SHAXX), a becomes significantly different from 0, implying that con- tinuous domestic price changes in a steady state, and thus contradict- ing the notion of steady state equilibrium. (The same result occurs when the regression is estimated in terms of prices rather than the logarithm of the prices.) When the exchange rate distortion variable (E*/Eo - 1) is used in the error correction term (,y equation) to see if it affects long-run direct protection, the variable is not significant-the hypothesis is strongly rejected by the data. Thus the choice of func- tional form and variables used to endogenize the parameters of the model is far from arbitrary. We suspected that these variables (agriculture's share in GDP and in total exports and exchange rate distortion) were important in the determination of the domestic price for exportables (P,), but the pre- cise channel through which each variable affects that price was revealed only after estimating the model. The results have plausible economic and political- economy interpretations. Implications of Empirical Results The empirical results strongly support the use of the model to explain the determination of the prices of agricultural exportables in our sam- ple of eighteen developing countries. The results show a negative sign for the coefficients -y, 61, and 01, indicating a negative relation between the protection rate and the share of agriculture in total exports, between the share of agriculture in GDP and the speed of adjustment to the long-run target price, and between exchange rate distortion and short-run transmission of price changes from the inter- national to the domestic market. The Average or Typical Case We first examine the implications of our model for the average case or the typical country in our sample. We start by replacing the coeffi- cients in equation 9-7 with their estimated values: (9-8) AP, = 0.72 1 - 0.18(L - 1) AP - 0.34(1 - 1.17SHAYY) [Pt-1 - (1.02 - 0.08SHAXX) Pr'-] + Ut, where U, is the residual, and the coefficient a has been omitted since it is not significant. An Empirical Analysis of Patterns of Intervention 193 The average share of agriculture in total exports (SHAXX) is 52 percent (rather than 54.4 percent, as shown in table 9-1, which does not take into account variations across countries in the length of the period covered). Plugging in this value yields the following long-run relationship between P and P': (9-9) P = (1.02 - 0.08)0.52P' = 0.9784P'. Since P and P' are logarithms of relative prices, we have P, P, 0.9784 (9-10) . The average level of the price index P,/PNA is about 130, implying an average level for PxPNA of about 117 and an average long-run rate of direct taxation of exports of 10 percent. This value is very close to the actual average tax rate (12.6 percent, as shown in table 2-2), which provides further support for the model. Note that the long-term tax rate varies directly with the level of the international price PIlPNA Table 9-1. Determinants of Rate of Taxation of Agricultural Exports Agriculture's Exchange rate Agriculture's share in overvaluation share in GDP exports Country Period (E*/Eo -1) (SHAYY) (SHAXX) Argentina 1960-84 21.5 12.6 57.6 Brazil 1969-83 14.7 12.4 62.3 Chile 1960-83 21.3 7.5 4.5 Colombia 1970-83 25.9 24.6 60.0 C6te d'Ivoire 1960-82 42.0 61.5 92.6 Dominican Republic 1966-85 24.7 23.4 72.9 Egypt 1964-84 21.0 24.6 60.5 Ghana 1958-76 61.5 44.4 64.1 Malaysia 1960-83 7.9 22.4 40.4 Morocco 1963-84 26.5 20.0 33.2 Pakistan 1969-86 36.7 32.8 91.8 Philippines 1962-82 23.7 14.9 47.9 Portugal 1960-81 2.3 13.5 31.7 Sri Lanka 1960-81 18.1 28.8 83.1 Thailand 1962-84 19.1 30.1 - Turkey 1961-83 44.7 27.5 66.7 Zambia 1970-83 128.2 13.7 0.8 Averagea 1958-85 32.6 24.3 54.4 - indicates that data are not available. a. Excludes Thailand because of the lack of data on agriculture's share in exports. Source: Authors' calculations based on data provided in country studies. 194 Determinants of Agricultural Taxation; Conclusions because the prices are in logarithms,5 which implies, on average, some degree of stabilization over the long run, although extremely little. On average, the share of agriculture in GDP (SHAYY) is about 24 percent (table 9-1), so from equation 9-8, the coefficient of adjustment 6 is -0.24 (-0.34 [1-1.17SHAYYI = -0.34 [1-1.17 x 0.243 = -0.24). Thus, on average the price at time t is corrected for about one-fourth of its deviation from its target value at t - 1. From equation 9-8 we see that in the absence of any exchange rate misalignment (E*/E0 -1= 0), 72 percent of the variation in the border price of exports would be transmitted to the domestic producer price in the short run. With an overvalued exchange rate (E* greater than Eo), however, the coefficient of short-run price transmission (3) would fall and the degree of domestic price stabilization would rise accord- ingly. Thus, for the average case of an exchange rate overvaluation of about 32.6 percent, 0 falls from 72 percent to 67.8 percent (0.72 [1-0.18 x 0.3266]), implying somewhat greater price stabilization (from a degree of stabilization of 28 to 32.2 percent). This result is very close to the results presented in chapter 3 on the impact of direct intervention on the price variability of exportables. On average, we found that the standard deviation of P,JPNA is 0.68 of its value for P./PN,, the coefficient of variation of PIPNA is 0.82 of its value for P.'PNA, and the annual price change statistic of PIPNA is 0.67 of its value for PI/PNA (table 3-1). Once again, our results lend strong support to the behavioral model suggested here. Individual Cases In our model, the long-run rate of direct taxation is positively related to the share of agriculture in total exports-the higher the share, the higher the tax rate, and the lower the share, the lower the tax rate. This is confirmed by our results for most countries. In C6te d'Ivoire, agriculture's exports averaged about 93 percent of total exports (see table 9-1) and the rate of direct taxation averaged 28.7 percent (see table 2-2). For other countries in which agricultural exports were sub- stantial, the direct tax was also high in most cases. Thus, the corre- sponding figures were 73 percent (agriculture's share) and 24.8 per- cent (direct tax) in the Dominican Republic, 64 percent and 29.8 percent in Ghana, 61 percent and 25.6 percent in Egypt, and 58 per- cent and 17.8 percent in Argentina. Similarly, countries in which agriculture's share in total exports was smaller than average tended to have a lower rate of taxation. In Malaysia, agriculture's share was 40 percent and the direct tax was 12.7 percent, and in the Philippines, the share was 48 percent and the An Empirical Analysis of Patterns of Intervention 195 tax was 11.2 percent. In Zambia, where agriculture's share was a very small 0.8 percent, the rate of direct taxation was only 3.1 percent, and in Chile, where the share was 4.5 percent, agricultural exportables were protected on average at about 13.5 percent. For a few countries, the results are unexpected. Some of these results can be explained at least partly by our selection of only a few products in each country rather than all agricultural exportables. For instance, in Brazil, agriculture's share in exports is 62 percent, yet our results indicate that exportables received direct protection of 5.4 per- cent. Coffee, which is heavily taxed and constitutes an important share of agricultural exports, is not included in the product sample, however. Similarly, in Turkey and to a lessor extent in Portugal, agri- culture's share is high but exportables are protected, on average, while in Colombia and Pakistan, one would have expected a higher tax rate, given the high share of agriculture in total exports. The speed of adjustment to the long-run target price depends on the value of the coefficient 6 in equation 9-8, which is negatively related to the share of agriculture in GDP. Countries with a high share will adjust more slowly than countries with a low share. C6te d'Ivoire, where agriculture's share in GDP is 61.5 percent (table 9-1), has an adjustment coefficient value of about 0.1, whereas Chile, where the share is 7.5 percent, has a coefficient of about 0.3. This means that when the domestic price has deviated from its long-run equilibrium value (given a constant international price), C6te d'Ivoire has returned to equilibrium much more slowly than Chile. Other countries where adjustment is rapid are Argentina, Brazil, the Philip- pines, Portugal, and Zambia. Countries with a much slower rate of adjustment are Ghana, Pakistan, and Thailand. How much of the change in border prices is transmitted to domes- tic prices in the short run depends on the value of coefficient 0 in equation 9-8, which is negatively related to the degree of exchange rate overvaluation. The more overvalued the exchange rate, the lower the degree of transmission of world price changes to domestic prices in the short run and the greater the stability of domestic prices. For instance, C6te d'Ivoire, Ghana, Turkey, Pakistan, and Zambia, had a high average degree of overvaluation (table 9-1), which implies less transmission and more price stability. This finding is confirmed by our results on price variability (table 3-1) for C6te d'Ivoire, Ghana, Pakistan, and Turkey, where export price variability was below the average, but not for Zambia. However, the change in domestic prices (AP) depends not only on the change in world prices (AP[) but also on the error-correction term, which may explain the result in Zambia. Portugal and Malaysia, with the lowest rates of exchange rate over- valuation, show little reduction in export price variability (table 3-1). 196 Determinants of Agricultural Taxation; Conclusions General Patterns of Price Behavior Based on the country results for the two structural variables (share of agriculture in total exports and in GDP) and for exchange rate distor- tion, we classified countries according to four patterns of general price behavior. Category 1 includes countries with very high shares of agriculture in total exports (SHAXX) and in GDP (SHAYY) and with a high degree of exchange rate overvaluation. That implies low values for 7 (imply- ing a high rate of long-run taxation), for 0' (the short-run price trans- mission coefficient), and for 6 (the speed of adjustment to the long- run target price). C6te d'Ivoire and Ghana, with high tax rates, fit this category. We can see what this pattern means for price behavior by looking at what happens when there is a change in world prices. Starting from the long-run equilibrium where P = -yP', we assume a one-time increase or an increase over several years in the world price P'. Since the exchange rate is highly overvalued, the short-run change in the domestic price is slight, and so the tax rate increases. But since agri- culture's share in GDP is high, adjustment to the long-run target price is very slow, and the tax rate may remain above the long-run target (and the price below it) for a long time. Countries that fit this pattern-stabilizing the domestic price and changing it only slowly- may miss opportunities for export expansion that arise with changes on the world market. And when the world price falls dramatically and domestic prices remain stable, a tax may turn into a subsidy, as in C6te d'Ivoire in recent years, and cause serious fiscal problems. Category 2, at the opposite extreme, includes countries with low shares of agriculture in total exports and in GDP and little exchange rate distortion. A country with these characteristics would have a low long-run rate of direct taxation of exports, if any. And since the trans- mission of world price changes would be high, the domestic price would not deviate from its long-run target for long. This is essentially the case for countries with a free trade regime and some stabilization around the long-run target (but not much). Chile, which in recent years has kept domestic prices within a price band around the border price, fits this pattern. Category 3, an intermediate case, includes countries with a large share of agriculture in exports and a somewhat smaller share in GDP and little exchange rate overvaluation. This pattern implies a high rate of taxation in the long run, but also a high rate of transmission of world price changes to domestic prices in the short run, so that depar- tures from the long-run target would be slight and transitory. Malay- sia, Sri Lanka, and possibly Thailand fit this pattern. An Empirical Analysis of Patterns of Intervention 197 Category 4, another intermediate case, includes countries in which both agriculture's share in total exports and exchange rate overvalua- tion are high, but agriculture's share in GDP is low. Under these conditions, the long-run rate of direct taxation will be high and adjustment to the long-run rate will be swift, but the tax rate may fluctuate considerably in the short run in order to stabilize domestic prices. This description applies to Argentina, where Sturzenegger (1990) found that interest groups significantly increase their pressure whenever they lose relatively, but not (or not by as much) when they fail to gain. Thus, policymakers tend to pursue policies that maintain the status quo in the long run ("the conservative regulator") but dampen world price variability in the short run. This finding is con- sistent with our model, which predicts that deviations from the long- run target will not last (fast return to the status quo), but that the tax will fluctuate significantly in the short run. Our model, then, can be used to classify countries according to several types of direct price policies, with each case depending on the level of the structural variables (agriculture's share in total exports and in GDP), on the real exchange rate misalignment (resulting from macroeconomic and industrial trade policies), and on the behavior of border prices. Policy Reform The model can explain the evolution of price intervention as a func- tion of changes in the structural and policy variables and changes in border prices. But what can it tell us about reform efforts? The model indicates a relationship between exchange rate policy (and the trade and macroeconomic policies behind it) and direct price interventions affecting exportables. If monetary and fiscal policies are unsustainable, a macroeconomic reform that includes exchange rate adjustment will result in an increase in the short-run transmission of world price changes to the domestic prices of agricultural exportables. Although the model shows no direct relation between macro- economic reform and the long-run rate of direct taxation,6 there may be an indirect relation. If the real exchange rate is so overvalued that resources flow from the unprotected tradable sectors (mainly agricul- ture) to the nontradable and protected tradable sectors, agriculture's share in both GDP and total exports may fall. Macroeconomic reform will lead to increased short-run transmission of world price changes, as described previously, but it will also be accompanied by a fall in the long-term tax rate because of agriculture's reduced share in exports caused by the extreme indirect taxation. That occurred in Ghana in the early 1980s, with a reform taking place in about 1985. 198 Determinants of Agricultural Taxation; Conclusions If estimated values are substituted for the coefficients Tyo and yj in equation 9.4 for the long-term target rate of protection (so that Y = 1.02 - 0.08SHAXX), we see that when agriculture's share in total exports (SHAXX) is 25 percent, -y = 1 and there is no long-run inter- vention. But when agriculture's share is above 25 percent, exports are taxed, and when it is below 25 percent, exports are protected. Conse- quently, a reform occurs when agriculture's share falls below 25 per- cent. This reform may occur suddenly if the world price of agri- cultural exports falls dramatically or if new exports, such as oil, appear. This analysis is based on the point estimates of -yo = 1.02 and -yj 0.08, but their "true" values may be different. For instance, 'Yo is not significantly different from 1, and if we assume that it is equal to 1, then exportables are taxed as long as agriculture's share in total exports is positive. When the share reaches 0, there is no interven- tion, and when the share is negative, the exportables have become importables and are protected (this simulation is outside the sample values). There have, in fact, been cases where a good was protected when it was an importable and then was taxed as it began to be exported, such as rice in Colombia. We used a separate model for importables, however, the main findings are presented at the begin- ning of the chapter. Developing a model general enough to explain the domestic price behavior of both importables and exportables is a topic for future research. Summary and Concluding Remarks Governments in developing countries have strongly intervened in their economies, including the agricultural sector. As this study shows, for most of the past three decades, government policies regarding agriculture have adversely affected agricultural incentives, output, and farm income. The bulk of discrimination against agriculture, as is shown in the study, comes from indirect price interventions, which are rooted in the national development strategies of industrialization based on import substitution. The taxation of agricultural production was ratio- nalized by several economic arguments. One stated reason was based on the belief that agriculture was impervious to price incentives, so that taxing it would sacrifice little in terms of output. Moreover, taxing agriculture was an easy policy to administer and attractive in countries with a thin tax base. In addition, shifting scarce resources from agriculture to industry was thought to be justified by the pre- sumed secular decline in the terms of trade of agricultural exports relative to industrial imports. Furthermore, various arguments were advanced implying that intervention against agriculture was con- sistent with the welfare of the poor. By the 1980s this conventional wisdom was beginning to look less wise. The belief that incentives do matter had been gaining acceptance, but the conventional wisdom had never been put to a rigorous test. The comparative study of agricultural price interventions in eighteen developing countries reported here examines the effects of sectoral and economywide policies on the level and structure of incentives for agri- culture, and presents a quantitative assessment of the consequences of these policies on agricultural output, trade, consumption, government revenue, and the income of the urban and rural poor. The main findings, most of which do not support much of the conventional wisdom, are presented below: * The indirect tax on agriculture from industrial protection and macroeconomic policies was about 22 percent on average for the 199 200 Determinants of Agricultural Taxation; Conclusions eighteen countries over 1960-85-nearly three times the direct tax from agricultural pricing policies (about 8 percent). The total tax (direct plus indirect) was thus 30 percent. * Industrial protection policies taxed agriculture more than did real overvaluation of the exchange rate. * High taxation of agriculture was associated with low growth in agriculture-and low growth in the economy. * The transfers out of agriculture have been enormous. The net effect of total (direct plus indirect) interventions averaged 46 per- cent of agricultural gross domestic product (GDP) from 1960 to 1984. These transfers ranged from 2 percent for the protectors (Korea and Portugal) to 140 percent of actual agricultural GDP in Sub-Sahara African countries. Such enormous transfers must have severely depressed private investment and growth in agriculture. * Surprisingly, most countries protected importables. On average, the direct protection of importables was about 18 percent and the direct taxation of exportables about 16 percent, for an average impact on the relative price of importables to exportables of about 40 percent. These distortions within agriculture increased between the early 1960s and mid-1980s. * Direct price policies stabilized domestic agricultural prices rela- tive to world prices, with an average reduction in variability of 25 percent and even more when world prices were highly volatile. Indirect policies contributed little, if anything, to price stability. * Public investment in agriculture did not compensate for adverse price policies. * The effect of removing agricultural price interventions is not regressive. In most countries, removing direct (or total) interven- tions changed the real incomes of the poorer urban and rural groups by less than 5 percent (up or down). More often than not, the rural poor gained from the removal of the interventions. * The contribution of agriculture to fiscal revenues has fallen over time and is on average small. The Average Tax on Agriculture Was Huge Governments influence agricultural prices both directly, through agricultural sector policies, and indirectly, through industrial protec- tion and macroeconomic policies that tax agriculture relative to trad- ables and nontradables outside the agricultural sector. Indirect inter- ventions depress the prices of agricultural tradables relative to nontradables (through their impact on the real exchange rate) and relative to other tradables (due to industrial protection). These poli- Summary and Concluding Remarks 201 cies affect production incentives by making agriculture less attractive than other sectors of the economy. The effects of indirect interven- tions have until recently been outside the theoretical and political debate on agricultural price interventions in developing countries. What have been the effects of such direct and indirect interventions in eighteen developing countries over twenty-five or so years? On average, they taxed agricultural producers by about 30 percent (table 10-1). The indirect effects (22 percent tax) were by far the stronger, easily outweighing even the positive influence of direct interventions in the few cases where direct measures protected agriculture. Macro- economic policies caused the appreciation of the real exchange rate, raised the relative cost of nontradable inputs, and reduced the real purchasing power of income received from the sales of export and import-competing commodities. And protection for domestic indus- try hurt agriculture by raising the domestic price of importable agri- cultural inputs above world prices, reducing the purchasing power of farm households as consumers of manufactured goods, and by caus- ing further appreciation of the real exchange rate. Such intervention usually reduced agriculture's share of gross national product and was often related to slower growth in agricultural production and agri- cultural exports and to slower economic growth overall. Many direct measures have also taxed agriculture-on average, by 8 percent-depressing the prices received by agricultural producers. Governments intervened directly through procurement measures (government marketing boards are often the only legal buyers of agricultural outputs), quotas on exports of food crops and other agri- cultural commodities, and direct taxation of such exports. But some direct measures benefited agricultural producers. Governments often subsidized the costs of farm credit and important agricultural inputs, Table 10-1. Direct and Indirect Protection of Agriculture in Eighteen Sample Countries, 1960-84 (period averages in percentages) Tax due to Indirect industrial Direct Total Country group Period protection protection protection protection Extreme taxers 1960-84 -28.6 -25.7 -23.0 -51.6 Representative 1960-86 -24.2 -32.8 -12.0 -36.4 taxers Mild taxers 1960-83 -15.7 -22.9 -0.2 -15.8 Protectors 1960-84 -13.6 -13.9 24.0 10.4 Sample average -22.5 -27.9 -7.9 -30.3 Source: table 2-1. 202 Determinants of Agricultural Taxation; Conclusions such as fertilizer. Many developing countries, to increase their food self-sufficiency, protected domestic producers of import-competing food products through quantitative restrictions or tariffs on imported commodities. And some countries, responding to the instability of world markets, intervened to stabilize domestic producer prices rela- tive to prices on world markets. Yet other direct measures attempted to hold down the costs of food for urban consumers-by fixing retail food prices or imposing ceilings on producer prices. Sometimes governments set up dual pricing sys- tems to keep producer prices high and consumer prices low, making up the difference out of budgetary resources. Although the fiscal costs of food subsidies could be reduced by carefully targeting pro- grams, such programs sometimes proved impractical because of their complex administrative demands and the large size of the target groups. Explicit food subsidies often proved unsustainable (as in Sri Lanka), because of budgetary and balance of payments pressures, so some governments instead reduced consumer prices by taxing agri- cultural producers-or drastically reduced consumer subsidies. The Transfers Out of Agriculture Have Been Enormous In most industrial countries, the main objective of agricultural price policies is to maintain farm income and employment in the face of declining real world prices for cereals-with massive net income transfers to agriculture. In most developing countries, however, the primary objectives have been food self-sufficiency, domestic price stability, low food prices for urban consumers, and government reve- nue. On average the net effect of direct and indirect interventions has been an enormous income transfer out of agriculture, averaging 46 percent of actual agricultural GDP a year over 1960-84. These transfers ranged from 2 percent for the protectors to 140 percent for the extreme taxers. The average for the representative taxers was 37 per- cent. Such enormous transfers must have severely depressed private investment in agriculture and agricultural growth. An analysis of these income transfers-and of who gained and who lost from them-helps to reveal some of the motives for price inter- ventions and to identify the forces arrayed against reform. Agricul- ture was clearly the loser, while the big winners were government (net revenue gain), urban consumers (lower food prices), and indus- try (cheap raw materials and other inputs). The average net income transfer out of agriculture as a result of direct price interventions was about 11 percent of agricultural GDP during 1960-84 (it fluctuated between 4 and 19 percent, depending on the product coverage): direct price interventions on outputs reduced agricultural GDP by about 13 percent, and transfers into agriculture Summary and Concluding Remarks 203 through input subsidies boosted agricultural income by about 2 per- cent. Adding the transfers into agriculture from interventions not related to price policy (7 percent) results in a total net income transfer out of agriculture as a result of direct interventions of about 4 percent. However, indirect interventions had a much larger impact on income transfers out of agriculture. Taking into account the income effects of indirect price interventions raises the net transfers out of agriculture to an astounding 46 percent of agricultural GDP. For the representative taxers, total transfers out of agriculture averaged 37 percent of agricultural GDP. For the extreme taxers, the income effects of total interventions were staggering: if agricultural GDP had a value of 100 under total price interventions, it would have reached 240 if there had been no price interventions over the period. Also important is what we did not find. Input subsidies did not compensate, or compensated very little, for the substantial income outflows resulting from interventions in output markets, and in most cases, public investment in agriculture did not compensate for the negative effects of price interventions. For the representative taxers, the income transfer through input subsidies was never higher than 1.1 percent of agricultural GDP, and for all eighteen countries, the average for 1960-84 was only 2 percent (for direct interventions). Government compensation through higher investment was found, to varying degrees, in only five of fifteen cases with available data, with only Egypt and Morocco showing compensation for all six agri- cultural price policy or income variables tested. In sum: government and nonagriculture were the winners in most cases, and agriculture the loser. In the most frequent case, the govern- ment earned net revenues from direct price interventions (or trans- ferred resources out of agriculture), supporting the hypothesis that extracting resources was a frequent objective of direct price interven- tions, resulting in a strong antiexport bias in the trade regime. Agricultural Growth Was Slowed by Taxation To examine the impact of price policy on annual growth of real agri- cultural GDP, we compared the average agricultural growth rate in the group of countries in which nominal "taxation" rates were lower (nominal protection rates were higher) than the average with the rate in the group in which taxation rates were higher (protection rates were lower) than the average. The group with the lower taxation rates (higher protection) showed a higher average growth rate, a result that is statistically significant. We also used a regression estimate of the growth rate of real agricultural GDP to simulate its average growth rate in the absence of total price interventions. Here, too, the relationship between total taxation (protection) and agricultural growth was sig- 204 Determinants of Agricultural Taxation; Conclusions nificant: the lower the tax on agriculture, the higher the growth. The higher prices in agriculture reduce labor outmigration from the sector, increase investment, and encourage wider adoption of new techniques-and result in a higher growth rate. For the high and low direct protection groups, the difference in mean agricultural growth is small-1 percentage point, or 30 percent-and statistically not significant. But for the two total protec- tion groups, the difference is large-2.5 percentage points, or 90 percent-and significant (table 10-2). This provides strong evidence of an association between high total taxation and low rates of agri- cultural growth, but weaker evidence of such an association for the direct taxation rate. The finding seems to confirm that total (direct plus indirect) intervention rather than direct intervention is the rele- vant concept when examining differences in agricultural growth per- formance, because it also captures the substantial impact of industrial protection and exchange rate misalignment. Direct intervention, although used more commonly, captures only a small fraction of the impact of price interventions on agricultural incentives (about a quar- ter of the total impact on average). There are other reasons why total (direct and indirect) price inter- ventions might be more strongly associated with growth in real agri- cultural GDP than direct alone. The relation of interest is that between agricultural growth and agricultural incentives. Since indexes of agri- cultural incentives (relative prices) cannot be compared across coun- tries, we used taxation (protection) rates instead. And differences in total taxation rates tend to reflect differences in agricultural incentives more closely than do differences in direct taxation rates. This is so for several reasons. First, direct price interventions are used in part to stabilize domestic prices, and so rates of direct taxation (negative protection) tend to move inversely with border prices. Con- sequently, changes in direct taxation rates do not always reflect changes in producer prices. Indirect taxation rates, by contrast, would not be expected to be closely related to agricultural border Table 10-2. Agricultural Protection (Mostly Negative), and Agricultural Growth (percent) Direct Total intervention intervention Variable High Low High Low Nominal protection rate 10.8 -29.3 -8.3 -46.2 Annual agricultural growth rate 4.3 3.3 5.2 2.7 Source: table 4-1. Summary and Concluding Remarks 205 prices, so changes in total taxation (negative protection) tend more closely to reflect changes in producer incentives. This is confirmed by the finding that most of the reduction in producer price variability (relative to that of border prices) is due to direct price interventions, with indirect price interventions having a negligible impact. Second, the impact of direct intervention on agricultural GDP might be ambiguous. Some of the impact could be due to changes in output mix, since the impact of direct price interventions varies by product or product group-some are taxed and others are protected. So, the impact of low average direct taxation may be more important if it affects relative prices within agriculture more than a policy with higher but more uniform direct taxation. This is less likely to occur with indirect intervention: industrial protection policies reduce rela- tive incentives for the entire agricultural sector, and overvaluation of the real exchange rate taxes all tradable agricultural products, which constitute the bulk of the agricultural sector in most countries. Third, changes in direct intervention are not always positively cor- related with changes in total intervention. And the latter constitute a better measure of the impact of intervention on agricultural incentives than the former. It should be no surprise, then, that agricultural growth is more strongly associated with the sum of direct and indirect interventions than with direct intervention alone. A strong and sig- nificant association between agricultural output and total taxation was also obtained from an econometric model estimated to simulate the impact of total price interventions on agricultural GDP. Economic Growth Was Slowed by Agricultural Taxation The high tax on agriculture in many developing countries has been motivated in part by the belief that industry was the dynamic sector, whereas agriculture was static and not very responsive to incentives. Economic growth would thus be accelerated by shifting resources from agriculture to industry. This resource shift was to be achieved by taxing agriculture indirectly-by protecting the industrial sector and maintaining an overvalued real exchange rate-and directly. Several of the country studies explicitly mention such an industrialization strategy: Argentina, Chile, Colombia, the Dominican Republic, Ghana, Malaysia, Morocco, Pakistan, and Zambia. And most of the others suggest that such a policy was implicit in increasing govern- ment revenue by taxing agriculture and using it to increase spending outside agriculture. These policies did not accelerate growth, however. We examined the relation between price interventions and overall economic growth in three ways. First, we compared the various measures of interven- tion with the rates of growth of GDP across four groups of countries. 206 Determinants of Agricultural Taxation; Conclusions Second, we used regression analysis to relate the various measures of intervention to GDP growth for the eighteen countries. Third, we examined the relation between the various measures of intervention and GDP growth for the eighteen countries, dividing them into two groups based on the average size of the various measures of intervention. All three tests suggest a negative relation between the rate of total taxation of agriculture and GDP growth: policies that depress agricul- ture's terms of trade below international levels are associated with slower economic growth. We also found that GDP growth rose as the indirect taxation of agriculture fell across the four country groups. The analysis showed that greater exchange rate overvaluation was related to lower GDP growth-and the stronger the industrial protec- tion, the lower the GDP growth. Comparisons of price interventions and GDP growth for the four country groups show that higher rates of indirect and total taxation have been associated with lower economic growth (table 10.3). The GDP growth rate for the least interventionist group (the protectors) is twice as large (6.5 percent) as that (3.3 percent) for the most interven- tionist group (the extreme taxers). We cannot claim causality from higher taxation to slower growth, but this inverse relation between taxation and GDP growth is confirmed by the other two tests. In the regression analysis, GDP growth was negatively related to total taxation, to exchange rate distortion, and to indirect taxation. The regression results show that an increase of 10 percentage points in total taxation reduces annual GDP growth by 0.43 percentage points-and that an increase of 10 percentage points in the overvalua- tion of the exchange rate reduces growth by 0.28 percentage points. With an average annual GDP growth rate of 5 percent for the eighteen countries, a fall of 10 percentage points in total taxation would result in a 10 percent increase in GDP after twenty-five years. A reduction of Table 10-3. Price Interventions and GDP Growth, by Country Group, 1960-85 (percent) Nominal rate of protection Annual Country group Indirect Total GDP growth Extreme taxers -28.6 -51.6 3.3 Representative taxers -24.4 -36.4 5.1 Mild taxers -15.7 -15.8 5.3 Protectors -13.6 10.4 6.5 Source: table 4-3. Summary and Concluding Remarks 207 10 percentage points in overvaluation would result in a 7 percent increase in GDP. In examining the relation between GDP growth and intervention, we also looked at whether the average GDP growth rate was signifi- cantly lower for the countries with higher-than-average values of intervention measures than for the other countries. Indeed, GDP growth does fall with increases in the rates of total taxation (signifi- cant), in indirect taxation (not significant), and in industrial protec- tion (significant). The results show that policies that lower agriculture's domestic terms of trade below their international level are associated with a slowdown in economic growth-precisely the opposite of the intended effect from industry-led growth strategies that tax agricul- ture through direct and indirect interventions. The belief that eco- nomic growth could be accelerated by shifting resources from agricul- ture to industry rested on the premise that agriculture was a static sector, and it received further support from the prevailing pessimism about the prospects for agricultural exports. These were mispercep- tions from which both agriculture and the economy as a whole may have suffered in many developing countries. Farmers Do Respond to Price Incentives There has long been a presumption that the production of individual agricultural products responds quickly to higher prices-because of shifts between products-but that total agricultural production is slow to respond. The presumption is wrong. We have already shown that price interventions severely depressed agricultural prices during 1960-84 in the eighteen countries studied. Total taxation averaged more than 30 percent, reaching almost 40 percent or more in over half the countries. Direct price interventions reduced agricultural prices by an average of 8 percent, with the prices of exportables declining by more then the average and those of importables rising somewhat. Indirect price interventions led to huge reductions in agriculture's domestic terms of trade in most of the eighteen countries, averaging 22 percent. Without price interven- tions, agriculture's terms of trade would have been more than 40 percent higher during 1960-84. With such a large price impact from intervention, the effect of price reform on output is likely to be significant. How agricultural produc- tion responds to a price reform depends on how severely interven- tions have depressed prices, how extensive and credible the reforms are, how responsive output is to a given price change, and what time period is considered. Clearly, some of these elements are related. Responsiveness to a given price change depends on the credibility of 208 Determinants of Agricultural Taxation; Conclusions the reform, and credibility may depend on the extent and nature of the price reform. Aggregate output can expand even in the short run, through increased use of agricultural labor, capital, and variable inputs. The real value of agricultural GDP can also expand through the reallocation of agricultural labor, capital, and land to higher productivity uses in response to changes in relative agricultural prices. A price reform that affects relative prices within agriculture will induce a change in out- put mix. And the reallocation of a given set of resources to more productive uses within agriculture will generate real income gains. This can be illustrated with a simple example. On average, direct price interventions protected importables and taxed exportables, resulting in a net direct tax for the agricultural sector. Removing direct price interventions, then, would increase the price of exportables, reduce the price of importables, and lead to a net increase in prices for the sector. In other words, the sector would gain in the value of output. That is not the entire gain, however, because agricultural producers can move resources to the products whose relative price increased. In other words, an increase in agriculture's terms of trade leads to a rise in the value of output, and the change in output mix leads to an additional gain. Those are the short-term gains. Further increases in aggregate output are possible beyond those resulting from increased use of existing resources and changes in output mix. But they require bringing in more labor and capital from other sectors and wider adoption of new techniques. These processes take time. The efficiency gains from better agricultural price policies depend largely on the aggregate response of the agricultural sector to improved incentives-itself a function of the response of both existing products and new products. Few studies have directly analyzed aggregate supply response. Instead, much of the empirical work on supply response has dealt with individual products or groups of related products. Most of the debate, by contrast, has dealt with the aggregate response over time. The issue is important because if the aggregate response of agriculture is low, the cost of taxing agriculture is also low. We believe that the methods used to estimate the aggre- gate supply response have often severely underestimated that response. Estimating the long-run aggregate supply response requires an economywide perspective that takes into account intersectoral resource movements in response to changes in incentives across sec- tors. A few studies have done this, and they find that labor migration and investment respond to agriculture's terms of trade. Three observations on these results are worth emphasizing. First, even though the short-term and medium-term elasticities are small, the short-term and medium-term output effects are not, because of Summary and Concluding Remarks 209 the very large price increase that would have resulted from removing total price interventions. Second, the long-term effect is significantly larger with an approach that explicitly captures intersectoral resource movements. Single-equation regressions have resulted in lower esti- mates of aggregate supply response because they did not allow for intersectoral resource reallocation as a response to changes in incen- tives. Third, if estimation of the migration and investment functions underlying the elasticity estimates were based on price data reflecting large random price variations around a weak price trend, the parame- ter values obtained might strongly underestimate the response to a long-term (permanent) change in price policy. A general problem with simulating the effects of a permanent change in price policy using estimates based on time-series data is that this method is likely to underestimate the effects. Time-series data include price variations that are only temporary or that pro- ducers believe are temporary. Since the estimates are often obtained under conditions of general and relative price instability, they will underestimate the response to a permanent change in incentives under conditions of stability. Consider the issue of price policy changes and expectations about their duration. Farmers are less likely to respond strongly to price changes that reflect annual variations around a given mean price than to variations that reflect a change in the mean that is expected to last. Farmers would be likely to respond to temporary price changes only by altering their use of variable factors. If they expect a price change to last, however, they would also adjust the quantity of fixed factors, by investing in machinery and other improvements, and they might also adopt new techniques. In other words, the greater the variability in prices, the lower the response to a given change in price. A general climate of uncertainty can also dampen the response to a price change, and, again, models that fail to take this into account will underestimate the effects of credible price reform under more stable conditions. Macroeconomic instability, with high and variable rates of inflation and high variability in relative prices such as the real exchange rate, real interest rates, and asset prices, will discourage investment. Producers will try to hedge the value of their assets rather than invest in new production. Other types of uncertainty will have a similar effect. Chronic uncer- tainty about the definition and enforcement of property rights will tend to discourage long-term investments. This is the case in many tropical regions in developing countries, where lack of legal title to land has led to slash and burn agriculture, with harmful conse- quences for agricultural investment, the biomass, and the environ- ment. Periods of temporary uncertainty about property rights proba- bly have a similar, inhibiting effect on agriculture. During the Chilean 210 Determinants of Agricultural Taxation; Conclusions land reform program under the Frei administration, private invest- ment in cattle, tree crops, and structures fell dramatically, and farmers moved into crops with shorter production lags. The aggre- gate supply response in such circumstances would be expected to be low. Another reason that dynamic general equilibrium studies under- estimate output response is that they assume that total investment is given, so that price policy affects only the sectoral allocation of a given investment total. But eliminating price interventions, particu- larly indirect interventions, should result in a more stable macro- economic environment and more credible and sustainable policies. This change could well lead to higher domestic and foreign invest- ment, and thus to a stronger supply response than that estimated under conditions of macroeconomic instability. The fiscal effects of specific agricultural price policies have received much attention and generated intense debate. But there have been few (if any) systematic attempts to quantify the net fiscal impact of price interventions. Consumer food subsidies have frequently been cited as a major drain on government budgets. Although they may be a drain for some food-importing countries, however, they are not for most countries. Similarly, many countries have a tradition of subsi- dizing agricultural credit and inputs, conventionally interpreted as compensating producers for the heavy taxation of agricultural pro- duction. Yet such subsidies represent only a small part of government expenditure (averaging only about 2 percent over 1960-83), whereas taxing agricultural exports yields substantial revenue to the govern- ment (averaging 10 percent of government expenditure). Indeed, the net effect of direct price interventions in agriculture was a revenue gain to the government of nearly 7 percent over the period-and as much as 17 percent during the 1960s. Price Interventions Affect the Budget in Many Ways Most direct agricultural price interventions affect the government accounts. The effects of some interventions, such as outlays for agri- cultural input and production subsidies or revenues from explicit border taxes, are obvious. In other cases, such as the effects of price interventions on the operating deficits or surpluses of state-owned marketing boards, they are more complex, or even hidden. Whatever the price intervention, it is likely to have some fiscal impact, and that fiscal impact influences agricultural price policy. This dynamic relationship between price policy and fiscal policy is especially clear in agricultural export policy and food and input sub- sidies. The government's need for revenue to fund expanding devel- opment programs was probably the major impetus behind taxing Summary and Concluding Remarks 211 agricultural exports, and it remains the major constraint to reform. The government revenues from the export operations of the Cocoa Marketing Board in Ghana and from the export tax on grains and beef in Argentina are examples of the importance of the revenue motive- and the obstacles it poses for reform. In some countries the fiscal burden of input and food subsidies escalated so rapidly that it led to macroeconomic imbalances that could be corrected only through pol- icy reform. Changes in agricultural policies in Portugal were dictated chiefly by budget pressures in the early 1980s, when fertilizer and feed subsidies were essentially eliminated. Attempts to reduce food subsidies in Egypt, Morocco, and Pakistan also show the link between price policy and the budget. In most industrial countries the major impetus for reform of agri- cultural programs now comes from fiscal pressures, even though income transfers seem to be the primary objective of agricultural poli- cies in these countries. In developing countries, by contrast, the links between price policies and the government budget arise mainly through the contribution of agricultural exports to fiscal revenues and the cost of government outlays for food subsidies to urban consumers. This analysis of the budgetary effects of price policies considers only direct price interventions in agriculture. But these effects are only a part of the overall income gains and losses to agriculture from price interventions. Indirect interventions-industrial protection and overvaluation of the exchange rate-generate income transfers to other sectors from agriculture that are off-budget items, many of which yield no revenue to the government. Taxes that apply to other sectors as well as to agriculture, such as income taxes and value- added taxes, have no direct link with agricultural price policies and would continue independent of agricultural price reforms. There nevertheless are some indirect links between macroeconomic variables and the budget that could influence the design of direct price interventions. Most of these are difficult to measure and were not included in our calculations of the budgetary effects of price inter- ventions because they were not quantified in most of our country studies. One link involves the nominal exchange rate. For Zambia, especially since 1980, an overvalued exchange rate has resulted in substantial fiscal savings in government operations involving maize and fertilizer imports, given the government's targeted domestic prices for consumers. Another complex link sometimes arises between farm prices and the government budget in middle-income countries, where agricul- ture is usually a relatively small sector. In Chile, government lending at concessional rates to the nonfarm private sector was at least partly a compensating mechanism to offset the effects of its agricultural 212 Determinants of Agricultural Taxation; Conclusions policies. The credit program prevented a decline in nonfarm invest- ment as a result of the substantial rise in nominal urban wages during the 1960s-a consequence of higher farm prices and the downward rigidity of urban real wages resulting from strong trade unions. The Contribution to Revenue Can Be Considerable In most countries the net budget effect of price interventions in agri- culture was a gain in revenue, mainly through taxes on agricultural exports. For the representative taxers, price interventions in output and input markets during 1960-84 produced revenues of almost 8 percent of government spending (table 10-4). Output taxes alone pro- duced as much as 20 percent of government spending in the 1960s. Over time, however, the net contribution of the price interventions for this group of countries fell substantially, from 18 percent of gov- ernment spending in the 1960s to 5 percent in 1980-84. In some coun- tries, however, the revenue contribution remains important-and undoubtedly constrains any policy reforms to reduce direct agri- cultural taxation. In those countries, sectoral reforms that would elim- inate or reduce agricultural export taxes would need to be accom- panied by economywide tax reform and other fiscal reforms. Agricultural price interventions resulted in net outlays for a few countries-Brazil, Korea, Morocco, Pakistan, Turkey, and Zambia- and on average for the mild taxers in 1980-83 and the protectors. In Brazil the net budgetary effects of price interventions varied dramati- cally, from a revenue gain of 32 percent of government spending during 1966-69 to a net outlay of 43 percent during 1980-83, mainly because of a large increase in credit subsidies to agriculture. Export taxes, including the profits of state agencies active in exports, were the single most important source of revenue from gov- ernment interventions in agricultural markets. On average during 1960-84, export taxes amounted to about 10 percent of government spending (20 percent in 1960-69). And in three countries-Brazil, Ghana, and Sri Lanka-export revenues amounted to 20 percent or more of total government spending. Credit and food subsidies had the next largest impact on the bud- get, though at less than 2 percent of government spending it was much less than that of export revenues. These subsidies were nev- ertheless substantial in some countries. Credit subsidies were 20 per- cent of government spending in Brazil and 4 percent in Colombia. And consumer food subsidies constituted a fairly high share of gov- ernment spending in Brazil, Egypt, Morocco, Pakistan, and Sri Lanka (table 10-4). The interaction between price interventions and budget effects dif- fered systematically for food exporters and food importers. Food Table 10-4. How Direct Price Interventions Affected the Budget, 1960-84 (percentage of total public expenditures) Output policies Input policies Consump- Total Total Export Import Production tion sub- output Credit Input input State Net Country group Period tax tax subsidies sidies effects subsidies subsidies effects agencya effects Extreme taxers 1960s 20.6 0.0 0.0 0.0 20.6 0.0 0.0 0.0 0.0 20.6 1970s 13.2 0.0 0.0 0.0 13.2 0.0 0.0 0.0 -3.0 10.2 1980s 0.2 0.0 0.0 0.0 0.3 0.0 0.0 0.0 -4.3 4.1 1960-84 9.4 0.0 0.0 0.0 9.4 0.0 0.0 0.0 -3.4 6.0 Representatives 1960s 22.1 0.1 0.0 -2.2 20.0 -0.6 -0.5 -1.1 -0.7 18.2 taxers 1970s 12.0 0.1 -0.1 -2.8 9.2 -0.6 -0.8 -1.2 0.4 8.3 1980s 8.6 0.0 -0.1 -1.7 6.8 -1.3 -0.8 -2.1 0.0 4.8 1960-85 11.4 0.1 -0.0 -2.4 9.0 -0.7 -0.7 -1.3 0.2 7.9 0 Mild taxers 1960s 15.8 0.1 -0.0 0.6 16.5 -3.3 -0.0 -3.3 -0.0 13.2 1970s 9.4 0.3 0.1 -2.1 7.7 -6.7 0.0 -6.7 0.0 1.0 1980s 2.0 0.2 -0.1 -4.3 -2.1 -11.5 0.1 -11.4 0.0 -13.5 1960-83 9.3 0.2 0.0 -2.0 7.6 -7.1 0.0 -7.0 -0.0 0.5 Protectorsb 1960s - - - - - - - - 0.0 - 1970s 0.0 0.7 -1.2 -1.2 -1.6 -1.6 -1.6 -1.0 -2.60 -4.2 1980s 0.0 1.0 -0.4 -0.4 0.2 -1.0 -0.6 -1.6 0.OS -1.4 1960-84 0.0 0.8 -0.9 -0.9 -1.0 -1.4 -0.9 -2.3 0.0 -3.3 Sample 1960s 19.8 0.1 -0.0 -1.02 18.9 -1.4 -0.3 -1.7 -0.4 16.8 average' 1970s 10.9 0.2 -0.1 -2.21 8.8 -1.7 -0.5 -2.3 -0.1 6.4 1980s 5.8 0.1 -0.1 -1.90 3.9 -3.0 -0.6 -3.6 -0.5 -0.2 1960-83 10.0 0.1 -0.1 -1.95 8.2 -1.8 -0.5 -2.3 -0.3 6.6 - Data not available. a. A value of zero for state agency does not necessarily mean that the agency had no profits or losses. Instead it may indicate that any gains or losses were allocated to other categories. b. Republic of Korea only. c. Simple, unweighted group average. Source: table 6-2. 214 Determinants of Agricultural Taxation; Conclusions exporters like Argentina and Thailand kept domestic food prices lower than border prices through export taxes and gained revenue in the process. But food importers like Egypt, Morocco, Pakistan, and Sri Lanka incurred substantial fiscal costs for food subsidies to keep consumer prices low. In most of these countries, food imports gener- ated no revenues because the state agencies with a legal monopoly on staple imports paid no trade taxes-whereas other agricultural importables were protected through quotas rather than revenue- generating tariffs. Remember in all this that the revenue and expenditure effects described here were the result of direct interventions only. The net effect of these interventions was to provide revenue to the govern- ment, but a far larger impact on agriculture resulted from off-budget items not quantified here-indirect price interventions, most of which do not provide revenue to the government. Price Distortions in Agriculture Have Increased Government price interventions discriminated strongly and increas- ingly against the agricultural sector in developing countries during 1960-84. By far the largest impact came through indirect interventions (industrial protection and macroeconomic policies), which reduced incentives in agriculture relative to other sectors of the economy. On average, direct interventions taxed exportables and protected import- ables, predominantly food products. In most countries consumer prices for food products moved in the same direction as producer prices in response to direct interventions. So direct interventions, by protecting importables, raised consumer food prices in most cases. Indirect interventions, however, lowered both producer and con- sumer prices. Price interventions and their effects are not uniform across product categories-importables (mostly staples) and exportables-in part because policy objectives differed. Taxes on agricultural exports are an important source of revenue for many developing countries, and a desire to reduce a country's dependence on food imports tends to limit taxation of the domestic production of importable commodities, mainly food products. One would therefore expect to find that exportable products are taxed more than importables. The surprise is that direct interventions have actually protected the production of importables in eleven of sixteen countries studied (Argentina and Thailand have no important importable products), with an average protection rate of over 14 per- cent for the sixteen countries (table 10-5). In nine of those eleven countries, the average protection rate for importables was above 17 percent, implying that producers received over 17 percent more for Table 10-5. Direct and Total (Nominal) Protection, by Product Category, 1960-84 (period averages in percentages) Staples Importables Exportables Direct Total Direct Total Direct Total Country group Period protection protection protection protection protection protection Extreme taxers 1960-84 18.7 -9.9 17.6 -11.0 -20.5 -49.1 Representative taxers 1960-86 -4.5 -28.9 7.8 -13.0 -15.9 -40.2 Mild taxers 1960-83 14.2 -1.5 14.2 -1.5 2.1 -13.6 Protectors 1960-84 26.9 13.3 28.4 14.8 0.5 -0.8 Sample average 1960-86 6.0 -16.5 14.4 -8.6 -12.6 -34.8 Source: table 2-2. 216 Determinants of Agricultural Taxation; Conclusions their products than they would have in the absence of direct interven- tions. (With the usual method of calculating nominal protection rates, direct protection was over 20 percent in those nine countries.) Since most importables are staples (except barley and corn, which are used mainly as feed), staples have a tendency to follow the direct protection pattern of importables. A few staples are exportables, however, so direct taxation rates for staples and importables differ in some countries. Consequently, the cross-country average direct nom- inal protection rate for staples is lower than for importables, but it is still positive at 6 percent. About half the countries directly protect the production of staples, possibly to increase self-sufficiency. While direct interventions protected importables and staples in most countries, they taxed exportables in thirteen of seventeen coun- tries (Korea has no important agricultural exportables). The average direct tax on exportables was 12.6 percent. Adding indirect interven- tions brings the total tax to 35 percent. That means that farmers would have received more than 50 percent (35/65) more for their exportable products (relative to prices in the nonagricultural sector) had the government not intervened in agricultural prices. Since importables were generally protected while exportables were taxed, relative prices within agriculture were distorted, sometimes severely. In C6te d'Ivoire, the direct protection rate of rice in relation to coffee and cocoa was 77 percent on average over the period. In Malaysia, corn had a 51 percent protection rate relative to sugar, while in Sri Lanka rice's protection in relation to tea was 84 percent. On average for all the countries examined, the protection rate on import- ables in relation to exportables was over 30 percent. Clearly, direct interventions must have led to substantial distortions in resource allocation within agriculture in many countries. Protection rates tend to vary over time as external circumstances, policies, and policy objectives change. Both external factors (the com- modity price boom, the oil price shocks, and the debt crisis) and internal factors (changes in trade and other policies or in government ideology) might be expected to influence price intervention policies and their effects. The mid-1970s were a period of substantial change in international economic conditions-and a plausible demarcation point for changes in the pattern of price intervention. The commodity price boom of 1972-75 inflated the food import bill of most developing countries, giving rise to increased concerns about food security and self- sufficiency. In response, one might expect an increase in the rate of direct protection on importables in 1976-84 compared with 1960-72. At the same time, the two oil price shocks of 1973 and 1979 and the interest rate shock of the early 1980s caused increases in the current account deficits of developing countries-shocks that could be expected to increase the rate of indirect taxation. Summary and Concluding Remarks 217 Although we expected an increase in the rate of direct protection for importables in the second period because of concerns for food security, the cross-country average (excluding Ghana) actually fell from about 23 percent in 1960-72 to 19 percent in 1976-84. If we also exclude Korea, where direct protection increased dramatically from 35 to 91 percent, the average fell from 22 to 15 percent, or by close to one-third. For the representative taxers, direct protection to import- ables fell by about 60 percent, from about 25 to 10 percent. Direct protection to importables fell significantly in five countries-Brazil, Colombia, Pakistan, Portugal, and Sri Lanka- and increased signifi- cantly in six-Chile, C6te d'Ivoire, the Dominican Republic, Ghana, Korea, and Malaysia. By contrast, we did not expect any significant change in direct protection rates for exportables as a result of the economic changes of the mid-1970s. The average direct tax did increase somewhat, from almost 14 percent to 16 percent between 1960-72 and 1976-84, with the largest increase coming in the group of mild taxers (from 3.6 to 6.2 percent). In most countries, direct taxation (negative protection) rates rose overall from 1973 to 1976 as countries tried at least partially to stabilize domestic prices during the world commodity price boom. Direct pro- tection rates also fell during the boom of 1979-81 in many countries, but by less. On average, the direct tax rose from 4.9 to 6.2 percent between 1960-72 and 1976-84. As expected, the rate of indirect taxation increased from 21.5 per- cent on average in 1960-72 to 26.8 percent in 1976-84, or by 25 per- cent. The rate increased in twelve of the eighteen countries, with the largest increases in the three Sub-Saharan African countries, the extreme taxers. The average for that group almost tripled between the two periods-from 18.5 to 55 percent-giving them not only the high- est increase between periods but also the highest indirect tax of all groups in 1976-84. Of the countries whose indirect tax fell between 1960-72 and 1976-84, the largest drops came in Chile and Korea, which had introduced broad-ranging liberalization programs. Overall, total taxation from direct and indirect intervention rose between 1960-72 and 1976-84-from a positive rate of protection of 1.4 percent to a tax of 8 percent for importables and from a tax of 35 percent to a tax of 42.8 percent for exportables. The total tax for all products increased from 26.4 to 33 percent. Direct Interventions Sometimes Increased as Indirect Interventions Rose or Fell In some countries the relationship between changes in indirect taxa- tion and direct taxation was important. Countries that undertook major economic reform through trade liberalization and macro- 218 Determinants of Agricultural Taxation; Conclusions economic stabilization measures that reduced indirect taxation often made changes in direct price intervention policies as well. Other countries accompanied trade liberalization measures and lower indi- rect taxation with increased direct price intervention in agriculture. In some countries, indirect taxation rose and. was partly compensated by higher direct protection. For some, direct agricultural price policies seem to have had little relation to economywide policies or to external events affecting indi- rect taxes. And some countries exhibited considerable stability in all their economic policies. On average for the eighteen countries during 1960-84, actual rela- tive producer prices exhibited little trend. Producer prices fell by 0.2 percent a year for exportables and exhibited no change for import- ables. Over the same period world prices increased at an annual rate of almost 1 percent for exportables and 0.25 percent for importables. The trend in actual producer prices was lower than the trend in world prices (defined as producer prices in the absence of total interven- tion), principally because of the increase in indirect taxation (indus- trial protection policies and overvaluation of the real exchange rate) over the period. Direct Interventions Raised the Cost of Food to Urban Consumers It is commonly believed that developing countries tax agriculture by keeping food prices below their border-price equivalent to appease the more vocal and influential urban consumers and to lower urban wages for the benefit of private and public employers. Our results show, however, that direct price interventions slightly protect food production on average, implying a small tax on food consumption. And with few exceptions, such as Egypt and Zambia, consumption is taxed when food is imported. Direct protection of food prices in the thirteen countries for which data were available on the effects of direct and total interventions on consumer food prices was on average positive and small (6 percent), indicating a small tax on consumers. Direct interventions taxed con- sumers in Brazil, Ghana, Korea, Portugal, and Turkey. Whereas direct interventions taxed food consumption on average, indirect interventions lowered relative food prices in all countries, implying a subsidy to food consumption relative to nonagricultural prices. Total (direct and indirect) interventions also lowered food prices in all countries except Brazil, Portugal, and Turkey, resulting in an average consumer subsidy of 18 percent. In all countries (except Sri Lanka) with a rate of direct protection of consumer food prices larger than 5 percent (in absolute value), the rate of direct protection of consumer prices and the rate of direct protection of producer prices for staples have the same sign. That Summary and Concluding Remarks 219 means that when consumption has been taxed, the production of staples has been protected. And when consumption has been subsi- dized, production has been taxed-so both prices move in the same direction. In sum: direct price interventions have had a relatively small effect, an 8 percent tax on agriculture on average. But they had a strong effect within agriculture, protecting the production of importables (about 14 percent) and taxing the production of exportables (about 13 percent). Of the indirect interventions, industrial protection has penalized agriculture more than overvaluation of the real exchange rate in two-thirds of the countries examined. The effect of indirect taxation has dominated both direct taxation of exportables and direct protection of importables. Direct price intervention did not lower the cost of food to urban consumers on average; it even raised it some- what. In most countries that import food, direct intervention has raised prices for both consumers and producers. And in most cases in which direct intervention lowered prices to producers, consumer prices also fell. The analysis of agricultural price interventions in developing coun- tries has two related dimensions. One is defining what constitutes a sectoral (agricultural) price intervention, considering the various forms of domestic assistance to agriculture. The other relates to the influence on agriculture of economywide policies affecting the real exchange rate and industrial protection, and thus the competitiveness of agriculture vis-A-vis other domestic sectors and foreign markets. Domestic agricultural prices are influenced by such border mea- sures as tariffs and quantitative import and export restrictions and by such nonborder measures as input subsidies, subsidized credit, and irrigation and transport charges. Trade policies are the principal force behind internal agricultural price and income policies; without border barriers on the same products, nonborder policies to protect the pro- duction of importables (such as input subsidies) would have a rather limited effect on output and trade flows. Since World War 11, in both developed and developing countries, there has been an upsurge in the proportion of agricultural trade subject to nontariff barriers such as quantitative restrictions, govern- ment procurement policies, and technical barriers to trade (including health and sanitary regulations). Many of these nontariff barriers, including state trading, could render unimportant most of the tariffs and explicit export taxes. In developed countries, the production sub- sidies for agriculture are widespread. But in most developing coun- tries the most influential price interventions are border measures- and of these, quantitative restrictions (QRs), rather than tariffs are the principal policy instruments. Such QRs include state trading, foreign exchange licenses, quotas on imports, and licenses and quotas on exports. 220 Determinants of Agricultural Taxation; Conclusions Dismantling QRs and replacing them with tariffs and explicit export taxes has several important advantages and should be a condition of a trade liberalization package. QRS insulate the domestic market from world price changes. They make the trade and price regime highly discretionary and selective. And they are less visible than tariffs, masking the level of protection or taxation and permitting discretion- ary interventions. Yet another advantage of replacing QRS with tariffs is that tariffs generate government revenues, removing one of the obstacles to trade reform in some countries. Although they are declin- ing in importance, revenues from agricultural export taxes have been a significant share of government revenues in several developing countries, and they are still a high share in a few countries. A further advantage is that tariffs, if they are accompanied by the elimination of the legal monopoly in trade that state agencies have in several coun- tries, would contribute to reducing the role of state trading in agri- cultural commodities. Because low-income families typically spend half or more of their income on food, government policies affecting food prices could have a direct and powerful impact on the living standards of the poor. Providing a regular supply of the basic staples at low prices to con- sumers is a stated objective of agricultural price policies in most developing countries. But government policies cannot protect all con- sumers, and faced with budgetary pressures, most governments try to reduce the fiscal burden of food subsidies by transferring part of the cost to producers of farm products, through price controls. So not all low-income groups have benefited equally from price interven- tions, and some have borne the costs. Without price interventions, the relative incentives for agricultural production would have been much stronger, leading to a more dynamic sector with larger output and productivity gains over time, and positive effects for the economy as a whole. Thus, in assessing the net effect of price interventions on poverty in the long run, we look at the combined effect of the short-run gains to consumers from relatively lower food prices, and the cumulative negative effects on overall agricultural output, employment, and rural income. We find that, in the long run, the net effect of price interventions was in many cases to reduce the real income of the poor more than that of other income groups. For Urban Consumers, the Short-Run Income Losses Were Small in Most Countries Contrary to the widely held view that cheap food policies prevail in developing countries, the impact of direct price interventions on the real income of urban households during 1960-84 suggests that food price policies penalized urban consumers in six of fourteen sample Summary and Concluding Remarks 221 countries. Most countries protected the production of agricultural importables, mostly food. Only a few of these countries, however, also reduced food prices to consumers through explicit food sub- sidies. And while direct interventions subsidized urban consumers in eight of the fourteen countries, in only four of them (Argentina, Egypt, Pakistan, and Turkey) were the real income effects on both low- and middle-income households higher than 3 percent of their income. Indeed, the income gains or losses from direct interventions were small in most countries for all three income levels. So, despite widespread interventions in food prices, the impact on the real income of urban households was generally small. For most countries the income effects of indirect price interventions intensified those of direct interventions. In Argentina, Brazil, Chile, Colombia, the Philippines, Turkey, and Zambia, the effects of indirect interventions were so large that the total effects were at least double the direct effects. Among food exporters like Argentina and Thailand, the effects of subsidies to consumers represented by export taxes on food products were reinforced by the effects of indirect interventions. But in some countries that taxed consumers through direct interventions-such as Korea, Colombia, and Portugal-the indirect effects led to an implicit subsidy that counteracted the tax in several subperiods (but not so on average for Portugal). The total income gain for low-income urban consumers ranged from 0.2 to 5.3 percent in seven of fourteen countries and was 10 percent or more in three-Argentina, Egypt, and Turkey. Overall, these results confirm that in most countries the effects of what are commonly referred to as cheap-food policies take place mainly through exchange rate misalignment and not through food policies or other direct price interventions. The relatively small effect of direct government interventions on the consumers' mean real income in most countries, together with the greater stability of domestic food prices compared with border prices, suggest this: in addition to reve- nue and self-sufficiency, another big motive underlying food price interventions may have been to prevent sudden large real income losses in years of higher-than-average food prices rather than to raise the standard of living of the urban poor. Unstable world market prices for food products, if passed on directly to consumers, can lead to significant fluctuations in the real income of consumers, a politically sensitive outcome in most developing countries. For the Rural Poor, the Short-Run Income Losses Were Substantially Larger Differences in consumption patterns, production mix, and marketa- ble surplus as a proportion of farm output account for most of the differences in income effects among farm (or rural income) categories. 222 Determinants of Agricultural Taxation; Conclusions The results are for the short-term effects only-given levels of produc- tion and given consumption patterns. In Ghana and Korea, and less so in Brazil, small farmers (or poorer households) benefited from direct price interventions. In Ghana direct interventions benefited small farmers and taxed large farmers, whereas in Korea and Brazil large farmers benefited even more than small farmers. Ghana was the only country of seven analyzed in which the income effects (as a percentage of real income), whether positive or negative, were greater for small farmers than for larger farmers. In Egypt, Thailand, and Turkey, direct interventions taxed both small and large farmers during most of the period, although in Thailand the income effect was very small-less than 1.8 percent for all income groups. In all three countries, large farmers were taxed more than small farmers. In almost all cases, indirect interventions worsened the income effects for rural households, reducing or obliterating the income gains from direct interventions, as in Brazil, Ghana, and Korea, or increas- ing the losses, as in most other countries. The indirect effects reduced the income losses for some periods and some income groups in Ghana (larger farmers) and to a much smaller extent in Thailand (all farm-size groups). Price interventions had a relatively larger real income effect on farm producers than on urban consumers. Direct price interventions had considerably more impact on the real income of larger farms (or wealthier rural households), whether these effects were positive (Korea and Brazil [for the products examined]-except 1980-83-and Portugal) or negative (Egypt and Turkey). Farms of all sizes were taxed heavily in Egypt, especially larger farms, which are more export-oriented and tend to sell a higher proportion of their produc- tion. Ghana was an exception, with small farmers gaining from direct price interventions because they produce more rice, which is pro- tected, and larger farmers losing because they produce more coffee and cocoa, which are heavily taxed. Where direct interventions protected farmers, as in Brazil (for the products examined) and Korea, richer farmers or households bene- fited more than poorer ones. In Portugal direct interventions had almost no effect on the income of small and medium-size farms, but benefited larger farmers. Thailand was the only case in which the real income effects of both direct and total price interventions were neglig- ible during 1960-84. On the whole, the income effects for rural households were sub- stantially higher and in the opposite direction from those for urban households in the same countries. Removing all price interventions would have brought real income benefits to farm households at all levels of living and by a substantial proportion of household income Summary and Concluding Remarks 223 in Brazil (around 15 percent in 1980-83), Egypt (between 35 and 56 percent for all groups throughout 1964-85), Ghana (17 to 37 percent for larger farmers during 1962-76), and, to a lesser extent, Turkey (2.8 to 4.8 percent for all income levels during 1961-79). Because of the pattern of direct interventions, removing total price interventions would have reduced the real income of farmers in Korea (except 1962- 69), Portugal (for larger farmers), and Ghana (small farmers in 1970- 76). In the Longer Term, the Poor Have Probably Lost Disproportionately In the long run, the net effect of price interventions may in many cases have been regressive. To the extent that the benefits from food subsidies were captured mainly by urban households-and predomi- nantly employees in the government and formal sector-at the expense of the rural sector, the cost may have been a reduction in the real income of the rural poor and, indirectly, the urban poor in the informal sector. If taxing agriculture reduces rural demand for labor, rural employment and real wages will fall, leading to increased migra- tion to the cities and increased competition for employment, and thus to a fall in income or an increase in unemployment in the informal urban sector as well. The net effect would thus be regressive, since the bulk of the poor in developing countries are in rural areas, whereas the poor in urban areas are largely in the informal sector, which is unprotected by minimum wage or other labor legislation. Studies of Brazil, Chile, and Colombia show that removing inter- ventions would have raised the real income of landless labor, the poorest group in the rural sector and generally the poorest group in the economy. And by reducing rural migration, the removal of inter- ventions might also have raised the income of the urban poor, although this effect is less clear because of the possible benefits of food subsidies for the urban poor. In the long run, the cumulative net effect of price interventions on the real income of the poor depends on the combined effect of the very short-run impact on the cpi and the impact on overall economic growth, which affects employment and nominal income. Estimates of short-run effects for fourteen countries provide a rough indication of the direction of change in the real income of the poor as a result of price interventions. On average, total interventions for the fourteen countries taxed agriculture at a rate of about 26 percent, resulting in an average annual loss in GDP growth of 1.1 percent, or 23 percent after twenty years. For the urban poor, static distributional gains from total inter- ventions (the cpi impact) averaged 6.5 percent. Assuming a constant 224 Determinants of Agricultural Taxation; Conclusions share of labor income in total income, the net effect (static gains plus cumulative impacts on growth) on the urban lower-income group of total interventions from the early 1960s to the early 1980s was a real income loss of over 15 percent. For the urban poor in all countries except Egypt, the loss resulting from the negative impact on growth was higher than the static income gains from price interventions, so the urban poor lost over time. In Egypt after twenty years, the net effect on the poor was slightly positive because of static income gains of 43.9 percent and a cumulative loss due to slower growth of about 40 percent. Over twenty-five years, however, the net effect on the urban poor is nega- tive, and the effect becomes increasingly negative over time because the static gain is a once-and-for-all gain whereas the loss in growth accumulates over time. The only country in which the rural poor gained from interventions is Korea, where agriculture is a small share of the economy. In Korea, the static effects on the rural poor were about 7 percent, and the growth impact of the interventions was small. Although these results suggest that substantial benefits would accompany the removal of price interventions, efforts to do away with food subsidies have faced strong resistance from urban con- sumers. The loss to urban consumers as a result of such reforms is immediate and easy to see; the benefits of higher growth appear only over time, however. In Egypt, Morocco, Zambia, and elsewhere, gov- ernment attempts to remove food subsidies led to riots and forced a reversal of the reforms. Agricultural price policies clearly are a hot issue. Most developing countries lack the risk-diffusion mechanisms needed to help agricultural producers adapt to highly volatile world prices. That is why many governments intervene to stabilize domestic producer (and consumer) prices, and thereby to stabilize income, to improve production efficiency, to reduce macroeconomic instability, and sometimes to redistribute income among the rural, urban, and government sectors. The macroeconomic argument for price stabilization is that fluctua- tions in the world price of agricultural exports will lead to fluctuations in export revenues and so to fluctuations in government revenue, the real exchange rate, wages and prices, and other macroeconomic vari- ables. Stabilizing the domestic price of important exportables (say, through a stabilization fund) may partly insulate the economy from external shocks and so reduce macroeconomic instability. It is impor- tant, however, to distinguish the impact of price stabilization on urban consumers and on farmers and rural consumers-and to distin- guish the effects of stabilization of prices, nominal income, real income, and consumption. Stabilizing the price of staples that make Summary and Concluding Remarks 225 up a large share in the urban consumer budget, in order to prevent large fluctuations in urban consumers' real income, is a political imperative in many developing countries. Stabilizing the prices of food products will also tend to stabilize the real income of rural con- sumers, such as landless laborers. But if the payment for their labor closely approximates their consumption patterns (of food, say, or if they receive free housing and a plot to grow food), the stability of their real income will depend less on the price of food. For farmers there are two principal arguments for stabilizing prices: reducing rural income fluctuations and improving production effi- ciency. What matters to risk-averse farmers is stabilization of general consumption (or utility) over time. Traditionally, through output diversification and liquidity, farmers have been able (partially) to sta- bilize consumption, but at a cost of reduced average consumption. Those are the arguments. But what were the actual effects of inter- ventions on price variability? First, were direct and indirect agri- cultural price interventions actually successful in stabilizing domestic producer and consumer prices? Second, did direct and indirect inter- ventions have different effects on price variability? Third, did the price stabilization vary by product group? Fourth, did the stabiliza- tion differ more when world prices rose than when they fell? And fifth, did the degree of price stabilization differ in the 1960s and in the 1970s and 1980s, when the world prices were more volatile? To answer these questions, we used three measures of price vari- ability: the standard deviation, the coefficient of variation, and the Z-statistic. Each has its merits. The standard deviation measures the degree to which a variable is dispersed around its mean value. The coefficient of variation-the standard deviation divided by the sample mean-standardizes the scale of the variable, so that the degree of dispersion of variables with widely differing average values can be compared. The Z-statistic, a measure of the average deviation of the price from its value in the preceding period (annual price change), is probably a better indicator of producers' concerns; farmers are more likely to be concerned with price deviations from the previous year prices than with deviations from the sample mean. Direct Interventions Reduced the Variability in Producer Prices On average for all countries examined, direct and total interventions stabilized domestic producer prices in relation to world prices in each commodity group-importables, exportables, staples, and nonsta- ples. The standard deviation showed a reduction in variability of more than 30 percent in all product categories for both direct and total interventions, except nonstaples, where the reduction through direct 226 Determinants of Agricultural Taxation; Conclusions interventions was only 10 percent. The other two measures also showed the impact of direct and total interventions in reducing price variability. The results also hold for the three groups of taxers, but not for the protectors (because of price interventions in Korea). Direct and total interventions have tended to reduce annual price variations more than they did price variations around the mean. It seems likely that farmers are more concerned with annual price changes than with price variations around a twenty-five-year aver- age. If so, the larger reduction in annual price variations (over 40 percent for direct interventions and over 50 percent for total interven- tions on average) than in variations around the mean would indicate some success in reducing the price risks of special concern to producers. Total interventions reduced price variability more than did direct interventions, but direct interventions had a far larger impact on price variability than did indirect interventions. Adding the effects of indi- rect interventions to those of direct interventions somewhat reduces price variability. That indirect intervention did not substantially improve price stability should not be surprising. Real exchange rate overvaluation due to expansive macroeconomic policies (large budget deficits) is usually accompanied by high and variable inflation rates and high relative price variability. Moreover, the main component of indirect intervention is industrial protection, and there is no reason to expect industrial protection policies to reduce relative price vari- ability. Further, the world prices of industrial goods are much less volatile than those of agricultural products. In fact, we found that in most countries, indirect interventions resulted in an increase in price variability as measured by the coefficient of variation. Direct Interventions Stabilized the Prices of Staples More Than Those of Other Products Direct interventions reduced price variability considerably more for staples than for nonstaples. Since both consumers and producers have an interest in price stability for staples, whereas price stability for nonstaples is of more concern to producers, it is not surprising that direct price interventions stabilized staples more than nonsta- ples. The same reasoning applies to importables compared to export- ables, since importables consist mostly of staples and exportables consist mostly of nonstaples. Overall, then, producers of importables (and staples) benefited twice compared to producers of exportables, since producer prices for importables were both more stable and more highly protected by direct intervention. Did the authorities try to stabilize prices more during commodity price booms than during busts? No. But they did try more price Summary and Concluding Remarks 227 stabilization during periods of high price variability than at other times. Direct Interventions Reduced the Variability in Consumer Prices in Most Countries Direct and total interventions stabilized consumer prices for staples on average for the thirteen countries for which information was avail- able (the average ratios for the three variability measures are all smaller than one). As with producer prices, price policies stabilized annual price variations more than price variations around the mean. Direct interventions stabilized consumer prices more than indirect interventions, which slightly reduced consumer price variability. All three measures show a reduction in consumer price variability for staples in Brazil, Colombia, the Dominican Republic, Egypt, Morocco, the Philippines, Thailand, and Turkey-where price policies also stabilized the producer prices of staples. In the five other coun- tries, either consumer price variability increased or the results were mixed. On the whole, then, price interventions have achieved the objec- tive of stabilizing prices, mainly through direct interventions. Indirect interventions, the major source of agricultural taxation, did little to improve price stability that might have mitigated their harmful over- all effects on agricultural incentives. Countries tended to stabilize prices more during the period of volatile world prices in 1971-84 than when the world prices were more stable in 1960-70. Producer prices were stabilized more for importable and staple products than for exportable and nonstaple products. Variability was also reduced for staple food prices for consumers in most countries examined. These findings do nothing to support the suggestion that the harmful effects of indirect interventions in agriculture are mitigated by any substan- tial beneficial effects on price variability. The question is whether the government can provide stabilization at a lower social cost, more efficiently, than the private sector. Many developing countries either lack private insurance mechanisms (such as futures markets or forward contracts) against price risk or such mechanisms are not generally available to small farmers. The govern- ments in those countries may be better able to pool risks, so shifting some of the price risk to the governments may be socially optimal. But both the response of farmers and other private agents to govern- ment intervention and the substitutability of private and public stabi- lization efforts should also be considered. The breadth of price reforms and their credibility strongly influence the way producers respond to price changes. If price reform affects only a single product or a small group of products, the response for 228 Determinants of Agricultural Taxation; Conclusions that product will tend to be large because of changes in output, but the aggregate response will be smaller than it would be if the reform had been more general. That means that reforming the policies that create high rates of indirect taxation, which affect all of agriculture, should result in a broader, more substantial response over time than reform of direct interventions alone. Producers' expectations about whether the price change will last also will influence how much supply responds to a given price change. If producers believe that price changes are only temporary, they might, by using more variable factors, increase the output of the annual crops whose prices increased. But they would not increase investment to raise productivity or increase their output of tree crops as they might if they believed the new prices would last, because they would not expect a higher return on their investment. The response would therefore be much more limited. Many factors affect the credibility or expected longevity of the new set of prices that result from reform. If the prices of products subject to price control are increased once and for all in an inflationary envi- ronment, producers would expect the nominal price increase to erode in real terms, and they would have little incentive to increase output. Their response would be different if price controls were removed entirely. Macroeconomic uncertainty also dampens the response to price reforms because it creates instability in relative incentives in the econ- omy. When inflation in high and the real exchange rate is substan- tially overvalued, the rate of indirect taxation is high and highly vari- able. Removing direct price interventions may not lead to a large supply response under these conditions, because of the general uncertainty and because the impact of nominal price changes on rela- tive prices is hard to discern. To generate a substantial response by producers under these conditions, agricultural price reform would need to be accompanied by broader reform of macroeconomic policies. Institutional and infrastructural weaknesses can also limit the response of output to price reform. Producers are unlikely to respond strongly to a price increase given by a marketing board that lacks the infrastructure and human resources to efficiently market and trans- port inputs and output or the financial resources to buy the larger output generated by the price increase. To make reform credible, other measures would be needed to strengthen and liberalize market- ing and transport. Sometimes a poor response to price reform is related to the dis- parity between official prices and true incentives. Official prices often do not reflect marginal incentives because some transactions are car- ried out at other prices. In Kenya, when the marketing board was unable to buy the entire output of beans produced in response to its Summary and Concluding Remarks 229 official offer of higher prices, farmers had to wait months to get paid or had to sell some of their output on the domestic market at a lower price. In Ghana, the Cocoa Marketing Board often delayed payments because of corruption or lack of funds, issuing promissory notes or checks that farmers had difficulty cashing locally. Under these condi- tions, farmers may be responding to the lower market price rather than to the higher official price. The weak response to an increase in the official price does not mean that farmers do not respond to higher incentives-only that they realize that higher official prices do not always improve the incentives. The failure of poorer farmers to respond to input subsidies is also sometimes viewed as an indication of a low responsiveness to improved incentives in the agriculture sector. Often, however, small- scale farmers fail to take advantage of input or credit subsidies not because they are uninterested but because poorly designed programs make it difficult for them to do so, while allowing larger, better con- nected farmers or agricultural processors to capture the subsidies. When small farmers do manage to receive subsidized fertilizer, they often receive it at the wrong time or receive a variety unsuited to their soils or crops. Similarly, farmers are often reluctant to use "subsi- dized" credit because of the high cost (in travel and time-consuming administrative requirements) of obtaining the funds. Sometimes supply fails to respond as expected to, for example, subsidized inputs, because the subsidies are not really reaching those the authorities thought they were reaching or intended to reach. In Brazil, subsidized credit went not to its intended recipients but to new "farmers," attracted to agriculture by the credit subsidy, which they used for land speculation. In Egypt, farmers often evaded con- trols intended to ensure that subsidized inputs were used in major food crops, using them instead in more profitable crops. Unsurprisingly, the response of the major food crops was considera- bly less than expected. Under any of these conditions, the supply response is likely to fall considerably short of its potential. But if measures were taken to eliminate some of the barriers to improved input use and to ensure that prices reflected opportunities at the margin, the response to improved incentives would almost certainly be considerably larger than if prices did not reflect true opportunities and scarcities. What Happens If Many Developing Countries Reform Simultaneously? The assumption throughout this study is that international prices for fann products are given. But what happens if many countries stop taxing agriculture? Could that push down world prices? The macroeconomic stabilization and trade liberalization under the 230 Determinants of Agricultural Taxation; Conclusions structural adjustment programs of the 1980s have undoubtedly reduced the indirect taxation of agriculture, with direct interventions for importables positive and increasing in most developing countries. The total (direct and indirect) interventions for importables in devel- oping countries are probably close to zero or slightly positive. Remov- ing these interventions would thus have little impact on the world price. And for importables that are taxed in most developing coun- tries, removing the total intervention could raise output and lower consumption. This could lead to a lower world price and thus be a plus for developing countries as a whole. Removing the tax on agriculture will undoubtedly lead to an expan- sion of output and to a reduction in consumption for many export- ables that have been subject to a high tax. But the simultaneous removal of the total tax by all developing countries will dramatically suppress the world price only for a few traditional tropical products, such as coffee, cocoa, and palm oil, and punish the farmers growing those tree crops. For other exports, the effect would be much smaller because the extent of intervention, especially the indirect intervention, has fallen in recent years. The effect would also be smaller for four additional reasons. First, the share of developing countries in world trade of those products is significantly smaller than for tropical products. Sec- ond, the removal of trade barriers by developing countries will result in new markets and increased trade among developing countries and will constitute an additional source of demand. Third, economic reform, especially industrial trade liberalization, by most developing countries will put political pressure on OECD countries to reciprocate by liberalizing agricultural trade. Such reciprocity could bring sub- stantial income to developing countries. Fourth, another source of income related to removal of interventions is the potential for diver- sification. Agricultural price intervention has taxed exportables in relation to importables. Removing these interventions will reallocate resources within agriculture toward the higher valued products and toward new products, including nontraditional exports. Countries that have opened their economies-such as Brazil, Chile, Kenya, Thailand, and Turkey-have diversified their agricultural exports. These diversification possibilities are likely to reduce the adverse price effects of reform. For commodities where the problem of demand constraint is rele- vant, the implications for policy are easily drawn from economic the- ory. The theory of second best states that any distortion should be attacked at the source. So, any country that has market power on the world market for one of its exports (for example, cocoa in the C6te d'Ivoire) may want to impose an optimal export tax or quota. This would reduce output and exports and increase the world price for the product. With an optimal export tax or quota, growth in output Summary and Concluding Remarks 231 would not reduce welfare even if the price falls. Of course, an over- valued real exchange rate and industrial protection also tax agriculture-indirectly-reducing output and pushing up the world price. But such indirect taxation results in price distortions among the rest of agriculture, industry, and the nontradable sector. So the policy prescription is to remove indirect taxation, usually the largest element of the total taxation of agriculture, and use direct taxation (an export quota or tax) to deal with international market power. For those products for which developing countries can affect the world price, removing the indirect taxation on agriculture would thus depress the world price, but an appropriate export tax or quota could offset this effect. And although this would lower the domestic price, the export tax or quota would give farmers the incentive to diversify their output. It is true, however, that if all countries liberalize, they will have to deal with the complex issues of strategic interaction. That is, the appropriate export tax or quota in one country will depend on the level of the export tax or quota set by the other countries compet- ing in the world market. Reflections on Policy Implications The findings presented in this volume provide a solid base for exam- ining the implications for agricultural pricing policy in developing countries. Major implications that emerge are: * In order to accelerate agricultural growth-and overall growth- and reduce poverty, taxation of agriculture should be removed. * In order to reduce the discrimination against agriculture, the protection of industry should be low and the real exchange rate misalignment should be eliminated. The bulk of discrimination against agriculture came from these indirect price interventions. * In those countries where the revenues from direct price interven- tions in agriculture still provide a significant share of government revenues, agricultural price reform should be accompanied by a general tax reform in order to replace the revenues lost from removing taxes on agriculture. Converting quantitative restric- tions to tariffs will increase revenues and will reduce the fiscal constraint to price reform. * Direct interventions-though small on average (about 8 percent)-resulted in a substantial anti-export bias in agriculture, taxing exportables and protecting importables, with significant distortions in relative prices (a tax on exportables relative to importables of about 30 percent) and in resource allocation within agriculture. Removing the anti-export bias should result in a significant improvement in resource allocation and in large income gains. 232 Determinants of Agricultural Taxation; Conclusions * In order for the reform of agricultural price interventions to be effective and for its impact to be transparent, it is essential that agricultural quotas, licenses, and state trading mechanisms be dismantled, with possible tariffication during the transition. * Social objectives-such as protecting the poor-should not be used to justify selective agricultural price interventions, because agricultural growth and incomes will suffer and because the impact on the urban poor is small in the short run and unknown in the long run. * The output response to agricultural price and trade reform will be enhanced and accelerated if reform in related sectors-land and credit markets, and transport and communication-are launched simultaneously where appropriate. * Domestic price stabilization of key agricultural products has been and remains an important objective of price polices. The experi- ence with price stabilization schemes suggests they are costly and often subverted by pressure groups. Part of the unfinished business is the evaluation of the relative merits of various approaches and design of efficient risk management instru- ments. * In this volume, we have emphasized the importance of indirect interventions (industrial protection and exchange rate misalign- ment) for agriculture. There are at least three reasons why indi- rect interventions are an essential component for a successful agricultural price reform. First, they constituted the bulk (three quarters) of the taxation of agriculture. Second, the removal of indirect interventions should have no adverse impact on domes- tic agricultural price stability. And third, for those countries where agriculture's share of government revenue is still impor- tant, and where the motivation for agricultural trade barriers was generating fiscal revenue, a reform of indirect interventions (including tariffication of quotas) may be easier from the fiscal viewpoint. Appendix: Direct and Total Nominal Protection Rates, by Country, Product, and Year Note: NPRd stands for "adjusted" direct nominal protection rate (see the appendix to chapter 2 for a definition). NPRT stands for total nominal protection rate. Source: Country studies. 233 COUNTRY: ARGENTINA YEAR BEEF COTTON SORGHUM SOYBEAN SUNFLOWER WHEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -17.1 -31.4 -48.5 -25.9 -43.1 -25.1 -42.2 0.0 -17.1 0.0 -17.1 -30.9 -48.0 1961 -26.5 -28.1 -54.6 -12.0 -38.5 -26.6 -53.1 0.0 -26.5 0.0 -26.5 -17.3 -43.7 1962 -20.7 -32.0 -52.7 -1.8 -22.5 5.3 -15.4 0.0 -20.7 0.0 -20.7 -17.3 -38.0 1963 -13.7 -27.7 -41.5 4.1 -9.6 -6.3 -20.0 0.0 -13.7 0.0 -13.7 -11.5 -25.2 1964 -16.9 -14.7 -31.6 0.8 -16.1 -15.9 -32.9 0.0 -16.9 0.0 -16.9 -1.5 -18.4 1965 -28.7 -21.9 -50.6 -6.2 -34.9 -5.0 -33.7 0.0 -28.7 0.0 -28.7 -13.4 -42.1 1966 -31.5 -12.2 -43.7 7.2 -24.3 -15.2 -46.7 0.0 -31.5 0.0 -31.5 -4.7 -36.2 1967 -19.6 -25.0 -44.6 -27.7 -47.3 -23.2 -42.7 0.0 -19.6 0.0 -19.6 6.7 -12.9 1968 -23.6 -23.9 -47.5 -15.6 -39.2 -14.8 -38.4 0.0 -23.6 0.0 -23.6 -20.6 -44.2 1969 -27.0 -22.2 -49.2 -0.7 -27.8 -3.1 -30.1 0.0 -27.0 0.0 -27.0 -13.0 -40.0 1970 -20.4 -9.0 -29.4 -12.0 -32.4 -17.5 -37.9 0.0 -20.4 0.0 -20.4 -14.6 -35.0 1971 -18.5 -1.3 -19.7 -14.0 -32.4 -21.2 -39.7 0.0 -18.5 0.0 -18.5 -10.0 -28.4 1972 0.2 -23.2 -23.0 -28.7 -28.5 -38.0 -37.8 0.0 0.2 0.0 0.2 -49.3 -49.1 1973 -5.7 -27.5 -33.2 -23.5 -29.2 -26.9 -32.6 0.0 -5.7 0.0 -5.7 -40.7 -46.4 1974 -6.0 -27.4 -33.4 -26.1 -32.1 -18.7 -24.7 0.0 -6.0 0.0 -6.0 -60.4 -66.4 1975 -20.0 -29.2 -49.2 -36.5 -56.4 -28.1 -48.1 0.0 -20.0 0.0 -20.0 -48.2 -68.1 1976 2.2 -1.8 0.3 -59.6 -57.4 -54.5 -52.4 0.0 2.2 0.0 2.2 -41.4 -39.3 1977 -5.7 -21.5 -27.2 -19.0 -24.8 -23.2 -29.0 -16.6 -22.3 -21.8 -27.6 -12.4 -18.1 Z 1978 -22.8 -2.3 -25.1 -7.7 -30.5 -9.5 -32.3 -11.9 -34.6 -25.7 -48.5 -15.0 -37.8 1979 -32.8 8.0 -24.7 -8.1 -40.9 18.6 -14.2 -8.1 -40.9 -14.6 -47.4 -8.2 -40.9 1980 -49.7 1.0 -48.7 3.1 -46.6 -4.6 -54.3 -2.4 -52.1 -12.2 -61.9 -4.9 -54.6 1981 -48.6 -17.0 -65.6 -8.2 -56.9 -10.0 -58.6 -6.0 -54.6 -4.6 -53.3 -1.6 -50.2 1982 -26.6 -20.1 -46.7 -11.6 -38.2 -12.7 -39.3 -12.0 -38.6 -20.4 -47.0 -10.7 -37.3 1983 -21.6 -21.9 -43.5 -22.8 -44.5 -22.3 -43.9 -23.7 -45.3 -27.3 -48.9 -22.5 -44.1 1984 -30.0 -10.0 -40.0 -16.7 -46.8 -21.5 -51.5 -18.3 -48.3 -17.1 -47.1 -17.4 -47.4 1985 -28.1 -7.1 -35.2 -16.3 -44.3 -22.5 -50.5 -19.4 -47.5 -19.2 -47.2 -21.4 -49.5 AVERAGE: 1960S -22.5 -23.9 -46.5 -7.8 -30.3 -13.0 -35.5 0.0 -22.5 0.0 -22.5 -12.3 -34.9 1970-S -12.9 -13.5 -26.5 -23.5 -36.5 -21.9 -34.9 -3.7 -16.6 -6.2 -19.2 -30.0 -43.0 19805S -34.1 -12.5 -46.6 -12.1 -46.2 -15.6 -49.7 -13.6 -47.7 -16.8 -50.9 -13.1 -47.2 WHOLE PERIOD -21.5 -17.3 -38.8 -14.8 -36.3 -17.0 -38.5 -4.6 -26.1 -6.3 -27.8 -19.3 -40.8 COUNTRY: BRAZIL YEAR COTTON CORN SOYBEAN RICE WNEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1966 -12.8 -12.6 -25.3 33.0 20.2 0.0 -12.8 0.0 -12.8 38.0 25.3 1967 -13.7 -17.9 -31.6 42.5 28.7 0.0 -13.7 0.0 -13.7 34.9 21.2 1968 -16.7 -18.4 -35.1 20.0 3.3 0.0 -16.7 0.0 -16.7 31.7 15.0 1969 -12.5 -13.7 -26.1 29.3 16.8 0.0 -12.5 12.5 0.0 38.1 25.6 1970 -16.4 -6.9 -23.3 54.8 38.4 -1.2 -17.6 80.8 64.5 55.5 39.1 1971 -20.5 -7.4 -27.9 53.4 32.9 -6.6 -27.1 135.8 115.3 28.3 7.8 1972 -18.9 -14.1 -33.0 59.6 40.7 -13.5 -32.4 89.7 70.8 19.0 0.1 1973 -16.7 -6.7 -23.4 45.0 28.2 -22.7 -39.5 15.7 -1.1 -10.6 -27.3 1974 -22.3 -6.8 -29.1 2.7 -19.6 -15.0 -37.3 -2.4 -24.7 -22.8 -45.1 1975 -19.5 2.2 -17.4 8.8 -10.8 -11.8 -31.4 10.9 -8.6 8.2 -11.3 1976 -16.0 18.2 2.3 35.7 19.8 -24.7 -40.7 -0.2 -16.1 10.8 -5.2 1977 -14.8 6.5 -8.3 56.5 41.7 -22.3 -37.1 -9.3 -24.1 59.7 44.9 1978 -24.7 -0.8 -25.5 23.0 -1.7 -24.0 -48.7 11.6 -13.1 35.2 10.5 1979 -26.4 -17.0 -43.4 -9.7 -36.1 -13.6 -40.0 6.5 -19.9 4.3 -22.1 1980 -23.9 2.0 -21.9 3.3 -20.5 -25.5 -49.4 -5.6 -29.5 -19.2 -43.1 1981 -15.0 4.7 -10.3 30.4 15.5 1.9 -13.1 -7.3 -22.3 3.7 -11.2 1982 -20.2 3.0 -17.2 7.3 -12.8 -35.2 -55.4 6.9 -13.3 12.9 -7.3 1983 -8.7 13.8 5.1 5.4 -3.3 -4.9 -13.6 -2.4 -11.1 -10.6 -19.3 AVERAGE: 1960'S -13.9 -15.6 -29.5 31.2 17.2 0.0 -13.9 3.1 -10.8 35.7 21.8 1970'S -19.6 -3.3 -22.9 33.0 13.3 -15.5 -35.2 33.9 14.3 18.8 -0.9 1980'S -16.9 5.9 -11.1 11.6 -5.3 -15.9 -32.9 -2.1 -19.0 -3.3 -20.2 WHOLE PERIOD -17.8 -4.0 -21.7 27.8 10.1 -12.2 -29.9 19.1 1.3 17.6 -0.1 CMJNTRY: CHILE YEAR APPLES CATTLE GRAPES MIlK WHEAT INDIRECT NPRd NPRI NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -19.8 0.0 -19.8 -16.0 -35.9 0.0 -19.8 148.9 129.0 15.3 -4.6 1961 -22.3 0.0 -22.3 -10.9 -33.1 0.0 -22.3 144.4 122.1 5.0 -17.3 1962 -30.0 0.0 -30.0 -0.7 -30.7 0.0 -30.0 130.1 100.1 1.4 -28.5 1963 -28.8 25.7 -3.2 1.4 -27.4 26.7 -2.1 132.1 103.3 -3.4 -32.2 1964 -28.1 13.5 -14.6 -12.2 -40.3 16.2 -11.9 133.6 105.5 -6.5 -34.6 1965 -26.0 11.6 -14.4 -16.3 -42.3 11.3 -14.8 55.8 29.8 26.0 0.0 1966 -24.1 1.5 -22.6 -19.7 -43.8 1.5 -22.6 31.0 6.9 62.1 38.0 1967 -23.6 31.2 7.7 -19.9 -43.4 32.8 9.2 21.6 -2.0 12.4 -11.1 1968 -24.6 30.8 6.2 -22.6 -47.2 29.3 4.7 21.3 -3.3 10.3 -14.3 1969 -13.0 32.2 19.2 -21.8 -34.7 33.8 20.8 20.4 7.4 8.6 -4.4 1970 -15.3 28.2 13.0 -16.1 -31.4 29.8 14.5 39.9 24.6 16.1 0.9 1971 -21.2 40.6 19.4 -12.6 -33.8 32.2 11.0 11.8 -9.4 19.7 -1.5 1972 -33.5 52.2 18.7 -24.6 -58.1 42.5 8.9 2.8 -30.8 8.2 -25.3 1973 -39.0 19.3 -19.7 -19.5 -58.5 30.1 -8.9 -0.6 -39.6 -23.1 -62.1 1974 -27.8 9.8 -18.0 -13.0 -40.8 22.8 -5.0 10.8 -17.0 -5.3 -33.2 1975 -27.0 0.7 -26.3 0.0 -27.0 3.0 -24.0 27.0 0.0 13.9 -13.1 1976 -8.6 0.0 -8.6 0.0 -8.6 0.0 -8.6 25.8 17.2 -12.6 -21.2 1977 -11.7 0.0 -11.7 0.0 -11.7 0.0 -11.7 2.7 -8.9 66.6 55.0 1978 -11.6 0.0 -11.6 20.3 8.7 0.0 -11.6 10.9 -0.7 9.8 -1.8 1979 -8.7 0.0 -8.7 0.0 -8.7 0.0 -8.7 56.0 47.3 -10.5 -19.2 1980 -9.5 0.0 -9.5 7.2 -2.3 0.0 -9.5 19.9 10.4 0.9 -8.6 1981 -14.1 0.0 -14.1 0.0 -14.1 0.0 -14.1 7.5 -6.7 9.5 -4.6 1982 -10.9 0.0 -10.9 0.0 -10.9 0.0 -10.9 6.7 -4.2 12.2 1.3 1983 -10.4 0.0 -10.4 0.0 -10.4 0.0 -10.4 10.0 -0.5 8.9 -1.5 AVERAGE: 1960's -24.0 14.7 -9.4 -13.9 -37.9 15.2 -8.9 83.9 59.9 13.1 -10.9 1970'S -20.4 15.1 -5.3 -6.5 -27.0 16.0 -4.4 18.7 -1.7 8.3 -12.1 1980'S -11.2 0.0 -11.2 1.8 -9.4 0.0 -11.2 11.0 -0.2 7.9 -3.4 WHOLE PERICO -20.4 12.4 -8.0 -8.2 -28.6 13.0 -7.4 44.6 24.2 10.2 -10.2 COXJNTRY: COLOMBIA YEAR COFFEE COTTON RICE WHEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -15.8 -9.7 -25.5 8.3 -7.5 36.2 20.4 15.4 -0.4 1961 -23.0 3.9 -19.1 -0.9 -23.9 83.9 60.9 20.3 -2.7 1962 -27.2 4.5 -22.7 -6.5 -33.7 20.9 -6.3 11.1 -16.1 1963 -26.8 -9.1 -35.9 4.8 -22.0 6.3 -20.5 0.5 -26.3 1964 -32.5 -10.2 -42.7 8.2 -24.3 34.2 1.7 25.0 -7.5 1965 -26.4 -11.1 -37.5 -5.7 -32.1 59.7 33.3 35.0 8.6 1966 -28.2 -16.7 -44.9 14.6 -13.6 37.4 9.2 19.0 -9.2 1967 -18.1 -17.9 -36.0 3.3 -14.8 31.4 13.3 18.2 0.1 1968 -20.5 -12.7 -33.2 3.6 -16.9 24.5 4.0 20.6 0.1 1969 -20.5 -8.6 -29.1 1.6 -18.9 8.3 -12.2 21.6 1.1 1970 -26.2 -10.3 -36.5 3.0 -23.2 1.1 -25.1 14.6 -11.6 1971 -29.5 -7.5 -37.0 -3.9 -33.4 -3.0 -32.5 1.0 -28.5 1972 -24.6 -8.0 -32.6 -4.3 -28.9 -18.4 -43.0 13.2 -11.4 1973 -24.4 -7.4 -31.8 -4.8 -29.2 -37.2 -61.6 -18.4 -42.8 1974 -22.4 -6.5 -28.9 -7.6 -30.0 -41.3 -63.7 -24.5 -46.9 1975 -14.0 -5.9 -19.9 0.9 -13.1 -36.8 -50.8 -4.0 -18.0 1976 -19.2 -11.1 -30.3 2.9 -16.3 -22.1 -41.3 -8.5 -27.7 1977 -24.8 -10.0 -34.8 0.8 -24.0 3.3 -21.5 12.1 -12.7 1978 -28.0 -3.3 -31.3 0.2 -27.8 -9.9 -37.9 13.1 -14.9 1979 -29.5 -13.6 -43.1 -2.2 -31.7 -0.3 -29.8 1.8 -27.7 1980 -22.3 -4.0 -26.3 2.4 -19.9 -7.9 -30.2 -1.5 -23.8 1981 -27.7 -8.7 -36.4 6.1 -21.6 -5.0 -32.7 4.6 -23.1 1982 -34.0 -5.3 -39.3 10.6 -23.4 48.7 14.7 15.5 -18.5 1983 -38.9 -2.0 -40.9 1.2 -37.7 27.4 -11.5 18.7 -20.2 AVERAGE: 1960-S -23.9 -8.8 -32.7 3.1 -20.8 34.3 10.4 18.7 -5.2 19701S -24.3 -8.4 -32.6 -1.5 -25.7 -16.5 -40.7 0.0 -24.2 1980'S -30.7 -5.0 -35.7 5.1 -25.6 15.8 -14.9 9.3 -21.4 WHOLE PERIOD -25.2 -8.0 -33.2 1.5 -23.7 10.1 -15.1 9.4 -15.8 238 Determinants of Agricultural Taxation; Conclusions COUNTRY: COTE D-IVOIRE YEAR COCOA COFFEE RICE INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -18.2 -22.0 -40.1 -13.3 -31.4 29.8 11.6 1961 -17.8 -28.2 -46.0 -26.4 -44.3 2.8 -15.0 1962 -17.1 -25.4 -42.4 -27.7 -44.8 2.9 -14.2 1963 -17.3 -24.6 -42.0 -24.8 -42.2 5.9 -11.4 1964 -13.7 -27.5 -41.2 -27.4 -41.1 0.0 -13.7 1965 -18.1 -22.5 -40.6 -30.1 -48.2 9.1 -9.0 1966 -18.6 -20.0 -38.6 -29.6 -48.2 -11.6 -30.3 1967 -23.0 -30.6 -53.6 -29.2 -52.2 -0.7 -23.7 1968 -13.7 -43.3 -57.0 -31.7 -45.4 -10.6 -24.3 1969 -11.6 -52.2 -63.8 -30.1 -41.7 19.6 8.1 1970 -15.3 -39.8 -55.1 -37.3 -52.7 56.4 41.1 1971 -20.9 -27.4 -48.3 -36.3 -57.2 115.1 94.2 1972 -22.2 -24.9 -47.1 -28.6 -50.8 67.4 45.2 1973 -24.6 -26.9 -51.5 -23.7 -48.3 91.3 66.6 1974 -15.0 -29.3 -44.3 -23.0 -38.0 3.5 -11.5 1975 -27.0 -21.2 -48.2 -19.8 -46.8 9.4 -17.6 1976 -15.2 -37.5 -52.6 -41.2 -56.3 12.4 -2.7 1977 -31.9 -37.1 -68.9 -44.6 -76.5 42.0 10.2 1978 -45.0 -31.3 -76.3 -26.6 -71.6 36.7 -8.3 1979 -46.1 -26.9 -73.0 -24.3 -70.3 45.1 -1.0 1980 -39.7 -25.5 -65.2 -28.0 -67.7 15.7 -24.1 1981 -23.1 -19.3 -42.5 -25.1 -48.2 9.9 -13.3 1982 -40.7 -12.1 -52.8 -15.6 -56.3 14.6 -26.1 AVERAGE: 1960'S -16.9 -29.6 -46.5 -27.0 -43.9 4.7 -12.2 1970'S -21.5 -30.5 -52.0 -31.8 -53.3 49.7 28.2 1980-S -38.9 -23.0 -62.0 -23.9 -62.8 24.4 -14.5 WHOLE PERIOD -23.3 -28.5 -51.8 -28.0 -51.3 24.6 1.3 Appendix: Direct and Total Nominal Protection Rates 239 COUNTRY: DOMINICAN REPUBLIC YEAR COFEE RICE SUGAR INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT 1966 -27.4 1.5 -25.9 10.9 -16.5 -26.1 -53.5 1967 -26.2 0.0 -26.2 11.8 -14.4 -31.7 -57.9 1968 -26.6 2.2 -24.4 31.6 4.9 -32.3 -58.9 1969 -26.2 0.7 -25.5 26.6 0.3 -30.2 -56.5 1970 -26.3 -17.7 -44.0 17.0 -9.3 -28.8 -55.0 1971 -27.2 -2.9 -30.1 16.0 -11.1 -21.9 -49.0 1972 -22.4 -11.6 -34.0 3.1 -19.2 -28.0 -50.3 1973 -21.7 -12.5 -34.2 -26.6 -48.3 -33.7 -55.4 1974 -18.4 -4.1 -22.5 -27.8 -46.1 -44.1 -62.5 1975 -12.9 -5.2 -18.1 9.6 -3.3 -56.6 -69.5 1976 -16.9 -16.6 -33.5 35.7 18.9 -35.7 -52.6 1977 -15.8 -17.7 -33.5 20.2 4.4 -25.3 -41.0 1978 -20.3 -2.4 -22.7 18.3 -2.0 0.0 -20.3 1979 -21.7 -18.0 -39.7 28.2 6.5 -8.6 -30.3 1980 -22.2 -3.1 -25.3 6.2 -16.0 -31.9 -54.1 1981 -18.0 5.7 -12.2 5.7 -12.2 -42.6 -60.6 1982 -21.1 -21.3 -42.4 72.6 51.5 -4.7 -25.8 1983 -22.8 -20.1 -42.9 69.5 46.6 17.0 -5.9 1984 -23.0 -23.9 -46.9 -1.5 -24.6 -19.2 -42.3 1985 -8.8 -68.4 -77.2 54.7 45.9 -57.4 -66.3 AVERAGE: 1960'S -26.6 1.1 -25.5 20.2 -6.4 -30.1 -56.7 1970-S -20.3 -10.9 -31.2 9.4 -11.0 -28.3 -48.6 1980'S -19.3 -21.8 -41.2 34.5 15.2 -23.2 -42.5 WHOLE PERIOD -21.3 -11.8 -33.1 19.1 -2.2 -27.1 -48.4 COJNTRY: EGYPT YEAR COTTOM MAIZE RICE SUGAR WHEAT IMDIRECT NPRd NPRT NPRd NPRT MPRd NPRT NPRd NPRT NPRd NPRT 1960 -39.8 -13.8 -53.7 4.9 -34.9 -16.9 -56.7 157.1 117.3 15.6 -24.2 1961 -34.6 -15.7 -50.3 -1.0 -35.7 -22.9 -57.5 196.1 161.5 20.9 -13.7 1962 -25.9 -14.1 -40.0 -4.5 -30.4 -37.8 -63.7 182.3 156.4 9.6 -16.3 1963 -23.5 -25.3 -48.7 -21.1 -44.6 -42.9 -66.3 12.2 -11.2 -15.1 -38.6 1964 -27.2 -16.0 -43.2 -13.8 -41.0 -45.1 -72.3 44.4 17.2 -24.8 -52.0 1965 -30.9 -22.8 -53.7 -17.8 -48.7 -40.8 -71.7 196.9 165.9 -12.2 -43.1 1966 -34.3 -18.4 -52.7 -4.2 -38.4 -33.5 -67.8 240.6 206.4 -4.2 -38.4 1967 -5.6 -44.4 -50.0 7.5 1.9 -45.3 -50.9 303.0 297.4 1.5 -4.1 1968 -28.3 -17.9 -46.2 -8.4 -36.8 -41.6 -69.9 235.1 206.8 -1.4 -29.8 1969 -29.3 -22.6 -51.9 -2.8 -32.1 -35.3 -64.7 125.1 95.8 3.0 -26.3 1970 -33.9 -21.2 -55.0 1.3 -32.5 -21.8 -55.7 121.7 87.8 21.5 -12.3 1971 -36.9 -20.2 -57.1 -4.7 -41.5 18.3 -18.6 87.1 50.3 0.0 -36.9 1972 -30.6 -18.7 -49.4 2.6 -28.0 -22.9 -53.5 33.3 2.7 0.0 -30.6 1973 5.8 -65.6 -59.8 -3.1 2.8 -68.8 -63.0 42.3 48.1 -10.2 -4.4 1974 -1.0 -62.4 -63.4 -7.1 -8.1 -86.2 -87.1 -28.7 -29.7 -56.9 -57.8 1975 -9.3 -49-9 -59.2 -32.2 -41.5 -74.4 -83.7 0.0 -9.3 -36.5 -45.7 1976 -18.7 -29.3 -48.0 -31.8 -50.5 51.2 32.5 65.9 47.2 -27.4 -46.1 1977 -18.3 -40.0 -58.3 11.1 -7.2 31.0 12.7 128.3 110.0 -9.0 -27.3 1978 -29.1 -15.6 -44.7 2.8 -26.2 -30.5 -59.6 134.1 105.0 -4.0 -33.1 1979 -14.5 -44.4 -59.0 -5.4 -19.9 -59.0 -73.5 13.7 -0.9 -21.9 -36.4 1980 -5.7 -42.5 -48.1 -9.1 -14.8 -59.4 -65.1 -41.5 -47.2 -41.9 -47.5 1981 -16.7 -27.5 -44.2 -42.9 -59.5 -55.8 -72.5 -6.7 -23.3 -41.4 -58.1 sl 1982 -19.5 -9.7 -29.2 -3.2 -22.7 -20.1 -39.6 65.2 45.7 -22.2 -41.7 1983 -12.4 -6.1 -18.5 31.0 18.7 -19.3 -31.7 71.0 58.6 -5.8 -18.2 1984 -14.5 -19.7 -34.2 18.1 3.6 -10.3 -24.8 127.4 112.8 3.6 -10.9 AVERAGE: 1960's -27.9 -21.1 -49.0 -6.1 -34.1 -36.2 -64.1 169.3 141.3 -0.7 -28.7 1970-S -16.9 -38.5 -55.4 -7.5 -24.5 -26.8 -43.7 52.9 35.9 -18.4 -35.4 1980's -13.7 -21.1 -34.8 -1.2 -15.0 -33.0 -46.7 43.1 29.3 -21.5 -35.3 WHOLE PERIOD -21.4 -27.3 -48.7 -5.3 -26.7 -31.6 -53.0 100.2 78.9 -10.4 -31.7 Appendix: Direct and Total Nominal Protection Rates 241 COJNTRY: GHANA YEAR COCOA MAIZE RICE INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT 1958 -17.4 -43.8 -61.2 -46.3 -63.6 -0.8 -18.2 1959 -21.0 -34.8 -55.8 -45.0 -66.0 0.0 -21.0 1960 -26.9 -19.7 -46.7 -51.1 -78.1 -7.3 -34.2 1961 -31.9 -14.3 -46.2 -36.8 -68.7 -6.1 -38.1 1962 -29.4 -18.3 -47.8 -31.7 -61.2 -9.9 -39.3 1963 -33.7 -23.9 -57.6 74.9 41.3 67.6 34.0 1964 -39.5 -12.7 -52.2 -2.4 -41.9 63.5 24.1 1965 -56.5 -6.5 -63.0 84.4 27.9 55.7 -0.8 1966 -54.0 -16.6 -70.5 115.1 61.1 49.7 -4.3 1967 -34.0 -33.6 -67.7 19.8 -14.2 25.7 -8.3 1968 -31.0 -41.4 -72.4 -24.8 -55.8 -16.6 -47.6 1969 -30.6 -43.0 -73.6 2.1 -28.5 9.7 -20.9 1970 -32.2 -34.6 -66.8 268.0 235.8 44.1 11.9 1971 -33.0 -34.2 -67.2 49.6 16.6 75.1 42.1 1972 -21.9 -35.9 -57.8 -49.2 -71.1 115.6 93.7 1973 -20.3 -47.0 -67.3 190.6 170.3 41.5 21.2 1974 -22.4 -49.6 -72.1 11.6 -10.8 -10.1 -32.5 1975 -32.4 -32.4 -64.9 37.2 4.8 94.0 61.6 1976 -51.6 -32.9 -84.5 170.9 119.3 178.6 127.0 1977 -76.4 -12.7 -89.1 74.3 -2.1 -294.5 -370.9 1978 -79.9 -14.1 -94.0 23.5 -56.3 -126.5 -206.4 1979 -72.5 -11.8 -84.3 3.3 -69.2 -108.7 -181.2 1980 -81.8 8.6 -73.2 172.4 90.7 -131.5 -213.3 1981 -92.4 -464.0 -556.4 -146.0 -238.4 -21.9 -114.3 1982 -93.7 -2.4 -96.2 -46.6 -140.3 -26.9 -120.6 1983 -96.3 -1.4 -97.8 -53.1 -149.5 -17.3 -113.7 1984 -71.9 -6.8 -78.6 74.6 2.7 -504.1 -576.0 AVERAGE: 1960'S -36.8 -23.0 -59.8 14.9 -21.8 23.2 -13.5 1970-S -44.3 -30.5 -74.8 78.0 33.7 0.9 -43.4 1980'S -87.2 -93.2 -180.4 0.3 -87.0 -140.4 -227.6 WHOLE PERIOD -49.8 -40.1 -89.9 37.2 -12.6 -18.4 -68.3 COJNTRY: KOREA YEAR BARLEY BEEF PORK RICE SOYBEAN INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -46.0 53.8 7.8 45.2 -0.8 34.9 -11.1 45.8 -0.2 84.8 38.8 1961 -36.1 78.3 42.1 -3.3 -39.4 -1.0 -37.1 17.9 -18.3 7.3 -28.8 1962 -38.1 10.5 -27.6 2.7 -35.3 26.6 -11.5 9.0 -29.1 9.5 -28.6 1963 -44.6 66.0 21.5 5.8 -38.8 28.9 -15.7 45.7 1.1 51.8 7.2 1964 -34.2 36.9 2.6 -14.7 -49.0 5.7 -28.5 11.3 -22.9 33.9 -0.3 1965 -27.1 21.3 -5.8 -1.5 -28.5 18.0 -9.1 -2.8 -29.8 24.6 -2.4 1966 -31.7 -19.5 -51.2 -1.8 -33.5 5.7 -26.0 -4.2 -35.9 30.0 -1.7 1967 -32.5 -14.7 -47.1 13.0 -19.4 27.7 -4.8 7.2 -25.2 61.7 29.3 1968 -34.7 4.0 -30.7 30.1 -4.6 64.9 30.2 3.3 -31.3 18.7 -15.9 1969 -33.5 19.1 -14.4 23.6 -9.9 25.9 -7.7 9.9 -23.6 39.0 5.5 1970 -31.8 3.3 -28.5 25.7 -6.1 52.2 20.4 14.2 -17.6 76.0 44.2 1971 -33.1 43.8 10.8 26.9 -6.1 62.7 29.6 48.1 15.1 38.4 5.4 1972 -26.0 67.7 41.7 38.4 12.4 34.3 8.2 81.5 55.4 61.7 35.7 1973 -22.6 13.4 -9.2 15.5 -7.1 51.4 28.8 14.7 -7.9 21.8 -0.8 1974 -27.4 -3.7 -31.1 18.9 -8.5 10.9 -16.6 0.5 -26.9 18.4 -9.0 1975 -25.0 -3.7 -28.8 23.8 -1.3 34.6 9.5 8.2 -16.8 29.3 4.3 1976 -16.9 24.3 7.4 70.8 53.9 47.0 30.1 84.8 67.9 73.8 56.9 1977 -14.6 97.1 82.6 101.3 86.8 35.8 21.2 122.2 107.6 86.4 71.8 1978 -11.5 122.3 110.8 175.7 164.1 21.2 9.7 84.1 72.6 136.8 125.3 1979 -17.1 129.8 112.7 100.3 83.2 0.9 -16.2 154.5 137.5 54.8 37.7 1980 -17.3 82.7 65.4 62.2 44.8 15.0 -2.4 112.5 95.1 150.3 133.0 1981 -15.0 71.4 56.4 100.8 85.8 8.2 -6.8 96.6 81.6 197.8 182.8 1982 -11.2 134.8 123.6 151.2 140.0 6.9 -4.2 92.0 80.9 237.2 226.0 1983 -8.5 149.2 140.7 143.8 135.3 14.1 5.6 72.3 63.8 249.3 240.8 1984 -7.5 82.3 74.8 157.8 150.4 1.7 -5.8 60.3 52.9 219.0 211.5 AVERAGE: 1960-S -35.8 25.6 -10.3 9.9 -25.9 23.7 -12.1 14.3 -21.5 36.1 0.3 1970-S -22.6 49.4 26.8 59.7 37.1 35.1 12.5 61.3 38.7 59.7 37.1 1980'S -11.9 104.1 92.2 123.2 111.3 9.2 -2.7 86.7 74.8 210.7 198.8 WHOLE PERIOD -25.8 50.8 25.1 52.5 26.7 25.4 -0.4 47.6 21.8 80.5 54.7 COJNTRY: MALAYSIA YEAR PALM RICE RUBBER 1 RUBBER 2 INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -7.4 -7.4 -14.8 27.9 20.6 -10.2 -17.6 -14.8 -22.2 1961 -8.9 -7.3 -16.2 20.1 11.2 -8.2 -17.1 -13.7 -22.6 1962 -10.4 -7.2 -17.6 10.0 -0.5 -7.2 -17.6 -12.5 -23.0 1963 -10.2 -7.2 -17.4 13.2 3.0 -6.3 -16.5 -12.6 -22.8 1964 -8.8 -7.3 -16.1 21.9 13.1 -6.4 -15.2 -13.7 -22.5 1965 -8.1 -7.4 -15.4 23.7 15.7 -7.4 -15.4 -13.8 -21.9 1966 -8.1 -7.4 -15.5 10.9 2.8 -6.4 -14.5 -12.9 -21.0 1967 -10.8 -7.1 -17.9 -8.4 -19.2 -5.4 -16.1 -13.4 -24.2 1968 -10.3 -7.2 -17.4 -16.4 -26.7 -5.4 -15.7 -14.4 -24.6 1969 -5.8 -7.5 -13.3 -14.6 -20.4 -6.6 -12.4 -14.1 -19.9 1970 -8.8 -7.3 -16.1 3.0 -5.8 -6.4 -15.2 -15.5 -24.3 1971 -10.8 -7.1 -18.0 21.1 10.3 -5.3 -16.2 -15.2 -26.0 1972 -12.2 -7.9 -20.1 17.8 5.6 -5.3 -17.5 -16.7 -28.9 1973 -7.1 -10.2 -17.3 -3.1 -10.2 -10.2 -17.3 -17.7 -24.7 1974 -8.7 -20.1 -28.7 -20.0 -28.6 -13.7 -22.4 -20.1 -28.7 1975 -9.1 -20.0 -29.1 -15.7 -24.9 -13.6 -22.8 -15.4 -24.6 1976 -2.1 -14.7 -16.8 70.8 68.7 -20.6 -22.6 -21.5 -23.6 1977 -3.4 -20.3 -23.7 80.5 77.1 -22.2 -25.6 -23.2 -26.6 1978 -5.7 -10.4 -16.1 25.4 19.7 -25.5 -31.2 -26.4 -32.1 1979 -1.6 -9.8 -11.5 28.7 27.1 -29.5 -31.2 -31.5 -33.1 1980 -5.8 -6.6 -12.4 27.0 21.2 -27.3 -33.1 -28.3 -34.1 1981 -10.1 -5.4 -15.5 30.0 19.9 -18.9 -29.0 -19.8 -29.9 1982 -11.4 -2.7 -14.0 72.4 61.1 -10.6 -22.0 -11.5 -22.9 1983 -10.8 -1.8 -12.5 142.7 131.9 -12.5 -23.3 -13.4 -24.1 AVERAGE: 1960'S -8.9 -7.3 -16.2 8.8 0.0 -6.9 -15.8 -13.6 -22.5 1970's -6.9 -12.8 -19.7 20.8 13.9 -15.2 -22.2 -20.3 -27.3 1980'S -9.5 -4.1 -13.6 68.0 58.5 -17.3 -26.8 -18.2 -27.8 WHOLE PERIOD -8.2 -9.0 -17.2 23.7 15.5 -12.1 -20.3 -17.2 -25.3 COUNTRY: MOROCCO YEAR BARLEY SUGAR HARD WHEAT SOFT WHEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -13.4 -26.8 -40.3 . -13.4 -26.8 -40.3 -17.3 -30.7 1961 -11.4 -20.4 -31.8 . -11.4 -40.7 -52.2 -13.3 -24.7 1962 -10.3 -11.7 -21.9 - -10.3 -18.8 -29.1 -17.1 -27.3 1963 -10.8 -39.2 -50.1 41.9 31.1 -25.9 -36.7 -9.8 -20.7 1964 -11.8 -43.2 -55.0 -32.6 -44.4 -5.3 -17.1 -8.8 -20.6 1965 -11.1 -36.5 -47.5 -44.5 -55.5 5.3 -5.7 -5.3 -16.4 1966 -12.2 -37.7 -50.0 20.2 8.0 2.6 -9.6 -11.4 -23.6 1967 -10.8 2.7 -8.2 68.7 57.8 16.0 5.2 -6.2 -17.1 1968 -13.0 5.2 -7.8 112.2 99.2 9.6 -3.4 -6.1 -19.1 1969 -17.3 -40.5 -57.8 80.2 63.0 -9.1 -26.4 -3.3 -20.6 1970 -12.1 -36.0 -48.1 58.0 45.9 -10.6 -22.6 -10.6 -22.6 1971 -18.4 -36.7 -55.1 25.3 6.9 5.7 -12.6 1.6 -16.7 1972 -11.6 -32.7 -44.3 -20.3 -31.9 -10.6 -22.2 -12.4 -24.0 1973 -9.7 -29.8 -39.5 -28.0 -37.7 -56.0 -65.7 -47.8 -57.6 1974 -13.2 -26.1 -39.2 -65.1 -78.3 -41.7 -54.8 -37.3 -50.5 1975 -10.8 -17.0 -27.7 -72.3 -83.0 -20.5 -31.3 -26.8 -37.5 1976 -18.3 13.1 -5.2 -66.2 -845 27.8 9.5 -17.2 -35.5 1977 -26.6 -18.3 -45.0 -12.5 -39.1 13.2 -13.4 19.8 -6.8 1978 -28.4 23.6 -4.7 57.3 28.9 26.5 -1.9 10.7 -17.6 1979 -25.1 -6.0 -31.1 93.7 68.6 3.7 -21.3 8.2 -16.8 1980 -21.1 -22.9 -44.0 -29.2 -50.3 -15.8 -36.9 18.9 -2.1 1981 -26.4 -10.3 -36.7 -51.5 -77.9 2.9 -23.5 11.8 -14.6 1982 -28.8 12.1 -16.7 -17.1 -45.9 34.2 5.4 13.5 -15.3 1983 -28.1 -31.7 -59.7 9.4 -18.7 -5.8 -33.8 -4.3 -32.4 1984 -18.3 -13.1 -31.3 11.4 -6.8 -9.0 -27.3 -9.0 -27.3 AVERAGE: 1960-S -12.2 -24.8 -37.0 24.6 12.4 -9.3 -21.5 -9.9 -22.1 1970'S -17.4 -16.6 -34.0 -3.0 -20.4 -6.2 -23.6 -11.2 -28.6 1980'S -24.5 -13.2 -37.7 -15.4 -39.9 1.3 -23.2 6.2 -18.3 WHOLE PERIOD -16.8 -19.2 -35.9 5.6 -11.2 -6.0 -22.7 -7.2 -23.9 COUNTRY: PAKISTAN BASMATI IRRI YEAR COTTON RICE RICE SUGAR WHEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -40.1 -9.0 -49.1 . . . . . . 1.8 -38.3 1961 -36.9 -13.2 -50.2 . . . . 343.1 306.2 -3.8 -40.7 1962 -37.7 -11.2 -48.9 . . - . 162.7 125.0 -6.2 -43.9 1963 -38.3 -8.6 -47.0 -8.0 -46.3 . . 77.7 39.4 -4.3 -42.6 1964 -39.1 -1.2 -40.3 -11.6 -50.7 . . 112.7 73.6 2.4 -36.7 1965 -38.8 -8.6 -47.4 -6.1 -44.9 . . 76.5 37.7 1.2 -37.6 1966 -38.4 -9.2 -47.6 3.7 -34.7 . . 140.5 102.1 14.2 -24.2 1967 -41.0 -8.8 -49.9 -18.3 -59.3 . . 247.7 206.7 4.7 -36.3 1968 -44.5 -8.9 -53.4 -8.3 -52,8 . . 608.0 563.5 28.3 -16.2 1969 -42.4 -8.6 -51.0 -1.2 -43.5 . . 215.5 173.2 30.5 -11.8 1970 -42.4 -2.9 -45.3 -0.6 -43.0 27-1 -15.4 138.7 96.3 25.9 -16.5 1971 -47.2 -17.4 -64.7 -21.6 -68.9 12.1 -35.1 -16.4 -63.6 4.2 -43.0 1972 -12.8 -20.1 -32.8 -44.5 -57.3 -48.9 -61.6 -16.6 -29.3 -26.2 -38.9 1973 -27.3 -25.4 -52.8 -48.7 -76.0 -57.4 -84.7 -30.5 -57.9 -34.9 -62.2 1974 -28.7 -22.1 -50.8 -35.0 -63.6 -49.2 -77.9 -47.8 -76.5 -34.2 -62.9 1975 -32.3 -14.9 -47.2 -37.3 -69.5 -30.5 -62.7 -12.2 -44.5 -25.7 -58.0 1976 -30.5 -27.8 -58.3 -8.3 -38.8 -6.3 -36.7 -10.4 -40.9 -12.5 -43.0 1977 -32.1 -11.5 -43.6 -26.5 -58.6 -27.2 -59.3 -13.6 -45.7 -18.3 -50.4 1978 -28.5 -5.7 -34.2 -45.0 -73.5 -33.6 -62.1 -15.0 -43.5 -20.0 -48.5 1979 -28.4 -18.6 -47.0 -43.7 -72.1 -35.8 -64.2 -40.1 -68.5 -15.0 -43.4 1980 -29.6 -23.9 -53.5 -36.6 -66.2 -38.0 -67.6 -40.8 -70.4 -17.6 -47.2 1981 -29.7 -12.0 -41.6 -35.9 -65.5 -29.5 -59.2 -33.8 -63.4 -21.1 -50.8 1982 -25.3 -19.4 -44.7 -39.6 -64.9 -14.2 -39.5 -16.4 -41.8 -15.7 -41.0 1983 -29.0 -4.3 -33.2 -37.6 -66.6 -12.8 -41.7 -12.8 -41.7 -30.5 -59.5 1984 -25.2 -16.4 -41.7 -44.9 -70.1 -21.7 -46.9 13.5 -11.8 -24.7 -49.9 1985 -24.4 -14.4 -38.8 -46.8. -71.3 -0.8 -25.2 3.8 -20.7 -8.3 -32.8 T 1986 -21.8 -6.3 -28.1 -46.9 -68.7 -2.3 -24.2 27.4 5.5 -25.0 -46.8 AVERAGE: 1960'S -39.7 -8.7 -48.5 -5.0 -33.2 . - 198.4 162.7 6.9 -32.8 1970'S -31.0 -16.7 -47.7 -31.1 -62.1 -25.0 -56.0 -6.4 -37.4 -15.7 -46.7 1980'S -26.4 -13.8 -40.2 -41.2 -67.6 -17.0 -43.5 -8.5 -34.9 -20.4 -46.9 WHOLE PERIOD -33.1 -13.0 -46.0 -24.0 -52,8 -13.7 -32.0 68.9 37.4 -8.6 -41.6 ul COUNTRY: PHILIPPINES YEAR COPPRA CORN RICE SUGAR iNDIRECT NPRd NPRT NPRd NPRT NPRd NPRT MPRd NPRI 1960 -25.0 2.5 -22.5 84.4 59.4 21.9 -3.1 -10.7 -35.7 1961 -22.8 15.4 -7.3 106.2 83.5 81.8 59.0 5.2 -17.6 1962 -20.5 -2.1 -22.5 4.7 -15.8 -8.4 -28.8 -15.6 -36.0 1963 -17.7 -1.9 -19.6 12.3 -5.4 -5.5 -23.2 -3.4 -21.1 1964 -19.6 -6.8 -26.5 42.5 22.9 16.1 -3.6 -10.5 -30.1 1965 -19.9 4.8 -15.1 44.5 24.7 22.3 2.4 -15.3 -35.2 1966 -18.6 -9.4 -28.0 30.0 11.4 27.1 8.5 -13.3 -31.9 1967 -21.8 -6.0 -27.9 57.3 35.5 11.6 -10.2 -11.4 -33.2 1968 -23.7 -3.9 -27.6 32.1 8.4 1.5 -22.2 -15.5 -39.2 1969 -23.6 -1.5 -25.1 4.9 -18.7 1.5 -22.1 -3.9 -27.5 1970 -22.0 -15.3 -37.3 32.1 10.1 1.4 -20.6 -18.4 -40.4 1971 -20.8 -8.9 -29.7 46.6 25.8 61.1 40.3 -16.2 -36.9 1972 -20.7 -12.8 -33.5 -1.4 -22.1 18.0 -2.7 -14.0 -34.7 1973 -13.5 7.5 -6.1 -10.8 -24.4 -38.4 -52.0 -20.9 -34.5 1974 -20.8 -14.1 -34.9 0.0 -20.8 -36.8 -57.6 -40.8 -61.6 1975 -24.3 -25.0 -49.4 17.9 -6.4 -12.8 -37.2 -41.4 -65.7 1976 -26.1 -3.6 -29.7 27.7 1.6 12.2 -13.9 -15.4 -41.5 1977 -26.3 -17.9 -44.2 20.4 -5.9 -6.5 -32.8 -13.9 -40.1 1978 -30.7 -9.4 -40.2 21.9 -8.8 -13.2 -43.9 4.7 -26.0 1979 -29.5 -0.2 -29.7 13.8 -15.7 -8.4 -37.9 26.0 -3.5 f 1980 -28.9 -28.7 -57.6 14.6 -14.3 -19.3 -48.2 -16.6 -45.5 1981 -28.0 -24.1 -52.2 51.7 23.6 -15.3 -43.3 -26.6 -54.6 1982 -28.6 -25.2 -53.8 18.8 -9.7 15.7 -12.9 -5.3 -33.9 1983 -27.0 -3.4 -30.4 14.0 -13.0 3.8 -23.2 -20.6 -47.6 1984 -26.5 -9.6 -36.1 39.1 12.6 0.4 -26.1 -4.9 -31.4 1985 -22.8 -27.7 -50.5 54.7 31.9 27.5 4.7 -25.2 -48.0 C 1986 -19.8 -22.0 -41.8 0.0 -19.8 1.5 -18.3 -38.2 -58.0 AVERAGE: 1960'5 -21.3 -0.9 -22.2 41.9 20.6 17.0 -4.3 -9.4 -30.8 1970's -23.5 -10.0 -33.5 16.8 -6.7 -2.3 -25.8 -15.0 -38.5 1980's -25.9 -20.1 -46.1 27.6 1.6 2.0 -23.9 -19.6 -45.6 WHOLE PERID -23.3 -9.2 -32.6 28.9 5.6 5.9 -17.4 -14.2 -37.5 COUNTRY: PORTUGAL YEAR BEEF CORN MILK RICE SHEEP TOMATOES WHEAT INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 16.0 41.5 57.6 45.1 61.2 0.0 16.0 2.3 18.4 8.1 24.2 0.0 16.0 55.5 71.5 1961 12.6 52.4 64.9 12.0 24.6 0.0 12.6 3.8 16.4 16.0 28.6 0.0 12.6 52.1 64.7 1962 9.2 53.0 62.2 13.8 23.0 0.0 9.2 -36.4 -27.2 11.8 21.0 0.0 9.2 47.5 56.7 1963 -0.9 58.4 57.5 19.2 18.3 0.0 -0.9 -2.8 -3.7 15.6 14.7 0.0 -0.9 50.5 49.6 1964 14.4 26.3 40.7 18.9 33.3 0.0 14.4 5.3 19.6 17.5 31.9 0.0 14.4 45.5 59.9 1965 11.7 23.5 35.2 39.0 50.7 0.0 11.7 23.9 35.6 13.5 25.2 0.0 11.7 82.9 94.6 1966 9.1 40.4 49.4 27.4 36.4 0.0 9.1 11.2 20.3 25.0 34.0 0.0 9.1 65.2 74.3 1967 13.5 66.1 79.6 27.2 40.8 21.8 35.3 5.7 19.2 30.5 44.1 0.0 13.5 60.3 73.8 1968 -3.3 76.8 73.5 24.1 20.8 18.6 15.3 8.9 5.6 29.4 26.1 0.0 -3.3 67.5 64.2 1969 -8.0 67.1 59.1 23.3 15.3 17.7 9.7 30.8 22.9 32.5 24.5 1.7 -6.2 79.4 71.5 1970 -2.1 45.7 43.7 7.1 5.1 18.8 16.7 27.3 25.3 30.3 28.2 5.5 3.4 55.5 53.5 1971 -3.4 41.7 38.4 32.7 29.3 9.1 5.7 -12.1 -15.5 32.8 29.4 10.7 7.3 87.8 84.4 1972 -4.7 42.1 37.4 46.6 41.9 11.9 7.2 37.4 32.7 29.4 24.7 12.4 7.7 58.5 53.8 1973 6.7 66.2 59.5 0.0 -6.7 11.7 4.9 -16.2 -22.9 29.9 23.1 0.0 -6.7 -10.1 -16.8 1974 0.3 54.6 54.9 -22.3 -22.0 11.4 11.7 -38.8 -38.5 26.6 26.9 62.0 62.3 -25.5 -25.2 1975 -1.0 123.3 122.3 29.8 28.8 4.7 3.7 -11.5 -12.5 31.8 30.8 40.6 39.6 -8.2 -9.2 1976 -2.9 151.0 148.2 30.2 27.3 3.9 1.0 -0.3 -3.2 34.7 31.8 11.2 8.3 3.7 0.8 1977 -14.7 140.5 125.8 35.4 20.7 3.0 -11.7 -10.1 -24.8 26.7 12.0 0.0 -14.7 16.7 2.0 Q 1978 -15.1 121.7 106.7 31.9 16.8 2.1 -12.9 -6.5 -21.5 24.1 9.1 7.4 -7.7 20.0 5.0 1979 7.2 122.5 129.7 24.3 31.5 2.4 9.5 7.2 14.4 31.0 38.2 26.2 33.3 38.8 46.0 1980 -12.6 85.8 73.2 18.3 5.7 5.1 -7.6 12.8 0.2 15.9 3.3 37.0 24.3 24.3 11.7 1981 -8.1 63.0 54.9 14.4 6.3 7.3 -0.9 -20.7 -28.8 4.3 -3.8 24.3 16.1 10.6 2.4 1982 -13.2 41.4 28.2 33.7 20.5 5.5 -7.7 0.6 -12.6 7.6 -5.6 7.4 -5.8 26.9 13.7 1983 -16.2 69.0 52.8 34.5 18.3 31.1 14.9 13.2 -3.0 13.6 -2.6 3.9 -12.3 26.6 10.4 1984 -14.3 37.9 23.6 31.3 17.0 2.9 -11.4 17.7 3.4 4.6 -9.7 12.9 -1.4 37.8 23.5 AVERAGE: 19601S 7.4 50.5 58.0 25.0 32.4 5.8 13.2 5.3 12.7 20.0 27.4 0.2 7.6 60.6 68.1 1970'S -4.3 91.0 86.7 21.6 17.3 7.9 3.6 -2.4 -6.7 29.7 25.4 17.6 13.3 23.7 19.4 1980'S -12.9 59.4 46.5 26.4 13.6 10.4 -2.5 4.7 -8.2 9.2 -3.7 17.1 4.2 25.2 12.4 WHOLE PERIOD -1.3 68.5 67.2 23.9 22.6 7.6 6.2 2.1 0.8 21.7 20.4 10.5 9.2 38.8 37.5 COUNTRY: SRI LANKA YEAR COCONUT RICE RUBBER TEA INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 -12.4 -22,1 -34.5 101.7 89.2 -12.7 -25.2 -19.1 -31.5 1961 -25.8 -19,4 -45.1 93.9 68.1 -10.6 -36.4 -16.7 -42.5 1962 -28.0 -3.6 -31.6 42.8 14.8 -11.1 -39.2 -17.0 -45.0 1963 -31.8 -12,2 -44.0 78.0 46.1 -10.2 -42.0 -17.9 -49.7 1964 -13.5 -24,6 -38.1 73.5 60.0 -11.3 -24.7 -22.2 -35.7 1965 -35.8 -17,9 -53.7 35.9 0.1 -7.4 -43.2 -13.3 -49.1 1966 -26.1 -23.0 -49.1 60.2 34.1 -9.3 -35.4 -18.8 -44.9 1967 -36.2 -14.3 -50.5 47.9 11.7 -5.9 -42.1 -16.1 -52.3 1968 -38.6 -10,7 -49.3 15.6 -23.0 -1.8 -40.4 -14.2 -52.8 1969 -26.4 -22.9 -49.3 40.2 13.9 -11.4 -37.8 -21.3 -47.7 1970 -23.6 -22.7 -46.2 62.7 39.1 -14.2 -37.8 -19.8 -43.3 1971 -32.7 -22.1 -54.9 75.9 43.1 -12.6 -45.4 -13.7 -46.5 1972 -30.4 -8.5 -38.9 84.1 53.7 -2.4 -32.9 -16.6 -47.0 1973 -34.4 -6.6 -41.0 72.8 38.3 -14.7 -49.1 -15.2 -49.7 1974 -39.5 -27.7 -67.2 12.6 -26.9 -27.5 -67.0 -9.6 -49.1 1975 -44.6 -23.6 -68.2 29.9 -14.7 -17.6 -62.2 -13.7 -58.3 1976 -46.6 -6.3 -52.9 40.5 -6.1 -13.8 -60.4 -9.3 -55.9 1977 -42.3 -0.6 -42.9 40.5 -1.7 -15.5 -57.8 -13.0 -55.3 1978 -35.3 -31.8 -67.2 -7.0 -42.3 -31.3 -66.6 -39.7 -75.0 1979 -26.0 -39.7 -65.7 -13.9 -39.9 -35.3 -61.3 -44.0 -70.0 1980 -50.0 -23.8 -73.7 -3.9 -53.9 -22.5 -72.4 -20.5 -70.5 1981 -25.8 -13.6 -39.4 -3.4 -29.2 -36.2 -62.0 -32.6 -58.5 1982 -36.2 1.1 -35.0 9.2 -27.0 -23.0 -59.2 -18.2 -54.4 1983 -32.4 7.5 -24.9 7.8 -24.6 -21.6 -54.0 -14.7 -47.2 1984 -14.9 8.8 -6.1 16.3 1.4 -31.1 -46.0 -28.2 -43.1 1985 -19.0 -16.5 -35.4 21.4 2.5 -12.8 -31.8 -28.5 -47.4 AVERAGE: 1960'S -27.5 -17.1 -44.5 59.0 31.5 -9.2 -36.6 -17.6 -45.1 1970'S -35.6 -19.0 -54.5 39.8 4.3 -18.5 -54.0 -19.5 -55.0 1980's -29.7 -6.1 -35.8 7.9 -21.8 -24.5 -54.2 -23.8 -53.5 WHOLE PERIO0 -31.1 -15.3 -46.4 39.8 8.7 -16.3 -47.4 -19.8 -50.9 COUNTRY: THAILAND YEAR MAIZE RICE RUBBER SUGAR INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT NPRd NPRT 1960 . . . - . 1961 . . . - . - 1962 -11.6 -4.6 -16.1 -34.4 -45.9 -11.7 -23.3 171.5 159.9 1963 -12.5 -4.5 -17.0 -36.6 -49.1 -12.6 -25.1 19.9 7.4 1964 -9.1 -4.7 -13.8 -41.4 -50.5 -11.9 -20.9 0.4 -8.7 1965 -9.9 -4.6 -14.5 -40.0 -49.9 -12.4 -22.3 114.6 104.8 1966 -11.3 -4.6 -15.9 -31.2 -42.5 -11.6 -23.0 40.8 29.4 1967 -11.0 -4.6 -15.6 -35.9 -47.0 -10.5 -21.5 8.5 -2.5 1968 -15.1 -4.4 -19.4 -41.5 -56.6 -8.5 -23.6 35.8 20.7 1969 -17.8 -1.3 -19.1 -31.7 -49.5 -13.4 -31.2 9.1 -8.7 1970 -18.5 -2.2 -20.7 -26.2 -44.7 -10.6 -29.1 19.4 0.9 1971 -20.3 -2.9 -23.2 -28.0 -48.4 -8.4 -28.8 1.7 -18.6 1972 -16.2 5.2 -11.0 -28.6 -44.8 -9.2 -25.5 -18.4 -34.7 1973 -13.7 -8.4 -22.0 -48.8 -62.4 -14.8 -28.4 -25.3 -39.0 1974 -5.7 -2.4 -8.2 -59.4 -65.2 -17.8 -23.5 -58.7 -64.4 1975 -13.1 -5.6 -18.8 -40.3 -53.4 -14.4 -27.6 -48.4 -61.6 1976 -12.6 -2.9 -15.4 -18.5 -31.1 -18.2 -30.8 -18.2 -30.7 1977 -17.4 0.3 -17.2 -21.8 -39.2 -18.4 -35.8 -2.1 -19.5 1978 -16.0 -1.9 -17.9 -33.8 -49.9 -19.1 -35.1 29.3 13.3 1979 -18.0 -3.8 -21.8 -23.8 -41.8 -20.1 -38.1 32.1 14.1 1980 -19.1 -4.4 -23.5 -25.0 -44.1 -21.1 -40.2 -2.3 -21.4 1981 -20.7 -6.9 -27.6 -24.3 -45.0 -15.0 -35.7 -8.1 -28.8 1982 -15.1 0.0 -15.1 -11.6 -26.8 -11.1 -26.3 30.6 15.5 1983 -21.5 0.0 -21.5 -7.6 -29.1 -14.1 -35.6 40.5 18.9 1984 -18.9 0.0 -18.9 -5.8 -24.7 -12.2 -31.1 43.1 24.2 AVERAGE: 1960'S -12.3 -4.1 -16.4 -36.6 -48.9 -11.6 -23.9 50.1 37.8 1970'S -15.2 -2.5 -17.6 -32.9 -48.1 -15.1 -30.3 -8.9 -24.0 1980's -19.1 -2.3 -21.3 -14.9 -33.9 -14.7 -33.8 20.8 1.7 WHOLE PERIOD -15.0 -3.0 -18.0 -30.3 -45.3 -13.8 -28.8 18.1 3.1 COUNTRY: TURKEY YEAR BARLEY COTTON HAZELNUTS TOBACCO WHEAT INDIRECT NPRd NPRT NPRd NPRT NRRd NPRT NPRd NPRT NPRd NPRT 1960 . . . . . . - 1961 -39.6 -3.8 -43.4 14.7 -24.9 -6.1 -45.7 -20.6 -60.1 3.9 -35.7 1962 -22.9 -9.3 -32.1 26.5 3.6 2.0 -20.9 6.9 -16.0 24.4 1.5 1963 -28.7 -9.8 -38.5 25.9 -2.8 -8.6 -37.3 8.3 -20.4 15.0 -13.6 1964 -35.9 -10.6 -46.5 22.7 -13.2 3.3 -32.6 -12.1 -48.0 11.9 -24.1 1965 -35.1 -6.6 -41.7 25.2 -10.0 1.3 -33.8 -13.0 -48.1 20.8 -14.4 1966 -39.6 -4.9 -44.5 23.9 -15.7 3.4 -36.2 -0.8 -40.4 26.8 -12.8 1967 -47.2 -4.3 -51.5 22.B -24.4 -0.4 -47.6 -4.4 -51.6 -16.3 -63.5 1968 -45.3 -0.2 -45.5 17.7 -27.5 -3.6 -48.9 -2.1 -47.4 30.1 -15.2 1969 -46.3 7.9 -38.4 22.9 -23.4 3.1 -43.2 -0.2 -46.5 24.9 -21.4 1970 -41.1 21.1 -20.1 22.2 -19.0 -13.8 -55.0 1.4 -39.8 39.6 -1.5 1971 -38.9 11.7 -27.2 17.5 -21.5 -1.0 -40.0 8.1 -30.9 -1.4 -40.3 1972 -32.1 25.8 -6.3 23.0 -9.0 1.9 -30.2 3.9 -28.2 38.5 6.4 1973 -27.6 -37.9 -65.5 18.0 -9.6 2.4 -25.2 5.7 -21.9 -42.7 -70.3 1974 -30.0 -13.2 -43.2 19.6 -10.4 -8.9 -38.9 11.4 -18.6 -17.2 -47.2 1975 -31.2 -7.8 -39.1 56.2 24.9 3.4 -27.9 5.6 -25.6 -6.3 -37.6 1976 -33.6 6.6 -26.9 55.7 22.2 10.8 -22.7 -2.4 -36.0 1.3 -32.2 1977 -38.1 10.0 -28.2 33.4 -4.7 2.9 -35.2 8.3 -29.8 60.2 22.1 1978 -48.4 13.1 -35.3 20.2 -28.3 -8.6 -57.0 -3.4 -51.8 43.7 -4.7 1979 -49.7 -4.9 -54.6 31.3 -18.4 0.0 -49.7 0.6 -49.1 39.8 -9.8 1980 -42.7 -15.0 -57.7 -2.3 -45.1 -18.6 -61.3 -18.2 -60.9 27.3 -15.4 1981 -37.6 -7.9 -45.5 8.8 -28.8 -28.2 -65.7 -29.9 -67.5 -12.2 -49.8 1982 -29.1 -23.0 -52.1 15.0 -14.1 -22.7 -51.8 -30.3 -59.4 -15.3 -44.4 1983 -32.4 -15.7 -48.2 4.4 -28.1 -27.4 -59.9 -31.2 -63.7 -12.4 -44.8 AVERAGE: 19601S -37.8 -4.6 -42.5 22.5 -15.4 -0.6 -38.5 -4.2 -42.1 15.7 -22.1 1970-S -37.1 2.5 -34.6 29.7 -7.4 -1.1 -38.2 3.9 -33.2 15.6 -21.5 19805S -35.5 -15.4 -50.9 6.5 -29.0 -24.2 -59.7 -27.4 -62.9 -3.1 -38.6 WHOLE PERIOD -37.1 -3.4 -40.5 22.8 -14.3 -4.9 -42.0 -4.7 -41.8 12.4 -24.7 Appendix:-Direct and Total Nominal Protection Rates 251 COUNTRY: ZAMBIA YEAR COTTON MAIZE TOBACCO INDIRECT NPRd NPRT NPRd NPRT NPRd NPRT 1966 5.7 -24.6 -18.9 3.2 8.9 0.2 6.0 1967 1.9 -18.1 -16.2 -14.9 -13.0 -0.1 1.7 1968 -3.0 -23.9 -26.9 -3.2 -6.2 0.4 -2.6 1969 1.3 -13.7 -12.5 -46.3 -45.0 0.4 1.7 1970 -5.9 -17.9 -23.8 -53.8 -59.6 31.3 25.4 1971 -16.1 -17.0 -33.1 -43.1 -59.2 17.8 1.7 1972 -20.0 -20.6 -40.6 -30.4 -50.4 16.1 -3.9 1973 -14.7 -42.3 -57.0 11.7 -3.0 1.6 -13.1 1974 -24.3 -27.6 -51.9 -28.0 -52.3 -0.3 -24.6 1975 -34.4 -8.9 -43.2 -17.5 -51.9 5.0 -29.3 1976 -39.5 -9.7 -49.1 -9.8 -49.3 0.2 -39.3 1977 -41.6 -19.5 -61.1 -12.5 -54.1 -0.3 -41.9 1978 -44.7 -15.5 -60.2 -8.4 -53.1 -0.1 -44.8 1979 -47.5 -13.6 -61.1 -15.7 -63.2 -0.1 -47.7 1980 -54.4 6.4 -48.0 -10.0 -64.4 0.1 -54.3 1981 -59.8 -2.4 -62.2 -13.7 -73.5 0.0 -59.8 1982 -61.5 -2.6 -64.1 -7.6 -69.2 11.3 -50.3 1983 -57.8 -3.1 -60.9 -6.7 -64.6 13.7 -44.1 1984 -51.8 -21.4 -73.2 -5.9 -57.7 11.9 -39.9 AVERAGE: 1960'S 1.5 -20.1 -18.6 -15.3 -13.8 0.2 1.7 1970*S -28.9 -19.3 -48.1 -20.8 -49.6 7.1 -21.7 1980'S -57.1 -4.6 -61.7 , -8.8 -65.9 7.4 -49.7 WHOLE PERIOD -29.9 -15.6 -45.5 -16.5 -46.4 5,7 -24.2 Notes 2. The Effects of Intervention on Agricultural Prices 1. Moreover, the tax resulting from overvaluation of the real exchange rate is probably lower than the rate of overvaluation because not all agricultural products are entirely tradable, and even traded products have nontradable components, including some distribution and marketing costs. 2. It is important to note that our measure of the direct nominal protection rate is smaller in absolute value than the usual measure. We defined the direct protection rate as the difference between the producer price and the border price in the absence of direct interventions, relative to the border price in the absence of total interventions. We used the same denominator (the price in the absence of total interventions) for the direct protection rate as for the indirect and total protection rates so that the sum of the direct and indirect rates equaled the total rate of protection. For instance, if P, the agricultural producer price relative to the non- agricultural price index (PA PN), is 40, the border price in the absence of direct interventions (P') is 60, and the border price in the absence of total interventions (P*) is 100, then the usual measure of direct protection is (P - P')/P* = (40-60)/60 or -33 percent, while our measure is (P - P')IP* = (40-60)/100 or -20 percent. The rate of indirect protection is (P' - P*)/P* = 60-100)/100 or -40 percent, and the rate of total protection is (P - P*)IP* = (40-100)/100 or -60 percent-the sum of our measures of direct and indirect protection. 3. Since nominal protection rates are negative in many cases, indicating taxation of agriculture, we frequently refer to them as rates of taxation. Also, we often use the shorthand expressions "direct or indirect rates of protection or taxation" to mean the rates of protection or taxation that result from direct or indirect interventions. 4. The divisia price index at time t is a geometric average of the individual product prices, with weights equal to the share of each product in the value of output. For discrete time, the weights are equal to the average of the shares at times t and t - 1. For a formal derivation and discussion of the properties of the index, see Diewert (1976). 5. In a study of agricultural price policies in developing countries, Tolley, Thomas, and Wong (1982) examine the impact of direct interventions on 252 Notes 253 agricultural output, budget, welfare, and other variables in Bangladesh, Korea, Thailand, and Venezuela. 6. Note that the indirect tax is large for Korea because the current account deficits were included in the exchange rate misalignment calculations even though the deficits may well have reflected an optimal intertemporal invest- ment strategy. As we show later, the indirect tax declined steadily over time. 7. This result is corroborated in a study by Thomas (1985) of the impact of macroeconomic policies on agriculture in Colombia. He concluded that the impact of real exchange rate overvaluation on agricultural incentives domi- nated the impact of sectoral policies. 8. For some commodities, the countries examined may have some market power as a group. If so, eliminating the tax on producers (direct, indirect, or total) simultaneously in the various countries may lead to a fall in the world price over time. In the case of simultaneous liberalization of the trade regime, this would imply a smaller tax on agriculture than the one reported here. However, such simultaneous policy change would not affect the relative importance of direct and indirect intervention. The issue of the impact of simultaneous changes in producer taxes for cocoa is examined by Panagariya and Schiff (1991). 9. Recall that these rates of direct protection are smaller than those derived according to the usual definition, which is about 70 percent for corn and about 69 percent for rice in Ghana. The large difference between the two types of direct protection rates is due to the importance of the indirect tax (see note 2, and appendix 2-1 for a formal derivation). 10. For Brazil, the high tax on consumption resulted because beef, with its high consumer price subsidy, was excluded from our analysis, while rice, which is heavily protected, had a high share in the two products (rice and wheat) analyzed for Brazil. 11. For Ghana, calculating the equilibrium real exchange rate involved an additional complication. Depreciation of the real exchange rate to its equilib- rium value for a given world price of cocoa would lead to an increase in Ghana's cocoa output. But Ghana's output is such a large part of world cocoa trade that this supply rise would reduce the world price. We therefore esti- mated the equilibrium real exchange rate in a simultaneous system that assumes that the optimal export tax is applied and that the world price of cocoa is determined endogenously as a function of Ghana's real exchange rate. 3. The Effects of Intervention on Price Variability 1. Migration and marriage outside the village are other ways that farmers stabilize consumption. Rosenzweig and Stark (1989) found that as output and income uncertainty increase, rural households send daughters to marry in more distant villages as a risk-diffusion mechanism. 2. Barah and Binswanger (1981) found that in India price stabilization poli- cies stabilized revenue mainly in the richer, irrigated areas and increased income variability in the poorer, unirrigated areas. 3. One argument for public buffer stocks-the fear that tradable staples cannot be imported fast enough-may be exaggerated, especially if private 254 Notes storage is allowed. A study of Pakistan showed that private storage is very responsive to expected price changes and can complement a modest amount of public storage, which may be justified because of the lead time needed to import wheat (see Knudsen and Nash 1990). 4. For studies on the impact of uncertainty on production decisions, see Batra and Ullah 1974; Just 1974; Just and Pope 1976; Magnusson 1969; Nowshirvani 1971; and Schiff 1986. 5. Assuming that costs are given, variability of revenue is equal to vari- ability of income. 4. Supply Response and Growth in Agriculture and the Economy (in collaboration with Jorge Quiroz) 1. The assumption of a low aggregate supply response was one of the basic tenets of the Latin American "structuralist" school (Prebisch 1949 and 1950; Furtado 1965; and Sunkel 1967) of the 1950s and 1960s and motivated its recommendation for import-substitution-based industrialization. This view of a very low aggregate supply response to an increase in agriculture's terms of trade has also become quite prevalent in the literature on Sub-Saharan Africa (see, for example, Delgado and Mellor 1984 and de Janvry 1986). Yet this same literature on Africa argues that capital inflows and food aid, by leading to appreciation of the real exchange rate and high rates of industrial protection, turned terms of trade against agriculture and led to a massive labor outflow from the sector. This argument implies high aggregate supply response to a decrease in agriculture's terms of trade. The issue remains unresolved and requires further analysis. 2. This, of course, abstracts from the potentially high social cost of distor- tions of relative prices within the agricultural sector. We return to this issue in the section on the impact of price interventions on output mix and agri- cultural GDP. 3. It affects both tradable and nontradable agricultural products; it affects nontradables less because they gain from the removal of industrial protection but not directly from removal of exchange rate misalignment. 4. In this study of the evolution of agricultural pricing policies, the output effect is only one of several issues considered in the country studies. We were not trying to develop new analytical or methodological techniques for esti- mating output effects, but simply to use existing methods or estimates to measure these effects. As in most studies on output effects, the estimates (whether from the country studies or from other studies) were generally based on single-equation estimation, and most estimates are for the individ- ual products examined and not for the agricultural sector as a whole. Single-equation estimation generally fails to capture some of the dynamic elements that influence the long-run output effects for individual products and the sector as a whole, including the movement of resources within agri- culture and between agriculture and the rest of the economy. Doing this requires estimation of investment and migration equations and is extremely time- and data-intensive. We were able to estimate these effects for Chile, because intersectoral migration and investment equations had been previ- Notes 255 ously estimated. This is discussed in the section dealing with long-run aggre- gate supply response. 5. The Nerlovian approach assumes that output at time t depends on all past prices P,-j, with the weights attached to those prices diminishing expo- nentially as j increases. 6. This assumes no market power on world markets. The gains from a change in the output mix result from an improvement in the allocation of resources. The change in output mix implies a movement along the transfor- mation curve (within agriculture) and leads to gains in real income (measured at border prices) and consumption possibilities for the country. Using an ideal quantity index, which holds the transformation curve constant, would indi- cate a consumption or real income gain. If border prices are used, however, the quantity index of real agricultural GDP will show a positive output effect. 7. This is probably not a large value, since the output of exportables can be increased by switching labor, land, capital, and purchased inputs from importables to exportables. For tree crops, the short-run elasticity will be smaller and the elasticity of 1 may reflect the long-run response to a relative price change. 8. The Divisia quantity index uses as weights the average of the shares of each product before and after reform in the value of output. The Divisia price index uses the same weights. 9. The lower elasticity, Ex = 0.44, is chosen in order to avoid a corner solution when interventions are removed. 10. For this test, we used the common definition of the direct protection rate rather than our alternative definition (see chapter 2, note 2 and appendix) because it reflects the actual change in price resulting from direct interven- tions and so is the relevant measure to use in evaluating output response using a supply elasticity. (Our measure of direct protection was adjusted to make it comparable to the measure of indirect protection, so the two rates together would equal total protection.) 11. In this particular case, an additional problem with the direct protection rate is that it applies to the products selected rather than to agricultural GDP. 12. For that reason it may be impossible to distinguish this model from one in which the growth rate of agricultural GDP depends on the level of total protection, as posited by Mundlak (1985). 13. To check the adequacy of the model, we compared the actual values of agricultural GDP and the values obtained from the model using the actual rates of protection. In most cases, the model performs very well. The correla- tion between the two series is above 0.88 in thirteen countries and is between 0.54 and 0.66 in three countries. For Argentina (0.36) and Colombia (0.11) the correlation is low, so the results of the simulation are less reliable for them than for the other countries. 14. For instance, Bond (1983) uses a trend variable to "take into account the effect of long-run structural changes on equilibrium output" (720). But some of these changes are caused by price policy and cannot be captured with this approach or with the other single-equation regressions discussed in the pre- ceding material. A rigorous analysis of long-run aggregate price elasticity should incorporate the impact of price policy on factors of production that are fixed in the short run. 256 Notes 15. For an analysis of this issue, with an application to C6te d'Ivoire, see Lopez (1992). 16. The three tests indicate the same relationship between total interven- tion and growth. They are not, however, formal causality tests. 17. The critical t-value is 1.34 for a one-tailed test at the 10 percent confi- dence level and 1.75 at the 5 percent level. 18. The value of test statistic v has a t-distribution as long as annual GDP growth is distributed normally. The critical t-value with 16 degrees of free- dom for a one-tailed test is 1.75 at the 5 percent level, 2.12 at the 2.5 percent level, and 2.58 at the 1 percent level. 19. These findings are confirmed by several sources-of-growth studies. McCarthy, Hanson, and Kwon (1987) found that total factor productivity growth in Colombia was positively related to the degree of trade liberalization (with the highest growth occurring in 1967-74). Krueger and Tuncer (1982) obtained the same result for Turkey. 20. Using W,_, in the supply function for X, assumes that all input costs occur at time t - 1, either because factors are used at that time or because contracts for their services are made at that time. If some factor costs are determined at time t, then their price at time t - 1 must be replaced by their expected price at time t. 21. Purchased inputs at the farm level may not always be important, but those involved in transportation and marketing are, and they affect the pro- ducer price. A foreign exchange crisis may lead to rationing of gasoline, spare parts for trucks, and so on, resulting in a lower product price. In the cases where this problem is important, it should be taken into account. 22. For the agricultural sector as a whole, primary factors are given in the short run, and their price will tend to be affected by price policies. The same is true in the long run when those factors become variable if agriculture represents a significant share of GDP. 23. If there are several purchased inputs z, then AP,,,- should be thought of as a column vector and d as a row vector. 24. The derivation of the elasticity with respect to value added is found in Johnson (1969). An application of the concept to the impact of effective pro- tection, including the effect of the exchange rate, is presented in Valdds (1973). 25. AVAX and AVA,, are the proportional changes in VA (since the VA terms are in logs). 5. Consumption and Trade Effects of Intervention 1. The effects of industrial trade, macroeconomic, and fiscal policies on foreign exchange earnings were not included in the calculations except as they affected agricultural trade through their impact on incentives. 2. To determine the impact of interventions on consumption, consumer prices ought to be evaluated at the retail level. In some countries, however, the impact of interventions was measured at the wholesale level, and a markup was added to cover costs from the producer to the consumer, such as for milling, transport, and storage. Interventions often affect these intermedi- Notes 257 ary activities as well. Milling is often subsidized, parastatals may subsidize transport or storage, gasoline may be taxed, trucks and parts may be subject to import duties, and more. These interventions affect the overall interven- tion rate at the consumer level. They were taken into account in some coun- tries (Brazil, for example), but not in others. 3. Countries have also intervened in other ways that have affected foreign exchange earnings. Many of these interventions, which were designed to save foreign exchange, resulted in foreign exchange losses, but these effects were not quantified for this study. For instance, Sub-Saharan African coun- tries such as Zambia and Zimbabwe have required import and foreign exchange licenses for importing agricultural inputs (pesticides, herbicides, fertilizers, and tractors and other farm equipment). License approval has often taken so long that the inputs had to be imported at the last moment from a higher-cost source or were not imported at all. The result was often a fall in the output of tradables and a loss of foreign exchange. 4. Portugal was an exception in some years, but the degree of industrial protection entering the calculation of indirect intervention may have been underestimated. (See the country study on Portugal for details.) 5. The cross-country average excludes Ghana because the foreign exchange effects were so high, especially in 1980-85, that their impact on the un- weighted average is larger than that of the other countries combined. 6. The Fiscal Impact of Direct Intervention 1. Or they do not yield revenue to the government from the agricultural sector, although industrial protection measures could yield revenues if tariffs rather than quantitative restrictions applied. 2. Losses from operations in input markets are allocated to input items except when the information provided by country authors combined input and output operations flows. 3. Concessional rates on electricity use are excluded if similar rates apply to other sectors, since these rates do not represent a subsidy specific to agriculture. 4. Data were not available from the country studies for C6te d'Ivoire and Portugal. 5. Estimates of credit subsidies exclude losses from defaults and underesti- mate the fiscal cost when government credit agencies absorb that loss. 6. The impact on the budget of subsidies on purchases of domestically produced wheat was small because the quantity procured was small. 7. The credit subsidy in real terms in year t (S,) was measured by S'=C, (i* - il) S, =1, here C, is the credit given by a bank at time t (to be repaid at time t + 1), i, is the nominal interest rate, i* is the nominal reference interest rate in the absence of subsidy, and P,, is the deflator used to adjust the nominal sub- sidy for inflation. 258 Notes 7. The Effects of Intervention on Income Transfers to and from Agriculture 1. Workers' remittances from employment outside agriculture were excluded in our measure of farm income. 2. In his paper, Ishikawa advocates a policy of a net injection of resources into agriculture, combined with an appropriate choice of technology and institutions. 3. The concept of price-related transfers used in Ishikawa's work differs from that used in this project. Ishikawa measures price-related transfers by the changes in the actual domestic commodity terms of trade between the two sectors relative to its value in a base year; this project uses border prices as the benchmark. 4. Classroom lectures on price theory, University of Chicago. 5. Even when input policies are crop specific, information on their alloca- tion is hard to come by. Crop-specific information was supplied for credit subsidies in the country studies for Malaysia and Thailand, however, and so for these countries the estimated rate of input subsidy for the rest of agricul- ture excludes these credit subsidies. 6. Net transfers are reported using the results for full-sector coverage of output-related transfers. Excluding the rest of agriculture would substantially bias the measures on net transfers, since both the input subsidies and non- price transfers apply to the sector as whole. The results are also reported for the average of the two alternative assumptions. 7. Even though income transfers from direct price-related interventions are 0 under assumption 1, transfers as a result of total price interventions (direct and indirect) are substantial in many cases because of the large effect of indirect interventions-industrial protection and overvaluation of the real exchange rate. 8. Exceptions where input transfers were higher than output transfers were Brazil, Chile, and, for some years, Colombia and Turkey. 9. For industrial countries we know that assistance to agriculture on a per farm basis has been rising over time (Organization for Economic Cooperation and Development 1987). For most developing countries, however, we are unable to calculate assistance on a per farm basis, since few countries report information on the number of farms for more than one year. 10. Output-related transfers out of agriculture are understated for Brazil, however, because coffee was not one of the selected products and so the income effects of output-related policies affecting coffee were not included in the calculation. 11. Similarly, some economists have viewed public investment expendi- tures and improved incentives as alternative strategies for agricultural devel- opment (see Krishna 1984 and Delgado and Mellor 1984). Even if it were true, however, that government investment expenditures tend to compensate for agricultural taxation, it is not clear that the net effect on income in agriculture would be positive or that such a policy would be optimal. These are complex issues that go beyond the objectives of this volume. 12. Compensation at the crop level is clear in some countries for tree crops. Taxes on tree crops are used partly to subsidize replanting and to finance Notes 259 research to improve productivity (as for rubber in Thailand and rubber and palm oil in Malaysia). 13. The error-correction term implies that any deviation of nonprice trans- fers from their long-term value in the previous period is partly corrected for in the present period. 14. Ideally, the social rate of return should be used in evaluating invest- ment projects, but private, measurable rates of return probably influence decisions about public investments in many cases. 15. Because agriculture has been aggregated into one sector, this frame- work does not capture the effect of selective price policies within agriculture (the possibility of net transfers out of agriculture simultaneously with positive transfers to particular producer groups). If the analysis were extended to individual crops, one might find that in the case of rice, for example, both producers and urban consumers lose and a parastatal captures the gains. 16. CPI'/CPI was not reported for the Dominican Republic and Morocco. For the Dominican Republic, PNA*IPNA was used instead, and no adjustment was made for CPI%. For Morocco, a price index was built using the share of food in cpi (0.55) provided in the study by Laraki (1989). 17. For some countries only an urban cpi was available. In those cases, the cpi of low-income urban consumers was chosen as a proxy for rural cpi. 8. The Effects of Intervention on Income Distribution and Poverty 1. Using the actual open market price instead of the border price equivalent as a reference could yield different results for the size of the income transfer. 2. The results presented for total price interventions represent the short- run effects only. Even in the medium run, it is unrealistic to assume that the major overhaul of the trade and exchange rate regimes implied by the removal of indirect price interventions will leave the nominal income of households in the various income groups unchanged, since wages and employment are likely to change in the trade-liberalizing manufacturing sectors. 3. Several studies have examined at least some general equilibrium implica- tions of changes in agricultural price policies on real income. These include Quizon and Binswanger (1986) for India; Dethier (1985) for Egypt; Amranand and Grais (1984) for Thailand; Reardon (1984) for Peru; and Braverman, Ahm, and Hammer (1983) for Korea. Innovative as they are, these studies do not really address the issue of interactions between agricultural price inter- ventions and the income of urban and rural wage-earning consumers in a long-term perspective. Farm investment and employment are not endo- genized, and most of these studies treat urban wages as exogenous. Even studies that include market-clearing relations in the urban labor market do so by modeling it with respect to fixed nominal wages in the urban sector. 4. The effects of policies on nominal income and on the cost of the basket of nonagricultural products were not captured empirically. 5. The categories varied; some countries reported data by farm size and others by income level. 260 Notes 6. In the country study, results were disaggregated for cocoa and rice. Here, we have aggregated those results using the corresponding weights in total gross value of production. 7. The results presented here are an aggregation of those presented in the country report. In an analysis covering rice only, Deaton (1989) concluded that higher prices for rice in Thailand are likely to benefit rural households at all levels of living, but especially households in the middle of the income distribution. These results differ from ours largely because of differences in the commodity coverage in the analysis. 8. In Turkey, "small" includes the first three deciles, "medium" includes the next five, and "large" the upper two. 9. Estimates on government flows come from chapter 6 (price interventions only, excluding public investment and other nonprice interventions). Non- agricultural income for urban and rural households is obtained as a residual. The analysis assumes that transfers are captured fully by households as con- sumers of food, although price-related transfers include some nonfood prod- ucts, such as cotton. Food expenditure shares by income classes are assumed to remain constant through time. 10. The study also found that the rural capital stock and the urban wage rate have a positive impact on the rural wage, whereas urban unemployment and the rural labor force have a negative impact on the rural wage. 11. Looking at the impact on other special interest groups, such as regional and ethnic groups or groups belonging to identifiable organizations such as public sector employees, the military, and organized labor, would also be informative, but we had insufficient data to do so. 12. If large-scale farmers hire labor (or rent machinery), that cost should also be deducted to obtain that farm's income (or residual profits). This is important for calculations of income effects over time, as returns to labor may change and may affect farm income. 13. If off-farm income and purchased inputs are negligible, and if farmers consume only the product i that they produce (in addition to their consump- tion of nonagricultural products), then it is easy to show that 9D = (1 - )NPRD where NPRD = (Pi - P') / Pi'). This relationship does not hold for 9T, and Y, is generally not negligible. Moreover, farmers may consume products other than the ones they produce. 9. An Empirical Analysis of Patterns of Intervention 1. For the eighteen countries examined in this project, the rate of direct protection was found to be negative but small, although it is positive for importables (see chapter 2). 2. Measured either as agricultural output per male farm worker divided by GDP per male worker in the economy or as agricultural land area per farm worker divided by GDP per capita. 3. The change AP is proportional, since P is the logarithm of the relative export price. Notes 261 4. Variables reflecting comparative advantage in agriculture, which were shown to be important in various models, are not included here because they were not available for all countries and would have severely reduced the size of the sample. However, these variables were found to be highly correlated with measures of the relative size of the agricultural sector, which are included in the model. 5. In fact, the average tax rate t = 1 - (P1PN)-0216. 6. The exchange rate variable was not significant in the error-correction term (-y equation). References Alderman, Harold. 1988a. "Food Price Subsidies in Egypt: Benefit Distribu- tion and Nutritional Effects." In Per Pinstrup-Andersen, ed., Food Subsidies in Developing Countries. Baltimore: Johns Hopkins University Press. -. 1988b. "The Twilight of Flour Ration in Pakistan." Food Policy 13(3): 245-56. Alderman, Harold, M. Ghaffar Chaudhry, and Marito Garcia. 1988. "House- hold Food Security in Pakistan: The Ration Shop System." IFPR Working Paper on Food Subsidies 4. International Food Policy Research Institute, Washington, D.C. Amranand, Piyasvasti, and Wafik Grais. 1984. Macroeconomic and Distribu- tional Implications of Sectoral Policy Interventions: An Application to Thailand. World Bank Staff Working Paper 607. Washington, D.C. Anderson, Kim, and Yujiro Hayami. 1986. The Political Economy of Agricultural Protection. London: Allen and Unwin. Avillez, Francisco, Timothy J. Finan, and Timothy Josling. 1988. "Trade, Exchange Rate and Agricultural Pricing Policies in Portugal," World Bank Comparative Studies series entitled the Political Economy of Agricultural Pricing Policy. Washington, D.C. Balassa, Bela. 1971. The Structure of Protection in Developing Countries. Balti- more: Johns Hopkins University Press. _. 1987. "Economic Incentives and Agricultural Exports in Developing Countries." Report DRD 250. World Bank, Development Research Depart- ment, Washington, D.C. Bale, Malcolm D., and Ernst Lutz. 1979. Price Distortions in Agriculture and Their Effects. World Bank Staff Working Paper 359. Washington, D.C. Bapna, S. L. 1980. Aggregate Supply Response of Crops in a Developing Region. Delhi: Sultant Chand and Sons. Barah, B. C., and Hans P. Binswanger. 1981. "The Effects of Crop Diversifica- tion, Technology, Price Stabilization and Crop Insurance on Revenue Risks: An Interregional Analysis for India." World Bank, Agriculture and Rural Development Department, Washington, D.C. Bates, Robert H. 1981. Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. Berkeley: University of California Press. 262 References 263 Batra, Raveendra N., and Aman Ullah. 1974. "Competitive Firms and the Theory of Input Demand under Price Uncertainty." Journal of Political Econ- omy 82: 537-48. Bautista Romeo, and Alberto Vald6s, eds. 1992. Trade and Macroeconomic Poli- cies in Developing Countries: Impact on Agriculture. San Francisco: Institute for Contemporary Studies. Bhagwati, Jagdish N. 1978. Foreign Trade Regimes and Economic Development: Anatomy and Consequences of Exchange Control Regimes. Lexington, Mass.: Ballinger Press for the National Bureau of Economic Research. Binswanger, Hans. 1990. "The Policy Response of Agriculture." Proceedings of the World Bank Annual Conference on Development Economics 1989. Washing- ton, D.C.: World Bank. Binswanger, Hans, and Jaime B. Quizon. 1988. "Distributional Conse- quences of Alternative Food Policies in India." In Per Pinstrup-Andersen, ed., Food Subsidies in Developing Countries. Baltimore: Johns Hopkins Uni- versity Press. Bond, Marian E. 1983. "Agricultural Responses to Prices in Sub-Saharan African Countries." International Monetary Fund Staff Papers, 30(4): 703-26. Branddo, A. S., and J. L. Carvalho. 1991. "Trade Exchange Rate and Agri- cultural Pricing Policies in Brazil." World Bank Comparative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C.: World Bank. Braverman, Avishay, Choong Yong Ahn, and Jeffrey S. Hammer. 1983. Their Implications for Government Deficits, Income Distribution, and Balance of Pay- ments. WorldBank Staff Working Paper 621. Washington, D.C. Cavallo, Domingo. 1989. "Agriculture and Economic Growth: The Experi- ence of Argentina 1913-84." In Maunder, Allan, and Alberto Vald6s, eds., Agriculture and Governments in an Independent World. Brookfield, Vt.: Gower. 20th Conference of the International Association of Agricultural Econo- mists, Buenos Aires, Argentina. CIMMYT (Centro Internacional de Org6ramiento del Orafs y Frigo). 1983. World Wheat Facts and Trends. Report 2: An Analysis of Rapidly Rising Third World Consumption and Imports of Wheat. Batan, Mexico. Coeymans, Juan Eduardo, and Yair Mundlak. 1992. "Agricultural Sectoral Growth: Chile, 1962-82." In R. Bautista and Alberto Valdds, eds., Trade and Macroeconomic Policies in Developing Countries: Impact on Agriculture. San Francisco: Institute for Contermporary Studies. Dardis, Rachel. 1967. "The Welfare Cost of Grain Protection in the United Kingdom." Journal of Farming Economics 49:597-609. Deaton, Angus. 1989. "Rice Prices and Income Distribution in Thailand: A Non-parametric Analysis". The Economic Journal 99(395):1-37. de Janvry, Alain. 1986. "Integration of Agriculture in the National and World Economy: Implications for Agricultural Policies in Developing Countries." In Agriculture in a Turbulent World Economy. Proceedings of the Nineteenth International Conference of Agricultural Economists. Brookfield, Vt.: Gower. 264 References de Janvry, Alain, and Kalihishi Subbarao. 1986. Agricultural Price Policy and Income Distribution in India. New York: Oxford University Press. Delgado, C. L., and J. W Mellor. 1984. "A Structural View of Policy Issues in African Agricultural Development." American Journal of Agricultural Eco- nomics 66:665-70. Dethier, Jean-Jacques. 1985. "The Political Economy of Food Prices in Egypt." Ph.D. diss., Agriculture and Resource Economics Department, University of California, Berkeley. _. 1989. Trade, Exchange Rate and Agricultural Pricing Policies in Egypt. World Bank Comparative Studies series on the Political Economy of Agri- cultural Pricing Policy. Washington, D.C. Diewert, W 1976. "Exact and Superlative Index Numbers." Journal of Econo- metrics 4:115-45. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper & Row. Falcon, Walter P., and Peter C. Timmer. 1975. "The Political Economy of Rice in Asia." Food Research Institute Studies 14(3)(whole issue). Furtado, Celso. 1965. Development and Stagnation in Latin America: A Structural- ist Approach. St Louis, Mo.: Social Science Institute, Washington University. Ghai, Dharam, and Lawrence D. Smith. 1987. Agricultural Prices, Policy, and Equity in Sub-Saharan Africa. Boulder, Colo.: Lynne Rienner for the Interna- tional Labor Office. Gulati,Ashok, and Garry Pursell. Forthcoming. Trade Policies, Incentives, and Resource Allocation in Indian Agriculture, World Bank, Washington, D.C. Hamid, Naved, Ijar Nabi, and Anjum Nasim. 1991. "The Case of Pakistan." In Anne 0. Krueger, Maurice Schiff, and Alberto Vald6s, eds., The Political Economy of Agricultural Pricing Policies in Developing Countries, vol. 2 Asia. Baltimore: Johns Hopkins University Press. Harberger, Arnold C. 1950. Classroom lectures on price theory. University of Chicago. Harris, John R., and Michael P. Todaro. 1970. "Organization, Unemployment and Development: A Two-Sector Analysis." American Economic Review 60(1):126-42. Hayami, Yujiro, and M. Kikuchi. 1978. "New Rice Technology and National Irrigation Development Policy." In Economic Consequences of the New Rice Technology. Los Bafios, Philippines: International Rice Research Institute. Herdt, Robert W 1970. "A Disaggregate Approach to Aggregate Supply." American Journal of Agricultural Economics 52:512-20. Honma, Masayoshi, and Yujiro Hayami. 1986. "The Determinants of Agri- cultural Protection Levels: An Econometric Analysis." In K. Anderson and Y. Hayami, eds., The Political Economy of Agricultural Protection. London: Allen and Unwin. Hurtado Hernin, Alberto Vald6s, and Eugenia Muchnik. 1990. Trade, Exchange Rate, and Agricultural Pricing Policies in Chile. World Bank Compar- References 265 ative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C. Intal, Ponciano, and John H. Power. 1990. Trade, Exchange Rate and Agri- cultural Pricing Policies in the Philippines. World Bank Comparative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C. Ishikawa, Shigeru. 1967. Economic Development in Asian Perspective. Tokyo: Kinokuniya Company. -. 1987. "Patterns and Processes of Intersectoral Resource Flows: Com- parison of Cases in Asia." In Gustav Ranis and T. Paul Schultz, eds., The State of Development Economics, Progress and Perspectives. London: Blackwell. Jansen, Doris. 1988. Trade, Exchange Rate and Agricultural Pricing Policies in Zambia. World Bank Comparative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C. Jenkins, Glen P., and Andrew Lal. 1990. Trade, Exchange Rate and Agricultural Pricing Policies in Malaysia. World Bank Comparative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C. Johnson, Harry G. 1969. "The Theory of Effective Protection and Prefer- ences." Economica, 36, (142):119-138. Josling, Timothy E. 1969. "Agriculture and Britain's Trade Policy Dilemma," Thames Essays, Trade Policy Research Centre, London. Just, Richard E. 1974. "An Investigation of the Importance of Risk in Farmers' Decisions." American Journal of Agricultural Economics 56:14-25. Just, Richard E., and R. D. Pope. 1976. "On the Relationship of Input Deci- sions and Risk." Department of Agricultural Economics Working Paper. University of California, Berkeley. Knudsen, Odin, and John Nash. 1990. "Domestic Price Stabilization Schemes in Developing Countries." Economic Development and Cultural Change 38:539-58. Krishna, Raj. 1984. "Price and Technology Policies." In C. K. Eicher and John M. Staatz, eds., Agricultural Development in the Third World. Baltimore: Johns Hopkins University Press. Krueger, Anne 0. 1978. Foreign Trade Regimes and Economic Development: Liber- alization Attempts and Consequences. Lexington, Mass.: Ballinger Press for the National Bureau for Economic Research. _. 1983. Trade and Employment in Developing Countries, vol. 3, Synthesis. Chicago: University of Chicago Press. Krueger, Anne 0., Hal B. Lary, Terry Monson, and Narongchai Akrasanee. 1981. Trade and Employment in Developing Countries, vol. 1, Individual Studies. Chicago: University of Chicago Press. Krueger, Anne 0., and B. Tuncer. 1982. "Growth of Factor Productivity in Turkish Manufacturing Industries." Journal of Development Economics 11:307-25. Laraki, Karim. 1989. Food Subsidies: A Case Study of Price Reform in Morocco. LSMS Working Paper 50. Washington, D.C.: World Bank. 266 References Lattimore, Ralph, and G. Edward Schuh. 1976. "Un modelo de politica para la industria brasilefia de ganado vacuno." Cuadernos de Economia 13, no. 39:51-75. Little, I. M. D., Tabor Scitovsky, and Maurice Scott. 1970. Industry and Trade in Some Developing Countries. London: Oxford University Press. Lopez, Ram6n. 1992. "Resource Degradation, Community Controls and Agricultural Productivity in Tropical Areas." Department of Agricultural Economics, University of Maryland, College Park. Lutz, Ernst, and Yasmin Saadat. 1988. "Issues Relating to Agricultural Pric- ing Policies and Their Analysis in Developing Countries." Journal of Agri- cultural Economics 2:19-37. Magnusson, G. 1969. Production under Risk: A Theoretical Study. Uppsala, Sweden: Almquist and Wiksells. McCarthy, F. Desmond, and J.A. 1985. "Sources of Growth in Colombia, 1963-80." Journal of Economic Studies. (U.K.) 12(4):3-14. Mellor, John W. 1978. "Food Price Policy and Income Distribution in Low- Income Countries." Economic Development and Cultural Change 27:1-26. Mirrlees, James A. 1988. "Optimal Commodity Price Intervention." World Bank, Washington, D.C. Mundlak, Yair. 1985. "The Aggregate Agricultural Supply." Center for Agri- cultural Economic Research Working Paper 8511. Rehovot, Israel. Mundlak, Yair, Domingo Cavallo, and Roberto Domenech. 1992. "Agricul- ture and Growth: The Experience of Argentina, 1913-84." In Romeo Bau- tista and Alberto Vald6s, eds., Trade and Macroeconomic Policies in Developing Countries: Impact on Agriculture. San Francisco: Institute for Contemporary Studies. Myint, Hla. 1987. Comments on "Patterns and Processes of Intersectoral Resource Flows: Comparison of Cases in Asia." In Gustav Ranis and T. Paul Schultz, eds., The State of Development Economics, Progress and Perspec- tives. London: Blackwell. Newbery, David, and Nicholas Stem. 1987. The Theory of Taxation for Develop- ing Countries. New York: Oxford University Press. Newbery, David, and Joseph Stiglitz. 1981. The Theory of Commodity Price Stabilization. New York: Oxford University Press. Nowshirvani, V. F. 1971. "Land Allocation under Uncertainty in Subsistence Agriculture. Oxford Economic Papers 23:445-55. OECD (Organization for Economic Cooperation and Development). 1987. National Policies and Agricultural Trade. Paris. Panagariya, Arvind, and Maurice Schiff. 1991. "Commodity Exports and Real Income in Africa." In Ajay Chhibber and Stanley Fischer, eds., Eco- nomic Reform in Sub-Saharan Africa, World Bank, Washington, D.C. Peterson, Willis L. 1979. "International Farm Prices and the Social Cost of Cheap Food Policies." American Journal of Agricultural Economics 61:12-21. Pinstrup-Andersen, Per. 1988. "The Social and Economic Effects of Consumer-Oriented Food Subsidies: A Summary of Current Evidence." In Per Pinstrup-Andersen, ed., Food Subsidies in Developing Countries. Balti- more: Johns Hopkins University Press. References 267 Prebisch, Raul. 1950. The Economic Development of Latin America and Its Principal Problems. New York: United Nations. - -. 1949. Economic Survey of Latin America. New York: United Nations. Quizon, Jaime B., and Hans P. Binswanger. 1984. "Income Distribution in India: The Impact of Policies and Growth in the Agricultural Sector." World Bank, Agriculture and Rural Development Department, Washing- ton, D.C. . 1986. "Modeling the Impact of Agricultural Growth and Government Policy on Income Distribution in India." World Bank Economic Review 1:103- 48. Ravallion, Martin. 1990. "Rural Welfare Effects of Food Price Changes Under Induced Wage Responses: Theory and Evidence for Bangladesh." Oxford Economic Papers 42:574-585. Reardon, Thomas. 1984. "Agricultural Price Policy in Peru." Ph.D. diss., Agriculture and Resource Economics Department, University of California, Berkeley. Reca, Lucio G. 1980. Argentina: Country Case Study of Agricultural Prices and Subsidies. World Bank Staff Working Paper 386. Washington, D.C. Rosenzweig, Mark R., and Oded Stark. 1989. "Consumption Smoothing, Migration and Marriage: Evidence from Rural India." Journal of Political Economy 97:905-926. Sah, Raaj Kumar, and Joseph E. Stiglitz. 1987. "Price Scissors and the Struc- ture of the Economy." Quarterly Journal of Economics 102:109-23. Schiff, Maurice. 1986. "The Competitive Firm under Uncertainty: An Appli- cation to Canadian Wheat Production." Canadian Journal of Agricultural Eco- nomics 34:235-42. Schultz, Theodore W 1964. Transforming Traditional Agriculture. New Haven, Conn.: Yale University Press. Scobie, Grant M. 1981. "Government Policy and Food Imports: The Case of Wheat in Egypt." Research Report 29, International Food Policy Research Institute, Washington, D.C. Siamwalla, Ammar, and S. Setboonsarng. 1991. "The Case of Thailand." In Anne 0. Krueger, Maurice Schiff, and Alberto Vald6s, eds., The Political Economy of Agricultural Pricing Policies in Developing Countries, vol. 2, Asia. Baltimore: Johns Hopkins University Press. Stryker, Dirck. 1991. "The Case of Morocco." In Anne 0. Krueger, Maurice Schiff, and Alberto Vald6s, eds., The Political Economy of Agricultural Pricing Policies in Developing Countries, vol. 3, Africa and the Mediterranean. Balti- more: Johns Hopkins University Press. Sturzenegger, A. 1990. "Trade, Exchange Rate and Agricultural Pricing Poli- cies," World Bank Comparative Studies series on the Political Economy of Agricultural Pricing Policy. Washington, D.C. Sunkel, Osvaldo. 1967. "La inflaci6n chilena: un enfoque heterodoxo." In 0. Sunkel, ed., Inflaci6n y Estructura Economica. Buenos Aires: Paidos. Thomas, Vinod 1985. Linking Macroeconomic and Agricultural Policies for Adjust- ment with Growth: The Colombian Experience. Baltimore: Johns Hopkins Uni- versity Press. 268 References Timmer, C. Peter. 1986. Getting Prices Right: The Scope and Limit of Agricultural Price Policy. Ithaca, NY: Cornell University Press. Timmer C. Peter, and W Falcon. 1975. "The Political Economy of Rice Pro- duction and Trade in Asia." In L. G. Reynolds, ed., Agriculture in Develop- ment Theory. New Haven, Conn.: Yale University Press. Tolley, George S., Vinod Thomas, and Chung Ming Wong. 1982. Agricultural Price Policies and the Developing Countries. Baltimore: Johns Hopkins Univer- sity Press. Toye J. F. J., ed. 1979. Taxation and Economic Development. London: Frank Cass. Vald6s, Alberto. 1973. "Trade Policy and Its Effect on the External Trade of Chile. 1945-1965." American Journal of Agricultural Economics 55:154-164. ---. 1981. Food Security for Developing Countries. Boulder, Colo.: Westview Press. Vald6s, Alberto, and Ammar Siamwalla. 1988. "Foreign Trade Regimes, Exchange Rate Policy, and the Structure of Incentives." In J. W Mellor and R. Ahmed, eds., Agricultural Price Policy for Developing Countries. Baltimore: Johns Hopkins University Press. Wallace, T. D. 1962. "Measures of Social Costs of Agricultural Programs" (with discussion by J. P. Cavin). Journal of Farming Economics 44:580-97. World Bank. 1986. World Development Report 1986. New York: Oxford Univer- sity Press. . 1988. World Development Report 1988. New York: Oxford University Press. The complete backlist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list and indexes of subjects, authors, and countries and regions. The latest edition is available free of charge from the Distribution Unit, Office of the Publisher, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, or from Publications, The World Bank, 66, avenue d'IMna, 75116 Paris, France. THE WORLD BANK This new series of comparative studies from the World Bank examines how policies have affected agriculture in eighteen developing coun- tries. All five volumes consider the impact of both direct policies toward agriculture and of general development policies on incentives confronting agricultural producers and on agriculture's contribution to development. In all volumes, it is shown that general policies can have effects even more powerful than direct policies on incentives. Three regional volumes estimate price discrimination against agriculture in individual countries, how it has changed over time, and the politi- cal-economic factors that guided the evolution. They also evaluate the effects of this price discrimination on such key macroeconomic vari- ables as foreign exchange earnings, agricultural output, and income distribution. Two synthesis volumes draw on the full range of country experience to give insight into the motivations of policymakers. the economic and political factors determining agricultural interventions, and the attempts to reform unsuccessful policies. The Political Economy of Agricultural Pricing Policy 1 Latin America Edited by Anne 0. Krueger, Maurice Schiff, and Alberto Vaides 2 Asia Edited by Anne 0. KrUeger, Maurice Schilf, and Albetio Valdes 3 Africa and the Mediterranean Edited by Anne 0. Krueger, Maurice Schiff, and Alhcrto Vidd s 4 A Synthesis of the Economics in Developing Countries Maurice Schiff and Alberto Vald6s 5 A Synthesis of the Political Economy in Developing Countries Anne 0. Krueger When this work commenced. Anne 0. Krueger was vice president, Economics and Research, at the World Bank and is now Arts and Sciences Professor of Economics at Duke University, Maurice Scliff was an economist at the World Bank. Alberto Valdes was program director for International Trade and Food Security at the International Food Policy Research Institute. ISBN 0-8018-453 1-9